{ "generatedAt": "2026-10-30T01:10:02Z", "origin": "https://greenscreenremover.net", "source": "Extracted from the live HTML pages at build time by agent/build_agent.py", "pages": [ { "path": "/", "title": "Green Screen Remover — Remove Green Screen from Video & Image", "description": "Free green screen remover that runs inside your browser. Drop a video or a photo shot on a green or blue screen, key the backdrop out, and download a transparent PNG, a PNG sequence or a transparent WebM. No upload, no sign-up, no watermark.", "markdown": "/ai/pages/index.md", "sections": [ { "heading": "Green Screen Remover", "text": "Drop a video or an image shot on a green or blue screen. The backdrop is keyed out inside this browser tab — nothing is uploaded. Drop a video or an image here or choose a file — you can paste from the clipboard too Runs locally in this browser tab · no upload · no account · no watermark Raise it if patches of the screen survive. Lower it if part of the subject disappears. Takes the green cast off hair, skin and light clothing." }, { "heading": "What this green screen remover does", "text": "This page is a chroma key tool for video and images. It reads the colour of the backdrop you shot against, makes those pixels transparent, and pulls the green light that bounced onto your subject back out of the edges. You get the cutout as a transparent PNG, as a PNG sequence for a clip, or as a transparent WebM video. Everything is computed by the browser on your own machine. The file you drop never leaves the tab — there is no upload step, no account and no watermark." }, { "heading": "Supported formats", "text": "Video: MP4 (H.264), WebM (VP8/VP9), and MOV or OGV when your browser can decode them. Anything the browser can play will key. Image: PNG, JPEG, WebP, GIF (first frame), BMP, and AVIF where the browser supports it. Output: transparent PNG at the full resolution of the original, a ZIP of transparent PNG frames, or a transparent WebM (VP9) recorded from the keyed canvas. Your browser is the limit, not this page: if it cannot decode the file, the page says so instead of pretending to work." }, { "heading": "How to remove a green screen", "text": "Drop your file. Drag a clip or a photo onto the page, choose one from your device, or paste an image from the clipboard. The green screen is detected from the border of the frame and keyed out straight away. Fix what the key got wrong. Tolerance decides how far from the key colour a pixel can be and still count as background. Edge trims the thin green halo off the outline or softens a jagged one. Spill removal pulls the green cast off the subject itself. Check it. Drag the divider to compare the original with the result, and switch the preview background between a checkerboard, dark and light so you can see the transparency honestly. Take the result. Download a transparent PNG. For a clip you can also export the current frame, a PNG sequence for the whole clip, or a transparent WebM." }, { "heading": "How the key works", "text": "Each pixel is converted to YCbCr, the colour space broadcast equipment has always keyed in, where Cb and Cr carry the colour and Y carries the brightness. The distance from the key colour is measured in the Cb/Cr plane only, so a light shadow falling across the screen — the same green, just slightly darker — still keys out, though deep shadows and underexposed footage call for a higher tolerance, which you can set by hand. A plain \"delete every pixel near this RGB value\" comparison leaves those shadows behind. That reasoning breaks on a neutral backdrop: white, black and grey all sit at the same point in the Cb/Cr plane, so a chroma key cannot tell a white wall from a white shirt. When the key colour has almost no saturation this page measures brightness instead and keys on luminance, which is the only thing that can separate them." }, { "heading": "How do I remove a green screen from a video?", "text": "Drop the clip on the page. The key colour is read from the first frame, every frame is keyed as it plays, and the preview shows the result live. When it looks right, export a transparent WebM or a PNG sequence." }, { "heading": "How do I remove a green screen from an image?", "text": "Drop the photo and the background is gone on arrival. Adjust tolerance, edge and spill if the cutout needs help, then download the transparent PNG." }, { "heading": "Is this green screen background remover free?", "text": "Yes. No account, no sign-up, no watermark and no limit on how many files you key. There is nothing to buy because nothing runs on a server — your own machine does the work." }, { "heading": "Are my files uploaded anywhere?", "text": "No. The page decodes and keys the file with the Canvas API inside the tab. It keeps working if you go offline after it has loaded." }, { "heading": "Can I key a blue screen, or another colour?", "text": "Yes. Pick the blue or magenta preset, use the colour picker, or press Pick from media and click the backdrop in the preview. Any solid backdrop colour can be the key." }, { "heading": "Why is there still a green edge around my subject?", "text": "That is spill — green light from the screen bouncing onto hair, shoulders and light clothing. Drag Edge to the left to trim the contaminated rim, and raise Spill removal to neutralise the cast that landed on the subject." }, { "heading": "Does the exported video keep its transparency?", "text": "The WebM export is recorded from a canvas that has an alpha channel, and it keeps it — verified by decoding the file back and reading the alpha of a transparent pixel. Whether a player shows it is a different matter: VLC, Chrome, Edge and most editors show the transparency, while many video players and social platforms paint transparent video on black. The PNG and PNG-sequence exports always keep alpha everywhere." }, { "heading": "Is there a length limit on the video?", "text": "No cap is imposed by this page. The limits are your machine's memory and the browser's decoder. Two practical notes: the WebM export records in real time, so a 30-second clip takes about 30 seconds, and a PNG sequence is heavy — a one-minute clip at 12 frames per second is 720 files." }, { "heading": "Why did my subject lose a piece, or why is a green prop gone?", "text": "A key decides by colour alone, so anything the same colour as the screen goes with it. Lower the tolerance, or pick a key colour further from the subject. Unlike some tools on this page there is no manual brush, so colour is the only lever — that is stated here rather than hidden behind a button that does not exist." }, { "heading": "What happens after the cutout", "text": "This page stops at the key. It hands you a transparent PNG, a PNG sequence or a transparent WebM, and it deliberately does not trim the clip, put the subject on a new background, add titles or export an MP4. Those are editor jobs, and a browser tab is the wrong place to do them. If that is the next thing you need, a desktop option is Movavi Video Editor. It is a paid product made by Movavi — a company we are not part of, and this is not an official Movavi page. Movavi offers a free 7-day trial of Video Editor, so you can test it on your own clips before deciding whether to buy. The link above is an affiliate link: if you buy through it, we may earn a commission — the price you pay stays the same." }, { "heading": "Guides and answers", "text": "Longer takes on the parts that usually go wrong. Same rules as this page: it all runs in the browser, and nothing is uploaded." }, { "heading": "Video", "text": "Remove Green Screen from Video Transparent Background Video Transparent WebM Explained PNG Sequence vs Transparent WebM Green Screen Video Export Settings Green Screen for YouTube Shorts OBS Chroma Key Settings Green Screen for Zoom and Meetings Green Screen File Size Limits" }, { "heading": "Image", "text": "Remove Green Background from Photo Remove White Background Without a Green Screen Transparent PNG Alpha Keying Hair and Transparent Edges Blue Screen Remover Remove a Green Screen Without Uploading Anything Green Screen Remover Without a Watermark Green Screen for Product Photos: From Cutout to White Background" }, { "heading": "Guides & Setup", "text": "Green Screen Lighting Guide DIY Green Screen at Home Blue Screen vs Green Screen Green Screen Spill Suppression Green Screen Tolerance Explained Chroma Key Artifacts Chroma Key Benchmark Chroma Key vs Luma Key Green Screen Camera Settings: Shutter, Frame Rate and ISO Why Green Screen Edges Look Jagged: Pixels, Resolution and Feathering Green Screen vs AI Background Removal: Which Workflow Fits? Green Screen Wrinkles and Shadows: Fix the Backdrop Before Keying Green Screen Clothing: Colors and Materials to Avoid Green Screen Distance from Subject: Spill, Shadows and Focus Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green screen remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/no-upload-green-screen-remover", "title": "Remove a Green Screen Without Uploading Anything", "description": "How to key out a green or blue screen with nothing leaving your computer: no upload, no account, no queue. A browser-based green screen remover that works on video and photos, and why keeping the file local matters.", "markdown": "/ai/pages/no-upload.md", "sections": [ { "heading": "Remove a green screen without uploading anything", "text": "Most online green screen removers ask you to send the file to their server first. This one does not. The keying happens inside the browser tab you already have open, and the file never leaves your machine." }, { "heading": "Why \"no upload\" is the whole point", "text": "Uploading a clip to a stranger's server means three things you usually did not agree to: your footage sits on someone else's disk, it gets processed by a queue you cannot see, and the export is often capped unless you pay. For a client's unreleased product video, a take with a person's face in it, or anything under an NDA, that is a problem before you even get to the quality of the key. Running locally removes all of it. There is no upload step, so there is nothing to delete afterwards. There is no queue, so a 500 MB file is as fast as your own processor allows. And there is no account, so there is nothing to sign in to." }, { "heading": "What you can do here", "text": "Photos — drop a still shot on a green or blue backdrop and download a transparent PNG. Video — key a clip frame by frame and export a transparent WebM, or a ZIP of PNG frames if you need a sequence for an editor. Green or blue — the key colour is a parameter, not a hard-coded assumption, so a blue screen works the same way. Check your result — toggle a checkerboard behind the output so you can actually see the holes before you export." }, { "heading": "What will not work, honestly", "text": "A colour-distance key separates pixels by how far they sit from the key colour. That has limits, and it is worth knowing them before you blame the tool: Shadowed backdrops. A dark patch on the screen sits further from the key colour than the lit part. Raise the tolerance and it comes back; raise it too far and parts of the subject start to disappear. Sheer or motion-blurred edges. A pixel that is half backdrop and half hair is genuinely ambiguous. Softening the edge helps; nothing will make it perfect. Spill on bright, thin subjects. Green light bouncing off the screen lands on hair and light clothing. The spill control pulls most of it back out, but heavy spill needs a dedicated despill pass. White walls. A white backdrop has almost no colour in it, so there is no chroma to key on. The tool falls back to luminance in that case, which is the best available option and still imperfect." }, { "heading": "Getting a clean result", "text": "Light the screen evenly. This matters more than any slider. A flat, evenly lit backdrop keys at a low tolerance; a shadowed one needs a high tolerance that starts eating the subject. Leave the default tolerance first. It is tuned for a normal green screen. Only move it if you actually see leftover patches or missing edges. Use the checker toggle. Transparent areas look like nothing. Checkerboard makes them obvious. Shoot a bit wider than you need. Keying trims the fringe; having margin means you keep the subject's outline." }, { "heading": "Does anything get sent anywhere?", "text": "No. The page loads once and then works offline if you keep it open. There is no analytics script on the tool page, no account to create, and no watermark on the export. The full detail is on the privacy page. Open the green screen remover" }, { "heading": "Related image pages", "text": "Green Screen Remover (home) Green Screen Remover Without a Watermark Remove Green Background from Photo Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/blue-screen-remover", "title": "Blue Screen Remover — Key Out a Blue Backdrop in Your Browser", "description": "A blue screen remover that runs locally: drop a photo or clip shot against a blue backdrop and export a transparent PNG, PNG sequence or WebM. No upload, no account, no watermark.", "markdown": "/ai/pages/blue-screen.md", "sections": [ { "heading": "Blue screen remover", "text": "Blue backdrops are used whenever the subject is warm — skin tones, brown hair, warm clothing, anything with a lot of red and yellow in it. This tool keys blue the same way it keys green: locally, in your browser, with the key colour as a parameter rather than a fixed assumption." }, { "heading": "Blue or green — which one should you shoot on?", "text": "Both work. The usual reasons to pick blue: Warm subjects. Green screens spill green onto skin, which reads as sickly. Blue spill is easier to live with on warm tones. Night scenes. Blue is closer to the ambient colour of a night look, so the composite sits better. Blue clothing. The obvious reverse: if the talent is wearing blue, do not shoot on blue. Green is still the more common choice, mostly because digital camera sensors are more sensitive in the green channel, so a green screen keys more cleanly at the same light level." }, { "heading": "How to key a blue screen here", "text": "Open the tool and drop your file in — photo or video, either works. Under Key colour, pick Blue. Auto-detect usually finds it on its own; setting it by hand is the fallback. Leave Tolerance at the default and look at the result on the checkerboard. Raise it only if patches of blue survive; lower it only if parts of the subject vanish. Export a transparent PNG, a ZIP of PNG frames, or a transparent WebM." }, { "heading": "Why the key colour is a parameter", "text": "A lot of \"green screen remover\" tools hard-code green. That is fine until you have a blue screen, a magenta one, or a neutral wall. This one measures distance in the Cb/Cr plane from whatever key colour you give it, and falls back to a luminance measure when the backdrop has almost no chroma — which is what lets it handle a white wall, badly but not catastrophically. We published the measurement behind that on the benchmark page, including the two cases where a single global threshold fails." }, { "heading": "What you get out", "text": "Transparent PNG for stills. PNG sequence in a ZIP for bringing frames into an editor. Transparent WebM for video that needs to keep its alpha. No watermark, no resolution cap, no account. The file stays on your machine the whole time — see why that matters. Open the green screen remover" }, { "heading": "Related image pages", "text": "Green Screen Remover (home) Remove a Green Screen Without Uploading Anything Green Screen Remover Without a Watermark Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/chroma-key-benchmark", "title": "Chroma Key Benchmark — What We Measured and Where It Fails", "description": "A reproducible chroma key benchmark on synthetic 320x240 fixtures: Cb/Cr distance keying versus two naive baselines, with mean IoU, oracle thresholds, and the two fixtures a single global threshold cannot solve.", "markdown": "/ai/pages/benchmark.md", "sections": [ { "heading": "Chroma key benchmark", "text": "We measured the keyer behind this site against two naive baselines on six synthetic fixtures. Every number below is regenerated by one command from a seeded generator, and the full write-up is archived with a DOI. We claim no accuracy advantage over a naive RGB threshold — and this page explains why." }, { "heading": "Setup", "text": "Six synthetic fixtures, 320×240: flat green, green with a shadow gradient, green with spill on the subject, green with both, flat blue, and a near-white neutral backdrop. Metric: intersection over union (IoU) between the produced alpha and the known ground-truth matte. Baselines are given oracle thresholds — each is evaluated at its own best value, which is the fairest possible comparison. Comparing against a badly tuned baseline would be meaningless. Reproduce with node bench/run.mjs and node bench/baseline.mjs in the reference repository." }, { "heading": "What the table actually says", "text": "A naive RGB threshold matches us to within 0.002 mean IoU. That is the honest reading, and it is why we do not advertise the keyer as more accurate. On flat, evenly lit, synthetic backdrops, any colour-distance rule saturates — there are only two kinds of pixel, and any threshold separates them. A benchmark like this has almost no discriminating power. What does separate the methods is the case where the assumption breaks: The green-dominance rule collapses on non-green backdrops. It scores 0.000000 on the flat blue and near-white fixtures. That is the argument for making the key colour a parameter rather than hard-coding green. Our shipped default scores 0.873398, well below our own best tolerance of 0.994378. Two fixtures — the shadowed ones — drag it down to roughly 0.62. A single global threshold cannot cover both a lit backdrop and a shadowed one in the same frame; you either keep the shadow or lose part of the subject. This is a structural property of the method, not a tuning bug, and it is why the tolerance control exists." }, { "heading": "Limits you should hold us to", "text": "Everything here is synthetic. No real footage, no real camera noise, no compression artefacts. A number that looks good on a gradient does not promise the same on a phone video. IoU rewards area, not edge quality. It says almost nothing about hair, fur, motion blur or fine detail — the places where keyers actually differ. The baselines are weak. A modern learned matting model would beat every row in this table. We compared against naive rules because they are what a dependency-free browser implementation competes with." }, { "heading": "Full write-up", "text": "The preprint with the method, the fixture construction and the complete tables is archived at doi.org/10.5281/zenodo.22916767 (CC BY 4.0). The source and the CSVs are in the chroma-key-reference repository. Open the green screen remover" }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) Chroma Key vs Luma Key Green Screen Lighting Guide Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/remove-green-screen-from-video", "title": "Remove Green Screen from Video — Export Transparent WebM, No Upload", "description": "Remove a green screen from a video in your browser: the backdrop is keyed out frame by frame and you export a transparent WebM or a PNG sequence. No upload, no account, no watermark.", "markdown": "/ai/pages/video.md", "sections": [ { "heading": "Remove green screen from video", "text": "Drop a clip shot against a green or blue backdrop, watch the key run frame by frame, and take the result away with the transparency intact — as a transparent WebM, as a ZIP of PNG frames, or as a single full-resolution still. The file never leaves your machine." }, { "heading": "What you are actually exporting", "text": "Removing a green screen from a video is a different job from doing it to a photo. A still has one alpha matte; a clip has one per frame, and whatever you export has to carry all of them. That leaves three honest options, and they are not interchangeable: Transparent WebM (VP9). A real video file with an alpha channel. Small, and it plays in VLC, Chrome, Edge and most editors. Many other players and most social platforms ignore the alpha and paint the video on black. PNG sequence in a ZIP. One transparent PNG per frame at the resolution of the original. The safest choice: every editor and compositor understands it, and nothing can silently flatten it. Larger, and you choose the frame rate. Single transparent PNG. One frame of the clip, full resolution. Right when all you need is a still or a plate. Pick the sequence when the result is going into an editor, WebM when you need a video file that stays small, and the single PNG when you only need one frame." }, { "heading": "How to remove a green screen from a video here", "text": "Open the tool and drop the clip in — MP4 (H.264) and WebM (VP8/VP9) are the reliable ones; MOV and OGV work when your browser can decode them. If it cannot read the file, the page says so instead of pretending to. Check the key colour. It is read from the border of the first frame, so most clips need nothing. Set it by hand, or use Pick from media and click the backdrop, if auto-detect grabbed the wrong thing. Adjust Tolerance, Edge and Spill while scrubbing through the clip. Tolerance is the radius around the key colour; Edge trims or softens the outline; Spill pulls the green cast off hair, shoulders and light clothing. Blue backdrops work the same way — see the blue screen page. Export: transparent WebM, the PNG sequence, or the current frame as a PNG." }, { "heading": "What the WebM export is really doing", "text": "Worth knowing, because it explains the two things people notice first. There is no video encoder in this page. The keyed canvas is captured with captureStream and recorded by the browser's own MediaRecorder while the clip plays — VP9 where the browser supports it and plain WebM otherwise, at roughly 8 Mbps. Two consequences follow: It runs in real time. A two-minute clip takes about two minutes to export, and the tab needs to keep working while it does. MediaRecorder writes no duration into the WebM header. Players usually work it out, but some editors and tools show an unknown or wrong length until the file is remuxed. The PNG sequence does not have this problem. The exported frame size is also rounded down to even pixel dimensions, which is what VP9 requires — so a 1921×1080 source comes out 1920×1080." }, { "heading": "Shadows are the usual reason a key fails", "text": "A shadow falling across the screen is still the same green, only darker, which is the whole reason this keyer measures distance in the Cb/Cr plane and ignores brightness. Ignoring brightness is not the same as tolerating any amount of it. On our synthetic shade ramp, at the shipped tolerance of 25 the backdrop stops keying out once it falls to between 0.65 and 0.60 of full brightness: the pixel's alpha flips from 0 (gone) to 255 (solid), because its chroma distance from the key has reached about 33.5 — past the roughly 29.5 radius tolerance 25 allows. Raising tolerance to 50 keeps that same ramp keyed down to 0.40, the darkest row in the file. Those are generated fixtures — smooth ramps with no camera noise, no compression and no real fabric — so read them as the shape of the problem, not a promise about your footage. The full measurement is on the chroma key benchmark page, the CSV is in the reference repository, and the write-up is archived at doi.org/10.5281/zenodo.22916767." }, { "heading": "When this does not work", "text": "Every page on this site has to say where the tool stops. For video, that is: Long clips as a PNG sequence. The sequence is capped at 1200 frames; at the default 12 fps that is around 100 seconds. The page tells you to lower the frame rate rather than failing quietly. Clips with no readable duration. Live recordings and fragmented streams report no length. Re-encode to MP4 or WebM first. Browsers that cannot record a canvas. Without MediaRecorder and captureStream the WebM option is unavailable, and the PNG sequence is the fallback. Uneven, wrinkled or badly lit backdrops. One global tolerance covers a lit screen or a shadowed one, not both in the same frame — the benchmark shows the two fixtures where that breaks. Anything that needs rotoscoping. This is a colour key. Semi-transparent detail — smoke, glass, veils, flyaway hair over a moving background — is beyond it. Subjects that are themselves green. Green clothing, green props or a green cast in the lighting get keyed out along with the screen. Shoot on blue instead." }, { "heading": "Your clip stays on your machine", "text": "Every frame is decoded and keyed by your own browser, so there is no upload step, no render queue and no account. A two-minute clip in a phone codec is exactly the kind of file you would rather not hand to a stranger's server — the no-upload page explains what that avoids. Open the green screen remover" }, { "heading": "Related video pages", "text": "Green Screen Remover (home) Transparent Background Video Transparent WebM Explained Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/transparent-background-video", "title": "Transparent Background Video — Formats That Actually Carry Alpha", "description": "How to make a video with a transparent background: which formats carry an alpha channel, why MP4 cannot, why transparent clips often play on black, and how to export one in your browser with no upload.", "markdown": "/ai/pages/transparent-video.md", "sections": [ { "heading": "Transparent background video", "text": "A video with a transparent background is an ordinary video file in which every pixel of every frame carries an alpha value, so whatever sits behind it shows through. This page is about the part that goes wrong most: transparency computed correctly, then thrown away by the format on the way out." }, { "heading": "What “transparent” actually means in a video file", "text": "Removing a green screen and having a transparent video are two different steps. The first produces a matte — a per-pixel decision about how much of the subject is there. The second is storage: the file has to keep that matte. Most formats people reach for cannot, and when they cannot the alpha is flattened at export and the clip turns back into a rectangle. No keyer quality fixes that, because the information was discarded after the key. Partial transparency matters here. A hard cut-out has only two values — fully there or fully gone — which is what makes keyed edges look cut with scissors. Real footage needs the in-between values along hair, motion blur and soft fabric." }, { "heading": "Formats that carry alpha, and the one that cannot", "text": "WebM with VP9. The open combination that genuinely stores an alpha channel and plays in Chrome, Edge, Firefox and VLC. This is what the tool writes, because it is what the browser can produce on its own. A PNG sequence. One transparent PNG per frame, usually zipped together. Every editor, compositor and motion-graphics tool understands it, and nothing can silently flatten it. It is also the format you want for a timeline. ProRes 4444 in a MOV. The professional answer, and the one that needs a desktop editor to write. Browsers cannot produce it, so it is out of scope for this site. GIF. Technically transparent, and technically a video, but its transparency is one bit — a pixel is either there or not — with no partial edges. Fine for a sticker, wrong for footage. MP4 / H.264. No alpha channel. There is no setting that turns it on. An MP4 of your subject will always be a rectangle, whatever the keyer did. Apple’s HEVC-with-alpha is the one real exception, and it stays inside Apple’s own ecosystem — browsers give you no way to write it." }, { "heading": "Why your transparent video plays on black", "text": "This is the most reported “my export is broken”, and it usually is not broken. A player that ignores the alpha channel has to composite the frames against something, and that something is usually black. The file still holds the transparency; the player does not. Test it properly rather than trusting one player: drop the WebM into a page with a coloured background, or import it into an editor and put another clip underneath. If the subject floats, the alpha survived. If it floats in Chrome and sits on black in your social feed, the difference is the player, not the file. Uploading is the hard version. Any platform that re-encodes to H.264 destroys the alpha on the way in, and nothing in the file can stop it. If the platform documents an alpha-friendly format, use that; if it does not, assume the transparency will not survive." }, { "heading": "Making a transparent video here", "text": "Open the tool and drop in footage shot against a green or blue backdrop. MP4 (H.264) and WebM (VP8/VP9) are the reliable inputs; MOV and OGV work when your browser can decode them, and if it cannot, the page says so. Let it read the key colour from the border of the frame, then adjust Tolerance, Edge and Spill while scrubbing. Blue backdrops behave the same way — see the blue screen page. Export a transparent WebM, a ZIP of transparent PNG frames, or the current frame as a single PNG. The video page covers what each export does under the hood." }, { "heading": "Transparent WebM or PNG sequence?", "text": "Pick the WebM when you need one small file that plays in a browser or in VLC. It is recorded from the keyed canvas at roughly 8 Mbps with VP9 where the browser supports it, and it takes as long as the clip does — there is no video encoder in the page, so the export runs in real time. The recorder also writes no duration into the header, so some editors show an unknown length until the file is remuxed. Pick the PNG sequence when the result is going into an editor or when you cannot afford a surprise. It is larger, it has no duration problem, and you choose the frame rate. Here it is capped at 1200 frames — about 100 seconds at the default 12 fps — and the page tells you to lower the rate rather than failing quietly. Pick the single PNG when you only need one frame." }, { "heading": "Edges are where transparency lives", "text": "The middle of the subject is easy. The difficulty is the outline, and an outline is what a colour key can only approximate. A shadow across the screen shows why: it is the same green, only darker, so this keyer measures distance in the Cb/Cr plane and ignores brightness. Ignoring brightness is not the same as tolerating any amount of it. On our synthetic shade ramp, at the shipped tolerance of 25, the backdrop still keys out at 0.65 of full brightness (chroma distance about 29.3) and stops keying at 0.60 (about 33.5) — the alpha flips straight from 0 to 255. Raising tolerance to 50 keeps that ramp keyed down to 0.40, the darkest row in the file. Those numbers come from generated fixtures — smooth ramps with no camera noise, no compression and no real fabric — so read them as the shape of the problem, not a promise about your footage. The full measurement is on the chroma key benchmark page, the CSV is in the reference repository, and the write-up is archived at doi.org/10.5281/zenodo.22916767." }, { "heading": "When this does not work", "text": "You need an MP4. There is no transparent MP4 to be had; this tool does not write one and no setting will produce one. Your target platform re-encodes uploads. The transparency is gone the moment it is transcoded. Use the platform’s own alpha-friendly format, or composite the clip onto a background first. Your player ignores alpha. The file is fine. Verify over a coloured background before concluding the export failed. Long clips as a PNG sequence. Capped at 1200 frames here; lower the frame rate, or take the WebM instead. Browsers that cannot record a canvas. Without MediaRecorder and captureStream there is no WebM option and the PNG sequence is the fallback. Anything that needs rotoscoping. This is a colour key. Smoke, glass, veils and flyaway hair over a moving background are beyond it, and no file format adds detail the matte never had. Uneven or badly lit backdrops. One global tolerance covers a lit screen or a shadowed one, not both in the same frame." }, { "heading": "Your footage stays on your machine", "text": "Decoding, keying and recording all happen in your own browser tab: no upload, no render queue, no account, no watermark. The no-upload page explains what keeping it local avoids. Open the green screen remover" }, { "heading": "Related video pages", "text": "Green Screen Remover (home) Transparent WebM Explained PNG Sequence vs Transparent WebM Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/remove-green-background-from-photo", "title": "Remove Green Background from Photo — Export a Transparent PNG", "description": "Remove a green background from a photo in your browser: the backdrop is keyed out and you download a transparent PNG at the full resolution of the original. No upload, no account, no watermark.", "markdown": "/ai/pages/photo.md", "sections": [ { "heading": "Remove green background from photo", "text": "Drop a photo shot against a green or blue backdrop and the screen is keyed out on arrival. What you download is a transparent PNG at the full resolution of the original — not a downscaled web image, not a flattened JPEG. The file never leaves your machine." }, { "heading": "What a transparent PNG actually gives you", "text": "Most “remove background” tools hand back a JPEG, and a JPEG cannot store transparency at all — the holes get filled with white or black and the cutout is only ever as good as the moment you saved it. A PNG carries an alpha channel: every pixel keeps a fourth value saying how opaque it is, so the edges stay soft and the file can be dropped onto any other background later without re-doing the work. That is what makes the result usable beyond one place: a slide deck, a product listing, a print-on-demand upload that asks for a transparent cutout, or a compositing app. One file, no re-export." }, { "heading": "How to do it here", "text": "Open the tool and drop the photo in — PNG, JPEG, WebP, BMP and AVIF where your browser supports it, plus paste-from-clipboard. The key runs straight away. Check the key colour. It is read from the border of the frame, so a clean shot usually needs nothing. If auto-detect grabbed the wrong thing, set the colour by hand or use Pick from media and click the backdrop. Nudge Tolerance, Edge and Spill. Tolerance is how far from the key colour a pixel can sit and still count as background. Edge trims the thin green rim off the outline or softens a jagged one. Spill pulls the green light that bounced off the screen back out of hair, shoulders and light clothing. Download the transparent PNG. It comes out at the resolution of the original." }, { "heading": "Why photos are harder than they look", "text": "A still gives you one frame, which sounds easier than a clip — and it is, until the backdrop is not uniform. One photo has to be solved with one global threshold, and a real screen is almost never one colour: light falls off towards the corners, fabric wrinkles, a shadow crosses part of it. The keyer measures distance in the Cb/Cr plane and ignores brightness, which is what lets a shadowed screen still key out — but only up to a point. On our synthetic shade ramp, at the shipped tolerance of 25 the backdrop stops keying out once it falls to between 0.65 and 0.60 of full brightness: its chroma distance from the key reaches about 33.5, past the roughly 29.5 radius that tolerance 25 allows, and the pixel flips from fully transparent to fully solid. Raising tolerance to 50 keeps that same ramp keyed all the way down to 0.40, the darkest row in the file. So the usual fix for a screen with a shadow across it is more tolerance, not a different tool. The other thing a real photo adds is compression. A JPEG has already smeared colour across the boundary, so the keyer is staring at a soft ring of mixed pixels instead of a clean edge — and a chroma key has nothing to work with there. Shoot or export PNG when you can." }, { "heading": "When the auto-detected colour is slightly wrong", "text": "Auto-detect reads the border of the frame, and on a shadowed screen the border is darker than the rest of it. In our fixture run it recovers flat green and flat blue exactly, and a neutral white wall exactly — but on the shadowed variants it lands darker than the true backdrop: #00a23b where the screen is #00b140, a largest-channel difference of 15, and #009d39 with a difference of 20 once spill is in the picture too. That is small enough to key fine in practice, and it is worth knowing that “the tool picked the wrong green” is usually this, not a bug: pick the colour off a lit part of the screen by hand and the shadow stops mattering." }, { "heading": "About that Edge slider", "text": "Edge is not a blur you can set once and forget, and the default is not always the best value. In the edge sweep at tolerance 25, the flat spill fixture scores a perfect IoU of 1.000 at edge 0, and drifts down to 0.889 at the far end of the range. The shadowed fixture behaves differently: its best score is 0.964 around edge −30, and it falls off a cliff after that — 0.905 at −20, 0.702 at −10, and 0.623 at the default 0. Same slider, opposite answer, because the two images have different problems. Move it while watching the outline, not by rule. Those are generated fixtures — smooth ramps with no camera noise, no fabric texture and no compression — so read them as the shape of the problem, not a promise about your photo. The full measurement is on the chroma key benchmark page, the CSV files are in the reference repository, and the write-up is archived at doi.org/10.5281/zenodo.22916767." }, { "heading": "When this does not work", "text": "Every page on this site has to say where the tool stops. For stills, that is: Semi-transparent detail. Flyaway hair over a busy background, fur, veils, smoke, glass, anything with motion blur in it. A colour key produces one matte; those need a hand-drawn mask or a rotoscope. This page does not have a brush, and it does not pretend to. A subject that is itself green. Green clothing, a green prop, or a green cast in the lighting gets keyed out along with the screen. Raising tolerance makes it worse. Shoot on blue instead — see the blue screen page. Backdrops that are lit and shadowed in the same frame. One global threshold covers one or the other. The ramp above is exactly this failure, measured. Neutral backdrops. White, grey and black all sit at the same point in the Cb/Cr plane, so a chroma key cannot tell a white wall from a white shirt. When the key colour has almost no saturation the page switches to a luminance key, which can separate them by brightness — and cannot separate a white shirt from a white wall of the same brightness. Wrinkled or uneven fabric. Every fold is a different shade of green. Steam it, stretch it, or light it flat; no slider fixes it. Heavily compressed or very small source files. There is no detail left at the edge to key against. Start from the largest, least compressed version you have. Anything involving a new background. This page stops at the key. It will not place your subject on another image, add a shadow, or export a JPEG. That is an editor's job." }, { "heading": "Your photo stays on your machine", "text": "The whole key runs in the browser tab: decode, key, PNG encode, download. No upload step, no account, no queue, no watermark. For product photos, IDs and anything else you would rather not hand to a stranger's server, that is the part that matters — the no-upload page spells out what it avoids. Open the green screen remover" }, { "heading": "Related image pages", "text": "Green Screen Remover (home) Remove White Background Without a Green Screen Transparent PNG Alpha Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-lighting-guide", "title": "Green Screen Lighting Guide — Light the Backdrop So the Key Comes Out Clean", "description": "How to light a green screen so a chroma key comes out clean: light the backdrop on its own, light the subject separately, keep them apart, kill the bounce — plus the measurements behind why an uneven screen fails.", "markdown": "/ai/pages/lighting.md", "sections": [ { "heading": "Green screen lighting guide", "text": "A chroma key is only as good as the light on the backdrop. Light the screen evenly by itself, light the subject separately, keep the two apart, and stop the green bouncing back onto the person. This page is about what to fix before you record — and it ends with what lighting cannot fix." }, { "heading": "Why the light matters more than the keyer", "text": "The key measures how far each pixel sits from the key colour in the Cb/Cr plane and ignores brightness, which is why a shadow falling across a screen often still keys out. “Often” is the whole problem. A green screen at 65% of its full brightness is still close enough to the key colour to be removed; the same screen at 60% is not. On a synthetic brightness ramp, the backdrop keys cleanly all the way down to 0.65 of full brightness at the shipped tolerance of 25, and flips to fully solid at 0.60 — its chroma distance from the key reaches about 33.5, past the roughly 29.5 radius that tolerance allows. One global threshold, one screen, and half of it stops working. That is what uneven lighting does. It is not a subtle quality difference; it is the difference between a pixel being transparent and being solid. A screen lit flat from two sides has one colour; lit from one side, a gradient of them." }, { "heading": "The rule: two separate lighting jobs", "text": "Lighting a green screen is two jobs, not one, and the usual mistake is doing both with the same lamp. Job one: the backdrop. Its own lights, aimed only at the screen, even from edge to edge. Job two: the subject. Their own light, aimed only at them, shaped the way you would light them anyway. Then keep them apart. Distance is what stops the subject casting a shadow on the screen and stops green light bouncing back into their hair and shoulders. With one light, put it on the subject and light the screen as evenly as you can, then raise tolerance instead. The honest trade: more tolerance forgives an uneven screen and eats into fine detail." }, { "heading": "Setting it up", "text": "Get the screen flat first. A wrinkle is a shadow you cannot light away, and every fold is a different brightness — the same failure as the ramp above. Stretch it, steam it, or weight the bottom edge. Light the screen from both sides. One light from each side, aimed at the backdrop at roughly 45°, feathered so the two overlap in the middle. Two matched lights are ideal; two different ones are fine as long as the screen reads as one colour. Do not let the subject light touch the screen. Flag it off, or move the person forward. That spill is what makes the bright patch behind one shoulder. Push the subject away from the screen, as far as the room allows: their shadow falls behind them instead of on the backdrop, the green bounce falls off, and the screen keeps its own light. Expose the screen sensibly. Under-expose it and you get noise in the matte; blow it out and the green desaturates towards white, which cannot be separated from anything else of the same brightness. Block the bounce. Green light wrapping onto the subject is spill. The Spill slider pulls some of it back out of hair, shoulders and light clothing, but every bit you stop with distance or a flag is a bit it does not have to guess at." }, { "heading": "Check it before you shoot", "text": "Take one still with the subject in place and drop it into the tool. Toggle the transparency view and look at the corners against the centre: if the screen keys in one place and not in another, you have a lighting problem, not a slider problem. Look at the outline of the subject for a green rim — that is bounce reaching them. And check the auto-detected key colour: on a shadowed screen it reads darker than the real backdrop, in our fixture run #00a23b where the screen is #00b140, a largest-channel difference of 15. Small enough to key fine; for the exact colour, pick it by hand off a lit part of the screen. How much does fixing it actually buy you? On the two synthetic fixtures that carry a shadow, the shipped tolerance of 25 scores a mean IoU of 0.618 and 0.623 — roughly two thirds of the backdrop left in the frame. Raising tolerance to 40 takes them to 1.000 and 0.966. Same images, no code change: the shadow was a lighting artefact the keyer was being asked to compensate for. Those are generated fixtures — smooth ramps with no camera noise, no fabric texture and no compression — so read them as the shape of the problem, not a promise about your shot. The full measurement is on the chroma key benchmark page, the CSV files are in the reference repository, and the write-up is archived at doi.org/10.5281/zenodo.22916767." }, { "heading": "When this does not work", "text": "Lighting fixes evenness, spill and shadow. It does not fix these, and no page on this site should pretend otherwise: Semi-transparent detail. Flyaway hair, fur, veils, smoke, glass, anything with motion blur. A colour key produces one matte; these need a hand-drawn mask or a rotoscope. There is no brush here. A subject that is itself green. Green clothing or a green prop gets keyed out with the screen, and more tolerance makes it worse. Shoot on blue instead — see the blue screen page. Specular highlights. A shine blown out to white on a shirt or on the screen itself sits at the neutral centre; a chroma key cannot separate it from any other white. Flags, diffusion and a polariser help by removing the highlight, not by making the key smarter. Wrinkles you cannot get out. Every fold is a different brightness. Steam it, stretch it, or accept a manual mask. A room too small for distance. With no separation, the shadow lands on the screen and the bounce lands on the subject. A blue backdrop reflects less light than a green one, which is one reason people switch — a mitigation, not a cure. Mixed or drifting colour temperature. Daylight through a window next to a tungsten lamp gives you two different greens, and the whole shoot drifts as the sun moves. Close the blinds and light it yourself. A screen that is too dark overall. Noise at the edge is not something a threshold can read. Add light, or raise the camera’s exposure. Shooting through heavy compression. A highly compressed recording smears colour across the boundary before the keyer ever sees it. Lighting cannot undo that; the video page covers what to feed the tool instead." }, { "heading": "Then key it locally", "text": "Once the screen is lit, drop the clip or the still into the tool and the backdrop is keyed out in the browser tab — no upload, no account, no watermark. The no-upload page says what keeping the file on your machine avoids. Open the green screen remover" }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) DIY Green Screen at Home Blue Screen vs Green Screen Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/blue-screen-vs-green-screen", "title": "Blue Screen vs Green Screen — How to Choose, and What Breaks", "description": "Blue screen or green screen: what each one is good at, the luminance gap between them, why most green-only keyers score zero on blue, and the cases where neither works.", "markdown": "/ai/pages/blue-vs-green.md", "sections": [ { "heading": "Blue screen vs green screen", "text": "Both key out fine in this tool. The difference is not in the software — it is in what the two colours do to your subject, to your lights and to your camera. Here is the honest split, with the measurements behind it." }, { "heading": "The short answer", "text": "Shoot on green unless you have a specific reason not to. Green is the default because a green backdrop puts more light into the channel a digital sensor is most sensitive to, and because green sits further from human skin tone than blue does — a green spill is easier to separate from a face than a blue one. Pick blue when the subject is warm: skin, brown hair, wood, warm clothing. Green spill on skin reads as sickly; blue spill is easier to live with. Blue also suits night scenes, because it is closer to the ambient colour of a night look and the composite sits better before you grade it. And pick blue when the talent is wearing green, or when foliage is in shot. Do not pick blue when the talent is wearing blue, when you want to keep a blue sky in frame, or when you are short on light. That last one is the one people get wrong." }, { "heading": "The luminance gap is the real difference", "text": "Colour separation is not the hard part. In our own preset table, the green screen preset #00b140 carries a chroma magnitude of 83.66 and the blue screen preset #0047bb carries 83.16 — nearly identical. The two colours are equally far from neutral. Their luminance is not identical. The green preset sits at luma 111.20; the blue preset sits at 62.99, about 57% as bright. On the broadcast presets the gap is wider: #00ff00 at 149.69 against #0000ff at 29.07. What that means on set: with the same lights and the same exposure, a blue backdrop lands much lower in your camera's range than a green one. Lower means closer to the noise floor, and a key that has to separate a colour from sensor noise is a key that speckles along the edges. If you shoot blue, budget more light for the backdrop than you would for green." }, { "heading": "Why most green screen tools cannot key blue", "text": "A large number of \"green screen remover\" tools never measure colour distance at all. They use a shortcut: if the green channel is the largest channel in a pixel, that pixel is backdrop. It works on green and it fails on everything else. We measured how hard it fails. On our synthetic fixtures, that green-dominance heuristic scores an IoU of 1.000 on a flat green fixture and 1.000 on a shadowed green one — and 0.000 on a flat blue fixture, even when you hand it the best possible threshold. It keys nothing whatsoever. It scores 0.000 on a neutral white wall too. The key in this tool measures distance in the Cb/Cr plane from whatever key colour you give it, so blue is not a special case. On the flat-blue fixture it scores 1.000 at the shipped default tolerance of 25, and auto-detection reads the backdrop as #0047bb with a maximum channel delta of 0." }, { "heading": "Blue tolerates a wider tolerance — on our fixture", "text": "One result we did not expect. In the tolerance sweep, the flat-green fixture holds an IoU of 1.000 up to tolerance 85, then collapses at 90: false positives jump to all 21,341 subject pixels and IoU drops to 0.722. The flat-blue fixture, with the same subject, holds 1.000 all the way to tolerance 100. The likely reason is mundane: on this fixture the subject sits closer to green in the Cb/Cr plane than to blue, so there is more room to open the blue tolerance before it starts eating the subject. Treat it as a property of these fixtures, not a rule. It is one synthetic subject at 320x240, not a survey of real footage." }, { "heading": "How to switch this tool to blue", "text": "Open the tool and drop your file in — photo or video. Under Key colour, pick Blue. Auto-detect usually finds it; setting it by hand is the fallback. Leave tolerance at the default and read the result on the checkerboard. Raise it only if patches of blue survive. Export a transparent PNG, a ZIP of PNG frames, or a transparent WebM. No upload, no account, no watermark. The file stays on your machine the whole time — see why that matters." }, { "heading": "When this does not work", "text": "Blue in the subject. A blue key is colour-blind. Blue eyes go transparent. So do blue jeans, a blue logo on a shirt, and a blue phone screen. Blue you wanted to keep. Sky, water, a blue LED. Same failure, but you notice it after the export. Under-lit blue. This is the big one. Because the blue preset sits at roughly half the luminance of green, a dimly lit blue screen gives the key far less signal. Light the backdrop properly or shoot green. Semi-transparent things. Smoke, glass, motion blur, flyaway hair. No chroma key recovers these cleanly and ours is no exception. A wrinkled or uneven backdrop. Every fold is a gradient the key has to chase. See the lighting guide. Heavily compressed footage. 4:2:0 chroma subsampling throws away exactly the colour resolution the key depends on. A single global threshold cannot save every frame. The benchmark page shows the two fixtures where it fails outright." }, { "heading": "What we did not measure", "text": "We measured brightness falloff on green only: at tolerance 25 a green backdrop down to 0.65 relative brightness still keys out, and 0.60 flips it back to solid. We did not run the equivalent sweep on blue, so we are not going to tell you where blue falls over — we do not know. Every fixture here is a synthetic 320x240 image, not real footage. The generator, the benchmark and every CSV referenced on this page are in the public repository on GitHub, and the write-up is published on Zenodo under DOI 10.5281/zenodo.22916767. Open the green screen remover" }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) Green Screen Spill Suppression Green Screen Tolerance Explained Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-obs-settings", "title": "OBS Chroma Key Settings — What Similarity, Smoothness and Spill Do", "description": "What OBS's chroma key sliders actually change: Similarity, Smoothness and Spill Reduction, described in terms of the threshold maths they run on, with the measurements behind each one.", "markdown": "/ai/pages/obs-settings.md", "sections": [ { "heading": "OBS chroma key settings", "text": "Every chroma key exposes the same three ideas: how far from the key colour still counts as backdrop, how soft the boundary is, and how hard the key colour gets pulled out of your subject. Here is what they do, with the measurements." }, { "heading": "The three sliders, one line each", "text": "Similarity is the threshold: how far a pixel can sit from the key colour and still be called backdrop. Smoothness is the feather: how wide the band of partial transparency around that threshold is. Spill Reduction is the clean-up: it attacks the green light that bounced off the screen onto your subject, which the threshold cannot touch. They are not three ways to do the same thing: no amount of feathering fixes a wrong threshold." }, { "heading": "What OBS actually exposes", "text": "OBS's own documentation lists the Chroma Key filter as Key Color Type (default Green), Similarity (default 400), Smoothness (default 80) and Key Color Spill Reduction (default 100), plus Opacity, Contrast, Brightness and Gamma. Those defaults are quoted from the OBS chroma key filter page. One caution before copying numbers: OBS's scales are its own. A Similarity of 400 is not 40% of anything, and it is not this tool's tolerance. Learn which slider moves which idea, then tune by eye." }, { "heading": "Similarity is the threshold", "text": "Each pixel is converted to YCbCr and its distance from the key colour is measured in the Cb/Cr plane only — brightness is left out, which is why a shadow on the screen still keys out. Raising Similarity raises that threshold. The cost is not symmetric. On 320x240 synthetic fixtures, our tolerance sweep shows both sides: On a flat, evenly lit screen, IoU holds at 1.000 up to tolerance 85 and then collapses at 90, where all 21,341 subject pixels are keyed out and IoU drops to 0.722. Too much threshold does not degrade gracefully. On a shadowed screen, the default tolerance of 25 leaves 38% of the backdrop behind — recall 0.617718 — and reaches 1.000 only at tolerance 40. On a subject with a green rim, precision falls from 1.000 at tolerance 0 to 0.885 at 80: 7,206 of 21,341 subject pixels erased, because the rim is tinted with the key colour and a wider band calls it backdrop. Raise it only until the last patch of screen is gone." }, { "heading": "Smoothness is the band, not the threshold", "text": "A hard threshold gives binary alpha: a pixel is either fully backdrop or fully subject. Feathering adds a band, so pixels just past the threshold get partial transparency instead of being snapped to one side. In the core we publish, the positive half of the Edge slider does that — it raises the upper bound and leaves the threshold alone. Honest caveat: our IoU metric counts any pixel with alpha above zero as subject, so feathering barely moves the number — on the spill fixture IoU stays 1.000 for every Edge value from 0 to 100. That is a limit of the measurement, not proof that Smoothness does nothing." }, { "heading": "Spill Reduction pulls the cast out", "text": "Green light bounces off the screen onto hair, shoulders and light clothing. Those pixels are genuinely your subject, so keying them out is wrong. Spill suppression pulls the dominant channel down toward the higher of the other two. Three properties follow, all asserted in the test suite: only one channel moves, and only downward; a subject with no cast is left exactly as it was; and the channel is clamped at the higher of the other two, so it cannot tint the subject the opposite colour. We have not swept this parameter — every row in our Edge sweep has spill at 0 — so there is no number for how much is too much. Tune it against a face." }, { "heading": "The other direction: trimming instead of widening", "text": "There is a fourth move people forget, the opposite of raising the threshold: push it inward so the key eats slightly further into the subject, removing a fringe that survived. Our Edge sweep, at tolerance 25 with spill off: On the spill fixture, Edge at -100 erases 6,902 of the 21,341 subject pixels and drops IoU to 0.889 — well past useful. On the shadow-plus-spill fixture, Edge at 0 leaves 20,925 backdrop pixels behind and scores 0.622672. At Edge -30 the backdrop is fully gone but 2,053 subject pixels went with it: 0.964303. Raising tolerance to 40 instead scores 0.966269 on that same fixture and costs 1,936 subject pixels. On our hardest fixture, trimming and widening land within about 0.002 of each other and both cost roughly two thousand subject pixels. Neither is free — if one is eating hair, try the other before you write the shot off." }, { "heading": "An order that works", "text": "Fix the lighting first — most slider problems are lighting problems. See the lighting guide. Set the key colour from your actual screen. Auto-detect samples a band along the frame border and takes the median, so a subject grazing one edge does not move it — but a shadow reaching the border does, and on our fixtures the detected colour lands up to 20 levels off. Raise Similarity only until the last patch of backdrop goes. Add Smoothness next, in small amounts, watching the outline rather than the middle of the subject. Add Spill Reduction last, checking skin tone while you do it. Judge it on motion — a paused frame hides edge flicker, which is what you see in the stream." }, { "heading": "When this does not work", "text": "A shadow that reaches the edge of the frame. The shadow becomes part of the sampled border, and the detected key colour lands between the lit and shadowed values. Deep shadow and spill at once. Our hardest fixture peaks at 0.966269 and never crosses 0.99 at any tolerance, because the two sit at similar distances from the key colour. Fixing it needs a spatial method — a matte, a brush, a second pass — which a live filter lacks. Anything semi-transparent. Smoke, glass, motion blur, flyaway hair. No chroma key recovers these. Key colour inside the subject. A green key is colour-blind: a green shirt, a green logo or green eyes go transparent. An uneven or wrinkled backdrop. Every fold is a gradient, and each one wants a different setting. Heavily compressed footage. 4:2:0 subsampling discards the colour resolution the key depends on, and a streamed source is usually compressed twice. A white or grey backdrop. A neutral surface has no chroma to measure. Our neutral path switches to luminance distance and has a narrow range: 1.000 up to tolerance 50, then 0.645560 at 60, where the subject itself is keyed out." }, { "heading": "Where the numbers come from", "text": "Every figure here comes from the CSVs in the public reference repository on GitHub; the write-up is on Zenodo under DOI 10.5281/zenodo.22916767. The fixtures are synthetic 320x240 images, not real footage, from a seeded random number generator. To see the same parameters on a file you shot, the tool runs it all in your browser with no upload. The benchmark page lists what was measured; blue vs green covers switching colour. Open the green screen remover" }, { "heading": "Related video pages", "text": "Green Screen Remover (home) Green Screen for Zoom and Meetings Green Screen File Size Limits Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-spill-suppression", "title": "Green Screen Spill Suppression — Why Edges Go Green and How to Fix It", "description": "What green screen spill is, why the rim of your subject turns green, what spill removal does to the pixels, and the cases where no slider fixes it — with the measurements behind each claim.", "markdown": "/ai/pages/spill-suppression.md", "sections": [ { "heading": "Green screen spill", "text": "Spill is green light that bounced off the screen onto your subject. It is real recorded light, not a keying mistake — which is why no threshold removes it. What it is, why it lands on the rim, and how far a slider can clean it up." }, { "heading": "What spill actually is", "text": "A green screen is a lamp. Lit to a usable level, it throws green light in every direction, and some of that light lands on the person standing in front of it. The camera records it, so hair, shoulders, ears and light clothing come back with a green cast that has nothing to do with the backdrop behind them. In one sentence: spill is light on the subject, not leftover background. Those pixels are your subject, fully opaque, and they should be. Keying them out punches holes in a person, and no threshold can tell \"green because the screen lit it\" from \"green because it is the screen\"." }, { "heading": "Why the rim goes green first", "text": "Three things put the worst of it at the outline: Geometry. The rim sits closest to the screen, so it catches the most bounce. Fine detail. Flyaway hair and fuzzy fabric pack huge surface area into a few pixels, and each strand catches bounce from both sides. Shine. Skin, satin, glasses and glossy plastic reflect the screen like a mirror, and a specular reflection of a green wall is pure green. None of it is fixable with a threshold: in every case the pixel really is your subject." }, { "heading": "Two problems, one name", "text": "People say \"green edge\" for two different failures, and they need different fixes: A cast. The subject is intact, but tinted green. This is spill. A rim. A thin halo of backdrop pixels survived around the outline, or the outline is jagged. That is a threshold problem, not a spill problem. Our measurements separate them. On the fixture built to carry a green cast, default tolerance keys the backdrop perfectly — IoU 1.000, no subject pixel lost. A cast alone does not make skin transparent, so if you see a tint on an intact person, tolerance and trimming are the wrong tools and will cost you real pixels for nothing." }, { "heading": "What spill removal does to a pixel", "text": "The operation is small and deliberately limited. Take the channel your key colour dominates — green, for a green screen — and pull it toward the brighter of the other two by however much the slider asks for. Three consequences follow, all asserted in the test suite: Only one channel moves, and only downward — it can never push a pixel toward magenta. It clamps at the higher of the other two. At 100, green falls to whichever of red and blue is brighter, so a pixel never ends up less green than the grey it came from. A pixel with no cast is left exactly as it was. If green is already below the other channels, nothing happens. It runs only where alpha is above zero, so the transparent backdrop is untouched and alpha is never modified. It changes colour, never shape. The tool ships at 50. That is a starting point, not an answer: every row in our edge sweep has spill at zero, so there is no number for how much is too much. Tune it against a face and stop as soon as skin looks right — over-doing it drains the warmth out of a person." }, { "heading": "Stop the light before you fix the pixels", "text": "Spill suppression is damage control. Preventing the bounce is cheaper and looks better: Put distance between subject and screen. Bounce falls off fast; two metres of separation beats any slider. Light screen and subject separately. A screen lit on its own need not be blasted, and a front-lit subject already outruns the bounce. Add a backlight or hair light. Rim light from behind separates subject from screen and out-competes the green wrapping the edge. Kill reflection. Matte clothing, less product on skin, and tilting glasses a few degrees. Flag what you can. Black wrap beside the screen stops it throwing light sideways onto the subject. The lighting guide covers the setup in detail; the OBS settings page explains the same three sliders in live-production terms." }, { "heading": "The order that works", "text": "Fix the lighting and the distance first. Most spill complaints are a setup problem wearing a slider costume. Set tolerance only until the last patch of backdrop disappears. Touch edge next, and only for a rim or a jagged outline — not for a tint. Raise spill removal last, watching skin tone, and stop early. Judge it in motion. A paused frame hides outline flicker, which is what a viewer sees." }, { "heading": "When this does not work", "text": "Trim will not fix a cast. On the cast fixture, pulling edge to its extreme erases 6,902 of 21,341 subject pixels and drops IoU to 0.889 — you destroy a third of the person to remove a tint that spill removal handles without touching shape. Deep shadow and spill together. Our hardest fixture scores 0.622672 at default settings with 20,925 backdrop pixels left behind. Trimming to edge -30 clears them and reaches 0.964303, but takes 2,053 subject pixels. Something always pays. Our metric cannot see the fix. IoU counts any non-transparent pixel as subject, so it does not move when spill suppression works. Absence of a number here is not evidence that nothing happened. A genuinely green subject. A green shirt, a green logo or a green prop is indistinguishable from spill by colour alone, and spill removal will drain it. Specular highlights. A mirror-like reflection of the screen is saturated green. There is no channel left to recover underneath it. Semi-transparent edges. Motion blur, smoke, glass and flyaway hair mix backdrop and subject in one pixel. Spill removal assumes a pixel is one thing or the other. Heavily compressed footage. 4:2:0 subsampling throws away half the colour resolution the key depends on, and smoothing smears green into the outline before you open the file. A cast from a colour you are not keying. Spill removal targets the channel your key dominates — green for our green preset, blue for our blue one." }, { "heading": "Where the numbers come from", "text": "Every figure here comes from the CSVs in the public reference repository on GitHub; the write-up is on Zenodo under DOI 10.5281/zenodo.22916767. The fixtures are synthetic 320x240 images, not real footage, built from a seeded random number generator — treat the numbers as the behaviour of the algorithm, not a promise about your shoot. To try the sliders on a file you shot, the tool runs it all in your browser with no upload. The benchmark page lists what was measured, and blue vs green covers switching key colour. Open the green screen remover" }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) Green Screen Tolerance Explained Chroma Key Artifacts Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/transparent-webm-explained", "title": "Transparent WebM Explained — Which Players and Editors Keep the Alpha", "description": "What a transparent WebM is, how VP9 carries an alpha channel, where the transparency gets dropped, why the file has no duration, and how to test whether the alpha survived.", "markdown": "/ai/pages/transparent-webm.md", "sections": [ { "heading": "Transparent WebM", "text": "A WebM is a Matroska container holding VP8 or VP9 video. Both of those codecs can carry a fourth value per pixel — how opaque it is — and most trouble comes from that value being dropped between the encoder and the screen. What it is, where it survives, how to check." }, { "heading": "What the alpha channel actually is", "text": "Ordinary video stores three numbers per pixel: red, green and blue. A transparent WebM stores a fourth, the alpha, running from 0 (fully invisible) to 255 (fully solid). It rides alongside the colour planes in the same bitstream, which is why the codec matters: VP8 and VP9 have an alpha mode, and a WebM written with a codec that does not is just a rectangle. The values in between are the ones worth caring about. A cut-out that only knows “there” and “gone” is what gives keyed footage its cut-with-scissors outline. This keyer produces a continuous alpha: inside the tolerance band, a pixel’s opacity is proportional to how far its colour sits from the key colour, so hair, motion blur and soft fabric keep a graded edge. The band only exists when the Edge slider is above zero. At zero and below it closes, and every pixel snaps to fully on or fully off." }, { "heading": "Three places the alpha gets dropped", "text": "At encoding. The encoder has to be asked for a codec with an alpha mode. If it is not, the frames are flattened onto a background before the first byte is written, and nothing downstream can recover what was never stored. At playback. A player that ignores alpha composites the frames against something, usually black. The file is intact; the player is not reading it. On upload. Any service that re-encodes to H.264 destroys the alpha on the way in. If a platform does not document an alpha-friendly format, assume the transparency will not survive." }, { "heading": "What this tool writes", "text": "Exporting a transparent WebM here is a recording, not an encode. The page draws each keyed frame onto a canvas, calls captureStream at 30 fps on that canvas, and hands the stream to MediaRecorder. Concretely, the export: asks for video/webm;codecs=vp9, and falls back to plain video/webm where that string is not supported; targets about 8 Mbps; rounds the frame dimensions down to an even number, which is what VP9 requires; runs in real time — there is no video encoder in the page, so a sixty-second clip takes about sixty seconds; saves as -transparent.webm. Without MediaRecorder or captureStream, the WebM button is hidden rather than offering an option that cannot work; the PNG sequence is the fallback. The video page covers the same exports from the shooting side." }, { "heading": "The file has no duration, and why that bites", "text": "A WebM written by MediaRecorder carries no duration in its header. The browser reports the length as infinity until something forces a scan to the end, and an editor that trusts the header will show an unknown length, refuse to scrub, or import a truncated clip. This is a container artefact, not damaged video. Remux rather than re-encode: copying the streams into a fresh container leaves the alpha alone, while any re-encode is a chance to lose it. Play the file once in a browser first — that is enough to make the real length appear in most tools that read it afterwards." }, { "heading": "How to check whether the transparency survived", "text": "Put the file on a page or a canvas with a coloured background. A subject that floats over the colour has alpha; a subject sitting on black does not. Import it into an editor and drop another clip underneath. Same test, better lighting. If you have ffprobe, read the pixel format of the video stream. An entry with an alpha component means the channel is there; a plain yuv420p means it is not. Do the test on the machine and in the version you are actually shipping to. “It played fine for me” on one browser says nothing about the player your client uses." }, { "heading": "Where it plays, and where it does not", "text": "We have not run a compatibility matrix across players and editors, so this page reports no playback percentages. What follows is the practical shape of the problem, and every third-party entry is something to verify in the version you use: Browsers. Chrome, Edge and Firefox read WebM and composite the alpha when the video is played in a page. That is the environment this format was built for. Desktop editors. Plenty of non-linear editors do not import WebM at all and expect you to supply something else — and a transcode is exactly where the alpha dies. If the result is going onto a timeline, export the PNG sequence instead. Safari and iOS. WebM support there has been uneven across versions. Check on the target device rather than assuming. Social and messaging platforms. Unless a platform documents an alpha-friendly format, plan for the upload to be re-encoded and flattened." }, { "heading": "When this does not work", "text": "You need an MP4. There is no transparent MP4. No setting here or anywhere else produces one. Your target re-encodes uploads. Composite the clip onto the background you want before uploading, instead of relying on alpha to survive. Your player ignores alpha. Verify over a coloured background before concluding the export failed. Your editor will not import WebM. Take the PNG sequence — it is larger, but nothing can silently flatten it. It is capped at 1200 frames here, about 100 seconds at the default 12 fps. The duration reads as unknown. Remux with stream copy; do not re-encode to fix it. Your browser cannot record a canvas. There is no WebM option and the PNG sequence is the fallback. Long clips. Real-time recording means a long wait, and a frame dropped during recording is gone — there is no re-encode stage to recover it. The alpha is only as good as the matte. Uneven lighting, compression and a badly lit screen all produce a bad matte, and no container adds detail the key never had. The slider page and the benchmark cover that side. We have no measured player or editor numbers. The figures published in our reference repository are keying quality on synthetic fixtures, not container or player compatibility. There is no honest number to quote here, so none is quoted." }, { "heading": "Where our published numbers come from", "text": "The measurements behind the keying claims on this site live in the CSVs of the reference repository on GitHub, and the write-up is archived on Zenodo under DOI 10.5281/zenodo.22916767. Those fixtures are synthetic 320x240 images, not real footage. They measure how well an algorithm separates a subject from a backdrop; they say nothing about containers, codecs or players, which is why this page carries none of those numbers. To make one, the tool keys the backdrop and records the result entirely in your browser — no upload, no account, no watermark. The transparent video page compares the formats side by side. Open the green screen remover" }, { "heading": "Related video pages", "text": "Green Screen Remover (home) PNG Sequence vs Transparent WebM Green Screen Video Export Settings Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/png-sequence-vs-webm", "title": "PNG Sequence vs Transparent WebM — Which Export to Pick", "description": "PNG sequence or transparent WebM: what each one keeps, why the sequence is lossless and heavy, why WebM is small and lossy, the 1200-frame ceiling, the real-time recording limit, and how to choose between them.", "markdown": "/ai/pages/png-vs-webm.md", "sections": [ { "heading": "PNG Sequence vs Transparent WebM", "text": "Both exports carry the same alpha channel out of the same keyed frames. They differ in what happens to those frames on the way to disk: one writes every pixel exactly as the keyer produced it, the other compresses them into a video. Which one you want depends on what happens next, not on which is “better”." }, { "heading": "The short answer", "text": "If the result is going onto a timeline in an editor, take the PNG sequence. Nothing can silently flatten it, every frame stays lossless, and any editor that can import an image sequence can read it. If the result is going onto a web page, into a browser-based compositor, or anywhere a single file is easier to move around, take the transparent WebM — it is one small file that plays where it was made." }, { "heading": "What the PNG sequence actually is", "text": "The page steps through your clip frame by frame at the rate you set, seeks the video to each timestamp, keys that frame, encodes it as a PNG, and packs the whole set into a ZIP named -frames.zip. Inside, the frames are numbered _0000.png, _0001.png and so on. Two properties matter. Each frame is written at the full resolution of your source — no rounding, no scaling. And PNG is lossless, so the alpha the keyer computed is the alpha you get, byte for byte. The ZIP is written stored, not deflated, so it is essentially the sum of the PNG sizes: compressing them again buys nothing worth the CPU time." }, { "heading": "What the transparent WebM actually is", "text": "The WebM export is a recording, not an encode. The page draws each keyed frame onto a canvas, calls captureStream on that canvas at 30 fps, and feeds the stream to MediaRecorder with VP9 at about 8 Mbps. The file lands as -transparent.webm. So it is a video: one file, small, playable in a browser tab. Two consequences follow from it being a recording. It runs in real time — a sixty-second clip takes about sixty seconds, because there is no faster-than-realtime encoder in the page. And a WebM written this way carries no duration in its header, which is why some editors show an unknown length until the file is played or remuxed. The WebM page covers that artefact." }, { "heading": "Where they really differ", "text": "Lossy versus lossless. VP9 drops detail to hit its bitrate. PNG does not. Fine hair and a graded edge are the detail a codec spends its bits on last. Resolution. VP9 needs even dimensions, so the recording canvas rounds width and height down. The sequence keeps the original pixel grid. Frame rate. The sequence honours your rate, default 12 fps. The WebM records at 30 fps regardless. File count. The sequence is many images inside one ZIP; the WebM is one file. For handing something over, that is often the whole argument. What can go wrong silently. A sequence has nothing to misread — an editor either imports the images or it does not. A WebM must survive a player that reads alpha, an importer that trusts its duration header, and any re-encode in between." }, { "heading": "The ceiling you will hit with long clips", "text": "A PNG sequence lives in memory before it is zipped, so it is capped at 1200 frames. At the default 12 fps that is about 100 seconds; at 24 fps, about 50. Go over and the export refuses, and the message names the frame rate that would fit. Your options: lower the rate, trim the clip, or take the WebM — which has no frame cap but costs real time equal to the clip length. The projected frame count sits next to the rate slider before you start." }, { "heading": "Which one to pick", "text": "Going into Premiere, Resolve, Final Cut or After Effects? PNG sequence. Import the folder as an image sequence and drop it on the timeline. Going onto a web page or into a browser demo? Transparent WebM. It is what the browser made it for. Going to a platform that re-encodes uploads? Neither will keep its alpha. Composite onto the background you want first. A single photo? Neither — take the plain transparent PNG, which is the photo export at full resolution. Not sure yet? Sequence. You can always encode it into a WebM later; you cannot recover the pixels a lossy encode already dropped." }, { "heading": "When this does not work", "text": "You need one file and lossless. There is no such option here. A sequence is many files; a WebM is compressed. Your clip is longer than the frame ceiling allows. Lower the frame rate or trim. Raising the ceiling is not an option — the frames have to sit in memory before the ZIP is written. You are in a hurry with a long clip. The WebM records in real time, so a ten-minute clip is a ten-minute wait, and a frame dropped mid-recording is gone. Your editor will not import WebM and shows an unknown duration. Use the sequence, or remux the WebM with stream copy — never re-encode to fix the header, that is where the alpha dies. Your target flattens uploads. No export format protects you from a service that re-encodes to H.264. Your browser cannot record a canvas. There is no WebM button at all in that case; the sequence is the fallback and the page says so rather than offering a broken option. The matte itself is fragile. If the key is already eating into the subject, exporting lossless will not repair it — it will just show you the damage more clearly. On our synthetic fixtures, tightening the edge slider to −100 removed 6902 of 21341 subject pixels, dropping IoU to 0.889. The slider page explains what that control does. We have no measured size or quality comparison. No bitrate table, no “sequence is N times larger”, no measurement of how much VP9 degrades a soft edge. We have not run it, so no number is quoted." }, { "heading": "Where our published numbers come from", "text": "The measurements behind the keying claims on this site live in the CSVs of the reference repository on GitHub, and the write-up is archived on Zenodo under DOI 10.5281/zenodo.22916767. Those fixtures are synthetic 320x240 images, not real footage, and they measure how well a key separates a subject from a backdrop — not file size, not codec quality, not player support. One caveat worth knowing before you read an alpha number off them: in the shadow sweep, every published alpha value is either 0 or 255. Those are binary mattes, so they cannot tell you anything about how a graded, semi-transparent edge behaves. That is precisely the part a lossy encode touches first, and we have no measured data on it. To make either export, the tool keys the backdrop and writes the result entirely inside your browser — no upload, no account, no watermark. The format comparison covers what each container can carry, and the video page walks through the exports from the shooting side. Open the green screen remover" }, { "heading": "Related video pages", "text": "Green Screen Remover (home) Green Screen Video Export Settings Green Screen for YouTube Shorts Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-tolerance-explained", "title": "Green Screen Tolerance Explained — What the Slider Actually Changes", "description": "What the tolerance slider on a chroma key really does: it moves one threshold measured in the Cb/Cr plane, in units you can read, plus the measurements that show where raising it fixes a shadowed screen and where it eats your subject.", "markdown": "/ai/pages/tolerance.md", "sections": [ { "heading": "Green screen tolerance, explained", "text": "Every chroma key has a tolerance control and almost none say what it does. This one moves a single distance threshold — which is why turning it up can rescue a badly lit screen or quietly erase your subject." }, { "heading": "The one number it moves", "text": "Tolerance does not widen the green range in any visual sense. It sets a radius. The slider maps onto a threshold called lo: lo = (tolerance / 100) × scale The scale is 118 for a green or blue screen and 150 for a neutral backdrop — a white or grey wall, where there is no chroma to measure and the keyer falls back to luminance. At the default tolerance of 25 that is 29.5 for a chroma key and 37.5 for a neutral one. Those are not percentages or degrees; they are distances in the colour plane, and everything below uses the same units." }, { "heading": "Distance from what, measured how", "text": "Each pixel is converted to YCbCr with the BT.601 coefficients, and only Cb and Cr are kept. Brightness is discarded on purpose: a shadow across the screen is the same paint with less light on it, and in RGB that moves the pixel a long way even though its colour has not changed. The distance is the straight-line gap between the pixel's colour point and the key colour's point. At or inside lo a pixel is background and gets alpha 0; at or beyond a second threshold hi it is fully opaque; in between, alpha is partial. With the edge control at 0, hi equals lo, so there is no soft band — tolerance alone gives a hard cut. Softening is the edge control's job." }, { "heading": "When turning it up helps", "text": "A shadowed screen is the one case where raising tolerance is correct. On shadow-green — a soft shadow across part of the backdrop, same hue, less light — the score at the default is IoU 0.617718, with 21,201 backdrop pixels surviving. Raise tolerance and it climbs: tolerance 25 → IoU 0.617718 (21,201 backdrop pixels left behind) tolerance 30 → IoU 0.671108 (18,240) tolerance 35 → IoU 0.770641 (12,720) tolerance 40 → IoU 1.000000 (clean) The subject sits further out than the shadowed backdrop, so nothing is lost on the way." }, { "heading": "When turning it up eats the subject", "text": "The threshold does not know what is backdrop and what is subject. It only knows distance. The subject in these fixtures is a skin tone about 105.94 units from the green key colour. At tolerance 85 the radius is 100.3, still short of it. At tolerance 90 it is 106.2 — past the subject — and on flat-green all 21,341 subject pixels are keyed out. IoU drops from 1.000000 to 0.722122. A neutral backdrop fails sooner, the scale being larger. A grey prop on a near-white wall sits 88 units of luminance away. The radius passes that at tolerance 60 (0.60 × 150 = 90), and from there the whole subject goes transparent: IoU 0.645560, unchanged to the top of the slider." }, { "heading": "The spill case: a peak, not a direction", "text": "On spill-green, where light bounced off the screen onto the rim of the subject, the cost is graduated. IoU is 1.000000 at tolerance 25, then 0.991136 at 30 (496 subject pixels gone) and 0.978579 at 35 (1,214 gone). The rim is closer to the key colour than the rest of the subject, so it goes first; the spill page covers why it exists. With both problems at once (shadow-spill-green, plus sensor grain) the shape is obvious: 0.622672 at 25, 0.869190 at 35, 0.966269 at 40 (1,936 lost), 0.955053 at 45 (2,610 lost), 0.777652 at 90 (15,857 lost). The best result sits at 40 and going past it makes things worse: tolerance has an optimum, it is not a control you push until the screen looks clean." }, { "heading": "Why blue survives more of it", "text": "That skin tone sits 118.19 units from the blue key colour rather than 105.94. On flat-blue the keyer still scores IoU 1.000000 at tolerance 100, where the radius is exactly 118.0 — a margin of 0.19 units. That is arithmetic luck at the top of the slider, not a guarantee: another subject colour would not get it, and it says nothing about which colour to shoot on." }, { "heading": "How to set it on a real clip", "text": "Start at the default 25 and judge the result on the backdrop, not the subject. Raise it a notch at a time and stop the moment the last of the screen is gone. Every notch past that only buys a thinner subject. If you need past roughly 40 to clear shadow, the shadow is the problem. Relight instead — that is what the lighting guide is for. Tolerance is one of three controls: it decides what counts as backdrop, edge decides how the boundary is feathered or trimmed, spill pulls key colour out of the subject. Moving one without watching the others is where bad keys come from." }, { "heading": "Where these numbers come from", "text": "Every figure comes from data/tolerance-sweep.csv in the chroma-key-reference repository, and the method is written up at DOI 10.5281/zenodo.22916767. The fixtures are synthetic: 320×240 flat colour fields with an ellipse subject in a skin tone, generated deterministically so the sweep is reproducible. They are not real footage. A real clip has grain, compression, motion blur and a subject of many colours, so read these numbers as the shape of the behaviour, not as settings to copy." }, { "heading": "When this does not work", "text": "A subject close to the screen in colour. A green shirt on a green screen is not a tolerance problem. Any radius that reaches the screen reaches the shirt. Deep shadow or a badly uneven screen. Tolerance absorbs a moderate falloff — the shadow fixture cleans up at 40 — but past that the radius reaches the subject first. Relight the backdrop. Transparent or semi-transparent subjects: smoke, glass, water, flyaway hair over the screen. Distance gives one answer per pixel, and a pixel that is half backdrop and half hair does not have one. Compression artefacts at the edge. Chroma subsampling smears the screen's colour into the edge, so the artefact sits close to the key colour and raising tolerance eats edge with it. A neutral backdrop with a mid-grey subject. The luminance fallback has no chroma to work with, and 88 units of luma is thin margin once the wall is unevenly lit. Motion blur. A blurred edge is a blend of subject and screen, and each blend value maps to a different distance. No single radius treats the band uniformly. Trying to fix spill with tolerance. It removes the rim by deleting those pixels, which shows up as a chewed edge in hair. Spill suppression changes colour; tolerance deletes pixels. Pushing the slider to the top. Nothing in the sweep improves all the way to 100: the best scoring tolerance here is 40, and the flat cases collapse at 60 or 90." }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) Chroma Key Artifacts Chroma Key Benchmark Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/remove-white-background-without-green-screen", "title": "Remove White Background Without a Green Screen — What Actually Works", "description": "No green screen? What a chroma key can and cannot do with a white wall: why a neutral backdrop falls back to a luminance key, the margin that leaves you, and where it stops working.", "markdown": "/ai/pages/white-background.md", "sections": [ { "heading": "Remove a white background without a green screen", "text": "A chroma key needs colour to separate on, and a white wall has none. This is what happens instead: the keyer gives up on colour and measures brightness, which works — until the wall stops being flat." }, { "heading": "Why a chroma key has nothing to hold on to", "text": "A chroma key does not separate subject from backdrop by brightness. It converts each pixel to YCbCr, discards the brightness, and measures the distance from the key colour in the two colour-difference channels. Discarding brightness is deliberate: a shadow on a green screen is the same paint with less light, and in colour terms it has not moved. A white wall breaks that. In key-presets.csv the white, black and grey backdrops all carry Cb 0.0000 and Cr 0.0000. Every pixel on a neutral backdrop — wall, subject, shadow, highlight — lands on the same point in the colour plane however light or dark it is. A distance in that plane is zero everywhere, and a threshold on zero sorts nothing. That is not a tuning problem. baseline-comparison.csv runs a naive green-dominance keyer on the neutral fixture with the best threshold available: IoU 0.000000. Not weak — empty. There was never a signal." }, { "heading": "What runs instead: a luminance key", "text": "The keyer checks the key colour's chroma magnitude first. Under 7, which is true for white, black and grey, it stops measuring colour and measures brightness: d = |luma(pixel) − luma(key)| Pixels inside a radius of the wall's brightness go transparent; everything further out stays. One number per pixel — the only lever that can tell a white wall from a white shirt, and what the tool falls back to it for. One other thing changes, and it matters if you move the slider. The radius is lo = (tolerance / 100) × scale, and the scale is 118 for a colour key but 150 for a neutral one. At the default 25 that is 29.5 on green and 37.5 here: the same slider position, covering more ground. The tolerance page has the full sweep." }, { "heading": "It works, right up to the point where it doesn't", "text": "The neutral-white fixture is a flat #f2f2f2 field with a solid #9a9a9a rectangle in front of it, 320×240. At the default tolerance the score is IoU 1.000000: 49,579 backdrop pixels keyed out, no subject pixels lost. Auto-detection reads the wall off the frame border and returns #f2f2f2 exactly — max channel delta 0. The two greys are 88 units of luma apart: the wall is 242, the prop is 154. One number, one gap. That makes the sweep a step, not a curve: tolerance 0 through 55 → IoU 1.000000, every time tolerance 60 → IoU 0.645560 — all 27,221 subject pixels keyed out tolerance 65 through 100 → 0.645560, unchanged The arithmetic explains the cliff. 0.60 × 150 = 90, and 90 is past 88. Both surfaces are flat, so every subject pixel sits at the same distance and all cross together — and raising tolerance only pushes the radius outward, never pulling them back. Against a green screen the difference is the whole story. On shadow-green the score climbs gradually — 0.617718 at 25, 0.671108 at 30, 0.770641 at 35, 1.000000 at 40 — because a shadowed screen spreads the backdrop across a range of distances. A flat neutral backdrop has no range: perfect or nothing, with one abrupt edge between." }, { "heading": "The margin you are actually working with", "text": "88 units out of 255 is about a third of the range, and the default radius of 37.5 uses less than half of it. On a flat synthetic wall that is comfortable. A real wall is not flat — it falls off towards the corners, takes the subject's shadow, and picks up whatever the room bounces onto it. Each of those eats into the same 88 units, and the subject sits at the near end of them. A black backdrop runs the same code with the distances flipped: key-presets.csv lists black-wall at luma 0.0 and grey-card at 128.0, both neutral, both with the 37.5 default. Which one you want stops being a colour decision — it is whichever end of the brightness range your subject sits further from." }, { "heading": "How to shoot so the key has something to measure", "text": "Light the wall evenly, on its own, brighter than the subject. Evenness is the whole game: a falloff across the wall eats the margin faster than anything else. The lighting guide is written for a green screen, but lighting the backdrop separately applies as-is. Put distance between subject and wall. A few feet back kills the shadow the subject throws on the backdrop — the one artefact a luminance key cannot tell from a darker subject. Separate the subject in brightness, not colour. Colour is discarded. A mid-grey prop on a white wall works; a white shirt on one, no setting will save. Light subject, dark backdrop. The same fallback runs, and the gap opens up at the other end. Watch for highlights. The distance is absolute, so a specular highlight brighter than the wall counts as close to it. Export as PNG. The result rides on an alpha channel, so the format has to carry one — the photo page covers what that means for the download." }, { "heading": "Where these numbers come from", "text": "Every figure comes from data/key-presets.csv, data/baseline-comparison.csv, data/auto-key-accuracy.csv and data/tolerance-sweep.csv in the chroma-key-reference repository. The method is written up at DOI 10.5281/zenodo.22916767. The fixtures are synthetic: 320×240 flat colour fields with a rectangle or ellipse subject in front of them, generated deterministically so the sweep reproduces exactly. They are not real photographs, and a real room has uneven light, sensor grain, compression and a subject made of dozens of colours. Read these as the shape of the behaviour, not as settings to copy." }, { "heading": "When this does not work", "text": "A light subject on a light wall. A white shirt, white packaging or pale skin against a white backdrop leaves almost no luma gap. No setting fixes it — change the backdrop or the lighting. An unevenly lit wall. The flat fixture is perfect to tolerance 55 because every backdrop pixel sits at the same distance. A real wall's darker corners move closer to a mid-grey subject than the radius expects. Shadows the subject casts on the backdrop. A shadow is darker, and darker is the direction the subject sits in. Move the subject away from the wall instead of raising tolerance. Colour differences alone. A yellow prop and a grey prop can share a luma value, and the luminance key never looks at colour. Specular highlights. The distance is absolute, so anything brighter than the key by less than the radius is treated as backdrop, and highlights punch holes in the subject. Fine detail over the backdrop — hair, fur, fabric edges, smoke, glass. One distance per pixel gives one answer per pixel, and a pixel that is part hair and part wall does not have one. Pushing tolerance to rescue a bad shot. On the neutral fixture there is no gradual zone: 55 is perfect and 60 has erased the whole subject. Expecting the spill control to help. Spill suppression pulls the key colour's dominant channel out of the subject, and a neutral key has all three channels equal — there is nothing to pull." }, { "heading": "Related image pages", "text": "Green Screen Remover (home) Transparent PNG Alpha Keying Hair and Transparent Edges Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-for-youtube-shorts", "title": "Green Screen for YouTube Shorts — Shoot Vertical Footage That Keys Clean", "description": "How to shoot and key green screen footage for vertical short video: why the tall frame makes the lighting worse, the 1200-frame ceiling on PNG export, and the shots that never come out clean.", "markdown": "/ai/pages/youtube-shorts.md", "sections": [ { "heading": "Green screen for YouTube Shorts: shoot vertical footage that keys clean", "text": "A Short is a tall frame, shot close, usually on a phone. All of that makes the key harder. Here is what goes wrong, what the measurements say, and how to export a clip that survives the upload." }, { "heading": "What the tall frame does to the key", "text": "A Short is a 9:16 vertical frame — 1080×1920 is the usual target — and Google's own help page puts the length ceiling at three minutes. That shape changes the shot before it changes anything technical: to fill a tall frame, the camera comes in close, which puts the subject near the backdrop. Distance is what keeps their shadow and the screen's bounce off them, and the vertical frame spends it. It also leaves less backdrop on screen: in a wide shot the green fills the sides, in a tight vertical shot it is a narrow strip the keyer must read the colour from. The tool samples the frame border to pick the key colour. When that strip is partly shadowed, the sample describes the shadow rather than the screen — on shadow-green, auto-detection returns #00a23b against a true #00b140, a max channel delta of 15. Not a misread: the border really is darker than the middle, so the key colour fits the darkest part of the screen." }, { "heading": "The shadow is the whole problem", "text": "A chroma key works in the two colour-difference channels and discards brightness, which is why a shadow on the screen usually still keys out. \"Usually\" has a limit. Dimmed to 0.65 of its lit brightness, the screen sits 29.3412 from the key and still comes out transparent at the default tolerance; at 0.60 the distance is 33.5337 and the pixel flips to fully opaque. The default radius is 29.5. Past roughly two-thirds brightness, an unevenly lit screen starts coming back as solid green. Tolerance is the fix people reach for, and it works: shadow-green goes from IoU 0.617718 at tolerance 25 to 1.000000 at 40. On a screen that was already clean, the same slider does the opposite — flat-green at tolerance 90 keys out all 21,341 subject pixels and falls to 0.722122. It widens one radius around one colour and cannot tell a shadowed screen from a green shirt. Light the screen flat and use tolerance as a small correction: the lighting guide and the tolerance page cover both." }, { "heading": "Spill is more visible here, not more likely", "text": "The other artefact is green light bouncing off the screen onto skin, hair and light clothing. A vertical Short shows it off: the subject is large in the frame, so a rim two pixels wide in a wide shot is a visible halo at phone size. What the measurements say is narrower than most advice. On spill-green, the shipped defaults score IoU 1.000000 — spill does not make the subject transparent. It is a colour problem, and a coverage metric cannot see it. Scrubbing the rim with the edge control does cost pixels: at edge −100, 6,902 of 21,341 subject pixels go and the score drops to 0.889322 — on a close-up, a bite out of a chin or a hand. The spill page covers what that control separates." }, { "heading": "How to shoot it", "text": "Light the screen on its own, evenly. Evenness is the whole game — the 0.65 cliff is a lighting failure, not a settings failure. Put distance behind the subject. It removes both the shadow on the screen and the bounce onto the subject. Leave green at the sides. The key colour is sampled from the border; if the subject touches both edges there is nothing clean to sample. Nothing green on the subject. Clothing, phone cases and props — the key separates by colour alone. Lock exposure and white balance. A phone that re-metres mid-take shifts the backdrop under a key set once. Keep the camera still. Motion blur smears the edge the key has to decide on. Record at the best quality the phone offers. Compression noise around hair is extra work for a key deciding one pixel at a time." }, { "heading": "Exporting: the frame ceiling and the clock", "text": "Two limits decide which export you can use, and both bite on longer clips. A PNG sequence is held in memory before it is zipped, so it is capped at 1,200 frames. The frame-rate control sets how many frames get pulled out of your clip — it has nothing to do with what the camera shot. At 30 fps, a 60-second Short wants 1,800 frames, over the ceiling. At 20 fps it is exactly 1,200; the default 12 fps gives 720 frames of the same minute. The tool warns you and suggests a rate that fits rather than failing at the end. A transparent WebM has no frame ceiling, but it records in real time: the keyed canvas is captured at 30 fps into VP9 at 8,000,000 bits per second, dimensions rounded down to even numbers. A one-minute clip takes about a minute; a backgrounded tab can stall it. For a Short that usually means WebM past about 40 seconds, the sequence for anything short you want lossless. The sequence-versus-WebM page goes through the trade-off, and the video page walks through the export." }, { "heading": "Where these numbers come from", "text": "Every figure comes from data/shadow-distance.csv, data/tolerance-sweep.csv, data/edge-sweep.csv, data/auto-key-accuracy.csv and data/key-presets.csv in the chroma-key-reference repository. The method is written up at DOI 10.5281/zenodo.22916767. The fixtures are synthetic: 320×240 flat colour fields with a rectangle or ellipse subject in front of them, generated deterministically so every sweep reproduces exactly. They are not real phone footage — a real room has uneven light, rolling shutter, sensor noise, heavy compression and a subject made of dozens of colours. Read these as the shape of the behaviour, not as settings to copy." }, { "heading": "When this does not work", "text": "A screen lit unevenly enough to fall below about two-thirds brightness. Past that the shadowed area stops keying, and no amount of tolerance brings it back without taking the subject with it. The subject wearing or holding anything close to the key colour. Colour is the only lever. A green shirt goes transparent before the backdrop finishes. Handheld motion blur on the edge. A smeared boundary pixel is half subject and half screen, and one pixel gets one answer. Fine detail over the backdrop — curly hair, fur, chiffon, smoke, glass, a spinning fidget spinner. Every one of them is the same failure at small scale. A clip longer than the frame ceiling if you need a PNG sequence. 1,200 frames is the cap; over it, lower the frame rate, trim the clip, or use WebM. Expecting WebM export to finish faster than the clip runs. It records in real time from playback, so a three-minute Short takes about three minutes. Aggressive edge cleanup to kill a spill rim. On the spill fixture that costs 6,902 subject pixels. Move the subject off the screen instead. Relying on auto-detection in a tight frame. It samples the border, and in a close vertical shot the border is where the falloff is. Pick the key colour by hand if the screen is not flat." }, { "heading": "Related video pages", "text": "Green Screen Remover (home) OBS Chroma Key Settings Green Screen for Zoom and Meetings Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/diy-green-screen-at-home", "title": "DIY Green Screen at Home — Materials, Size and Wrinkles That Actually Key", "description": "How to set up a green screen at home that actually keys clean: which materials work, how big to make it, why wrinkles are the one thing that wrecks the matte, and what a DIY screen cannot fix.", "markdown": "/ai/pages/diy.md", "sections": [ { "heading": "DIY green screen at home", "text": "You do not need a studio. You need the right fabric, enough of it, and not a single wrinkle. This page covers what to buy, how big to make it, and why folds wreck your key — with the measurements behind each claim." }, { "heading": "The material matters less than the wrinkles", "text": "The keyer does not care whether you bought a branded pop-up or taped green card to the wall. It cares about one thing: is every pixel of the backdrop the same colour. A cheap bedsheet pulled tight and lit evenly will key cleaner than professional muslin that is bunched up and lit from one side. The chroma key measures distance from the key colour in the Cb/Cr plane. On a flat green fixture the key colour is #00b140, chroma magnitude 83.66, luma 111.20. At full brightness every pixel sits at distance 0 and is removed. As the screen darkens — a shadow, a fold — the distance grows. At 65% of full brightness the distance is 29.34, still inside the default tolerance radius of roughly 29.5, so the pixel still keys out. At 60% the distance reaches 33.53, past the threshold, and the pixel turns solid. That is a 5% brightness drop. One fold can be more than that." }, { "heading": "What to buy", "text": "Fabric (muslin or polyester). The cheapest option. Look for matte green with no sheen. Polyester holds its shape; muslin is heavier and drapes flatter. Either keys fine flat. A green double bedsheet covers a head-and-shoulders shot; 3 metres covers a seated person. Pop-up collapsible screen. A wire-framed disc that folds into a bag. Flat by design, which solves wrinkles for you. Most are 1.5 to 2 metres — enough for one person, not a wide shot. Green paint on a wall. The flattest surface you can get and never needs folding. Use pure green with no blue or yellow tint. The preset on this tool is #00b140. Green paper or card. Poster board works for small product shots; butcher paper on a roll lays flatter. Avoid glossy paper: the sheen reflects light as a white hotspot the key cannot remove." }, { "heading": "How big to make it", "text": "Big enough that the subject and their shadow do not spill past the edge. Measure the widest the subject will be, add 50 cm per side, and the same above their head. For someone at a desk, 2 m wide by 1.5 tall is workable. For standing shots, 3 m wide and 2.5 tall. If the subject gestures past the edge, that part of the arm is gone — there is no backdrop to key, and the tool has no brush to paint it back. Do not confuse screen size with subject distance. Farther from the screen means less green bounce on the subject, but the screen stays the same size. The trade is distance against framing: the screen must cover every pixel behind the subject, including where their shadow falls." }, { "heading": "Wrinkles: the one thing that wrecks it", "text": "A wrinkle is a shadow you cannot light away. Each fold catches light on one face and casts a shadow on the next, so the fabric has two or three brightness levels side by side. The keyer sees them as different colours, and the dimmer ones fall past the threshold. On a synthetic shadow fixture — a smooth ramp with no fabric texture or camera noise — the default tolerance of 25 scores a mean IoU of 0.618: roughly two thirds of the backdrop left in the frame. Raising tolerance to 40 takes it to 1.000. Same image, no code change: the shadow was a lighting artefact the keyer was compensating for. The cost is fine detail — hair, translucent fabric and motion-blurred edges start disappearing. Kill wrinkles before you shoot; raise tolerance only as a last resort. Three ways: Stretch it. Clamp the fabric top and bottom, or use a frame. Tension removes most folds. Steam it. A garment steamer relaxes muslin or polyester in minutes. Weight the bottom. Sew a channel at the bottom edge and thread a dowel through it. Gravity pulls the fabric flat. If none of those work, switch to a pop-up or paint the wall. A wrinkled screen is a material problem, not a slider problem." }, { "heading": "Check it before you record", "text": "Take one still with the subject in place and drop it into the tool. Toggle the checkerboard and compare corners against the centre: if the screen keys in one place and not another, you have a wrinkle or lighting problem, not a tolerance problem. Check the auto-detected key colour: on a shadowed screen it reads darker — in our fixture run it lands on #00a23b where the screen is #00b140, a channel difference of 15. Small enough to key fine; for the exact colour, pick it by hand off a lit part of the screen. The lighting guide goes deeper on even lighting: two lights from each side at 45 degrees, aimed only at the backdrop, subject lit separately." }, { "heading": "When this does not work", "text": "A DIY screen solves material, size and wrinkles. It does not solve these: A room too small for distance. Subject close to the screen: their shadow lands on the backdrop, green bounce lands on them. Distance fixes both. A blue backdrop reflects less light — see the blue vs green page. Semi-transparent detail. Flyaway hair, fur, veils, smoke. A colour key produces one matte; these need a hand-drawn mask. There is no brush on this tool. A subject wearing green. Green clothing keys out with the screen. More tolerance makes it worse. Shoot on blue — see the blue screen page. Wrinkles you cannot get out. If the fabric will not steam flat, a pop-up or painted wall is the only fix. No tolerance setting saves a deeply creased screen. Bad light, not bad fabric. A flat screen lit from one side has the same gradient as a wrinkle. Fix the light — see the lighting page. Heavy compression. A phone recording smears colour across the boundary before the keyer sees it. See the video page. Glossy fabric. Satin reflects light as white hotspots the key cannot separate from any other white. Choose matte. A green that is not green enough. Teal, lime and olive sit at different points in the Cb/Cr plane. Auto-detect picks the wrong colour; set the key by hand." }, { "heading": "Then key it locally", "text": "Once the screen is up and flat, drop the clip or still into the tool and the backdrop is keyed out in the browser — no upload, no account, no watermark. The no-upload page covers what keeping the file local avoids. The fixture data is in the reference repository, and the write-up is archived at doi.org/10.5281/zenodo.22916767 — synthetic test fixtures, not real footage, so read them as the shape of the problem. Open the green screen remover" }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) Blue Screen vs Green Screen Green Screen Spill Suppression Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-video-export-settings", "title": "Green Screen Video Export Settings — Resolution, Frame Rate and Alpha", "description": "What resolution, frame rate and alpha a green screen export actually keeps: the three formats this tool writes, what each one does to your source, and the settings it does not offer.", "markdown": "/ai/pages/export-settings.md", "sections": [ { "heading": "Green screen video export settings", "text": "There is no settings dialog here. There are three export formats, and each makes a different set of promises about resolution, frame rate and alpha. This page is what those promises actually are — read off the code that does the writing, not off a spec sheet." }, { "heading": "The three things you can export", "text": "Transparent PNG — a still, keyed at the source image's native width and height. Alpha kept, losslessly. PNG sequence (.zip) — one PNG per frame, at the source video's native size. Alpha kept on every frame. Transparent WebM — a real-time canvas recording with an alpha channel, encoded as VP9 where the browser supports it. There is no MP4 export: the MP4 profiles that play everywhere carry no alpha channel, so an MP4 of a keyed clip would have to be composited onto something first. That belongs in an editor." }, { "heading": "Resolution: your source size, or nothing", "text": "The tool never resamples. There is no \"export at 1080p\" control, because there is nothing to control: the keyed frame is written at whatever the source actually is. For the PNG still and the PNG sequence that is exact. A 1080×1920 phone clip exports 1080×1920 frames; a 4K clip exports 4K frames, slower to key and much heavier to store. If you want a smaller file, downscale the source first — the export will not do it for you. The WebM path has one small exception: both dimensions are rounded down to even numbers before the recording canvas is created, so a 1081-pixel-wide source exports at 1080 and you lose one column and one row. The encoder works in 4:2:0 chroma subsampling, which wants even values. You will only notice this on cropped phone footage." }, { "heading": "Frame rate: two different rules", "text": "The PNG sequence and the WebM do not share a frame rate setting, and only one of them is under your control. PNG sequence. This one uses the frame rate slider. It defaults to 12 frames per second, and the number of frames written is the clip's duration multiplied by that rate, rounded down. There is a hard ceiling of 1200 frames, because every frame is held in memory before the archive is written. A 60-second clip at 12 fps is 720 frames and fits; the same clip at 30 fps would be 1800 and is refused, with the rate that would fit reported instead. Transparent WebM. You cannot set this one. The canvas is captured at up to 30 frames per second and the keying loop runs on animation frames while the clip plays in real time. Three consequences: Output is capped at 30 fps whatever your source is — a 60 fps clip comes out at roughly 30. It is a real-time capture, not a frame-accurate transcode. If the keying loop cannot keep up, frames are dropped and the result plays slightly faster than reality. Because it records while playing, a 30-second clip takes about 30 seconds to export. There is no offline render mode." }, { "heading": "Alpha: what survives each path", "text": "PNG still and PNG sequence: 8-bit per-pixel alpha, lossless. A pixel the key removed is fully transparent in the file, and there is no second encoding pass to soften the edge. WebM: the recording canvas has an alpha channel and the recorder keeps it, so the transparency is in the file. Whether you see it depends on what you open it with — that is a player problem, not an export problem. The transparent WebM page covers which players and editors keep it. Expect one difference on the WebM path: it is a re-encode. The keyed frames go through a lossy VP9 pass at 8 Mbps, so a fine alpha edge — hair, a soft rim — can come out slightly softer than it looked in the preview. The PNG sequence has no such pass. If the edge is the point, export frames." }, { "heading": "Audio", "text": "The WebM export has no audio track, and this surprises people. The video element is muted for programmatic playback and the recorded stream is taken from the canvas alone, so nothing carries the sound across. If you need the audio, keep the original and mux the track back in an editor, or record it separately and sync afterwards." }, { "heading": "File size and how long it takes", "text": "The PNG sequence is the heavy option. Each frame is a full-resolution PNG and the archive is written in stored mode — the frames are not compressed further on top of the PNG encoding they already went through — so the download is close to the sum of the individual frames. The 1200-frame ceiling exists for this reason: the frames are assembled in memory before the archive is handed to you. The WebM is the light option. Video is encoded at 8 Mbps, so half a minute lands somewhere around 30 MB, as a single file rather than a folder of stills. The trade is the real-time export and the lossy pass above." }, { "heading": "Which one to pick", "text": "Compositing over a new background: PNG sequence if your editor imports image sequences cleanly, WebM if it prefers a single clip. Test one short clip before committing to a long one. A single still: PNG. One file that plays with transparency: WebM — then check it in the player you actually intend to use. Frame-accurate timing, or you care about the edge: PNG sequence. No re-encode, no dropped frames, no rounding. The PNG sequence vs WebM page goes through the trade in more detail." }, { "heading": "When this does not work", "text": "You want MP4 with alpha. Not offered, and not possible in the common MP4 profiles. Composite in an editor. You want to choose an output resolution. There is no picker; the export matches the source. You want to set the WebM frame rate. Fixed at up to 30 fps, driven by real-time capture. You need the audio in the WebM. It is not there. Mux it separately. Your machine is too slow to keep up. The WebM drops frames and runs fast; there is no slower-but-complete option. Use the PNG sequence, which is not time-bound. Your browser cannot record a canvas. The WebM path refuses and says so, pointing you at the PNG sequence rather than producing a broken file. The clip is long and you wanted frames. Over 1200 frames the sequence is refused — lower the rate or cut the clip up. The source has an odd pixel dimension. The WebM loses one row and one column; the PNG exports do not." }, { "heading": "Then export it locally", "text": "None of this involves an upload, an account or a watermark — the key runs in the tab and the file is written straight to your disk. The no-upload page covers what keeping it local avoids, and the video page covers the keying side rather than the export side. The measurement fixtures behind the numbers on this site are in the reference repository, with the write-up archived at doi.org/10.5281/zenodo.22916767 — synthetic fixtures, not real footage, so read them as the shape of the problem. Open the green screen remover" }, { "heading": "Related video pages", "text": "Green Screen Remover (home) Green Screen for YouTube Shorts OBS Chroma Key Settings Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-remover-no-watermark", "title": "Green Screen Remover Without a Watermark — Clean Exports, No Logo", "description": "Why so many green screen tools stamp a logo on the free export, what running the key in your own browser changes, and what you give up instead of a watermark.", "markdown": "/ai/pages/no-watermark.md", "sections": [ { "heading": "Green screen remover without a watermark", "text": "A watermark on a keyed export is not a technical necessity — it is a pricing decision. This page is what that decision usually looks like, why a browser-side tool can skip it, and what you trade instead." }, { "heading": "What \"no watermark\" means here", "text": "Every export this tool writes is the keyed canvas and nothing else. There is no logo burned into a corner, no semi-transparent mark across the frame, no forced downscale on the free path and no \"made with\" badge added to the file. The download button hands the browser a blob built from the same canvas you were just looking at, and the browser writes it to your disk. That is not a promise on a pricing page; it is the shape of the code. The export path encodes the canvas and triggers a download — there is no drawing step in between where a mark could be added. The file is also unencumbered in the ways that matter for reuse: a transparent PNG, a ZIP of PNG frames, or a transparent WebM, all at your source resolution. Nothing in the file records that it came from here." }, { "heading": "Why so many tools put one there", "text": "Most online green screen removers render on a server. That has a real cost per file: a machine decodes your clip, keys it, encodes it again and stores the result long enough for you to fetch it. Anybody offering that for free has to recover the cost somewhere, and the two usual answers are a watermark on the free tier or a hard cap that pushes you into a plan. A watermark does two jobs at once. It withholds the usable version of the file, and it advertises the tool wherever that file ends up. For a service that pays for render time, it is a rational choice — not a trick, just the trade they picked. There are two other reasons you will see marks that have nothing to do with price: stock-footage previews that are watermarked because you have not licensed the clip yet, and tools whose licence requires attribution. Neither is something a keying tool can or should strip." }, { "heading": "Why this one can skip it", "text": "The keying happens in your tab, on your CPU. There is no render queue, no storage and no per-file bill, so there is nothing a watermark would be protecting the margin on. The no-upload page covers what keeping it local also avoids: no upload, no account, no retention window, no terms that let a service reuse what you dropped in. The site is sustained by one disclosed affiliate link, noted in the footer of every page. That is the whole model. No subscription unlocks a cleaner export, because there is no dirtier one to unlock." }, { "heading": "No watermark does not mean no limits", "text": "Skipping the watermark does not make the tool unlimited. The honest limits are about your machine, not your plan: A PNG sequence is assembled in memory before it is zipped, so it stops at 1200 frames. The page refuses and tells you the frame rate that would fit rather than writing a short file. The transparent WebM is a real-time capture at up to 30 fps, encoded at about 8 Mbps. It is not a frame-accurate transcode and it carries no audio. Everything runs at your source resolution — which means a long 4K clip is slow here, because your machine is doing all of it. The difference is that these are physics and memory, not a paywall. They do not change if you come back tomorrow or sign in. The export settings page lists them in full." }, { "heading": "What you are actually paying with", "text": "You pay in features rather than money. There is no manual brush, no rotoscope, no tracking mask and no batch queue. Colour is the only lever this key has, and the page says so instead of showing you buttons that do nothing. If your shot needs hand work, no free-tier watermark removal will save it — you need an editor. You also pay with your own hardware. A three-minute 4K clip is slower here than on someone else's render farm, and if the machine cannot keep up, the WebM capture drops frames. The PNG sequence vs WebM page covers the export that is not time-bound." }, { "heading": "Does \"free\" mean the key is worse?", "text": "No, and this is measurable. On the flat-green fixture in the reference set, the shipped default produces an intersection-over-union of 1.000000 — 55,459 subject pixels kept, zero background pixels left, zero subject pixels lost. Nothing is being held back for a paid tier, because the whole keyer is already in the page. What does cost you quality is the shot, not the tier. On the same fixture with a shadow falling across the screen, the default drops to 0.617718, and raising tolerance to 40 brings it back to 1.000000. Those are synthetic fixtures, not real footage — read them as the shape of the problem, not as a score your clip will get." }, { "heading": "How to check any tool before you trust it", "text": "Export the shortest possible clip first. Most watermarks are applied on the free path regardless of length, so a two-second test tells you the truth. Read the pricing page, not the headline. \"Free\" usually means free to process; the download is the part that is metered. Try it offline. Load the page, disconnect, and key a file. If it still works, no server was involved — and a server is where watermarking happens. Check the file, not the preview. Some tools show a clean preview and stamp the download." }, { "heading": "When this does not work", "text": "The watermark is already burned into footage you were given. A chroma key subtracts a colour; it cannot reconstruct what a logo covers. You need inpainting or the unbranded source. You need someone else's render farm. A long 4K clip on an old laptop is slow here. Local processing is the feature and the ceiling. Your clip is longer than the sequence ceiling. Past 1200 frames the PNG export is refused; lower the frame rate or cut the clip. You want MP4. The MP4 profiles that play everywhere carry no alpha, so there is nothing to composite yet. You want the audio in the exported video. The WebM has no audio track. Keep the original and mux it back. You want batch processing. One file at a time, in the tab. The mark is on a stock preview you have not licensed. That is a licensing thing, not a pixel problem. You need hand-worked edges. There is no brush here." }, { "heading": "Key it locally, keep the file clean", "text": "Open the tool, drop in a clip or a photo, and download a transparent PNG, a PNG sequence or a transparent WebM — no upload, no account, no mark on the result. The video page walks through the keying controls, and the transparent WebM page explains which players keep the alpha you just exported. The fixtures and measurements behind the numbers on this site live in the reference repository, with the write-up archived at doi.org/10.5281/zenodo.22916767. Open the green screen remover" }, { "heading": "Related image pages", "text": "Green Screen Remover (home) Remove Green Background from Photo Remove White Background Without a Green Screen Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/chroma-key-artifacts", "title": "Chroma Key Artifacts — Dark Rims, Green Fringes, Jagged Edges and Flicker", "description": "Every common chroma key artifact and what actually causes it: the dark rim, the green fringe, holes in the subject, stair-stepped edges and edges that boil on video — with the synthetic-fixture measurements behind each one.", "markdown": "/ai/pages/chroma-key-artifacts.md", "sections": [ { "heading": "Chroma key artifacts — what each edge problem actually is", "text": "A dark rim, a green rim, holes in the subject, stair-stepped edges, an outline that crawls when the video plays. They look alike in a thumbnail and have four different causes. Here is how to tell them apart." }, { "heading": "Where all of them come from", "text": "The keyer converts every pixel to YCbCr, measures how far it sits from the key colour in the Cb/Cr plane, and turns that distance into an alpha. At the default edge of 0 the two threshold edges coincide, so alpha is only ever 0 or 1 — a binary matte. Nothing in that loop looks at the neighbouring pixel or the previous frame. So every artifact on an edge is one of three failures: backdrop pixels that stayed opaque, subject pixels that got erased, or subject pixels that kept the wrong colour. Which one it is decides which slider helps." }, { "heading": "The dark rim: backdrop that survived", "text": "The outline hugs the subject, is grey rather than green, and thickens on the side that fell into shadow. Sometimes it is a whole patch of screen that refused to go. Those pixels are the screen, only darker. Distance in Cb/Cr survives a brightness drop far better than distance in RGB, but not completely: a backdrop at 55% brightness sits 85.0941 units from the key in RGB and 37.8083 away in Cb/Cr, while default tolerance of 25 opens a band of only 29.5 units. So it stays solid. On the shadowed fixture at tolerance 25, 21,201 of 55,459 backdrop pixels survive, a recall of 0.617718. At 35 it is down to 12,720 (0.770641), and the fixture reaches 1.000000 only at tolerance 40. Raising tolerance is the fix; the better one is lighting the screen so it never needs the extra room." }, { "heading": "The green rim: a colour problem, not a matte problem", "text": "The subject is intact and the cut is clean, but hair, skin and light clothing carry a green cast, strongest at the outline. This is not a hole in the matte. Light bounced off the screen onto the subject, those pixels really are partly green, and the keyer kept them correctly. On the spill fixture at tolerance 25 the matte scores 1.000000, precision 1.000000 — nothing erased, nothing left behind. Chasing the tint with tolerance instead erases the rim with the subject: at tolerance 80, 7,206 of 21,341 subject pixels are gone, precision 0.885008, and the rest are still tinted, because the tint is inside the subject rather than on its boundary. Two controls address it: the spill slider pulls the dominant channel down toward the brighter of the other two, and a negative edge trims the outer band. At edge −100 the trim costs 6,902 subject pixels, IoU 0.889322. There is no sweep for spill strength — every edge-sweep row was measured with it off — so this page can say what the slider does, not by how much." }, { "heading": "Holes in the subject", "text": "See-through patches in a person, thin fabric gone translucent, hair thinning out. The opposite failure: subject pixels fell inside the band. It is not gradual. On the flat-green fixture the score holds 1.000000 to tolerance 85, then at 90 the key erases all 21,341 subject pixels at once and IoU drops to 0.722122 — one step between a clean matte and an empty frame. A trim does the same by another route: edge −100 eats 6,902 subject pixels." }, { "heading": "Stair-stepped edges", "text": "On a still frame the outline is a staircase rather than a curve. That is the binary matte: with edge at 0, alpha has no intermediate value, so there is nothing to anti-alias with. A positive edge opens a ramp instead — 46 units of chroma distance wide at edge +100 for a colour key, 60 for a neutral one, with pixels inside it partly transparent. Soft edges composite more forgivingly, but a ramp laid over a spill rim leaves a translucent green band, so use it after the tint is handled, not instead." }, { "heading": "Edges that boil on video", "text": "The still frame looks fine and the outline crawls once it plays. Every frame is keyed with the same parameters and no knowledge of the previous one, so noise and grain push pixels back and forth across a single threshold. Thin detail suffers most, because hair, smoke and motion blur already sit near that threshold. No measurement backs this one: the fixtures are single frames, so there is nothing to score across time. This section is the mechanism, not a number. What helps is widening the ramp, so a flipping pixel changes opacity instead of vanishing, and cleaner footage." }, { "heading": "Telling them apart in one pass", "text": "Rim dark, worst in shadow — tolerance too low. The tolerance page shows the units. Rim green or blue, subject intact — spill. Use the spill slider, not tolerance: see the spill page. Holes inside the subject — tolerance too high or too much trim. Staircase edge on a still frame — edge is 0. Add a small positive edge. Edge crawls only in playback — noise crossing a hard threshold. Soften it, clean the source. Only one side is bad — lighting, not parameters: the lighting guide." }, { "heading": "Where the numbers come from", "text": "Every figure here is measured against ground truth. The reference set is six deterministic synthetic frames at 320×240 with hand-built masks, regenerated by a script so the tables can be diffed when the algorithm changes. A pixel counts as backdrop when the keyer gave it alpha 0; precision is the share of kept pixels that really are subject, recall is the share of backdrop actually removed, and IoU combines both. They are synthetic fixtures, not real footage — read them as the shape of the problem, not a score your clip will get." }, { "heading": "When this does not work", "text": "The artifact is baked into footage you were given. A key subtracts a colour; it cannot reconstruct what a bad matte already destroyed. Motion blur. Those pixels are a genuine mix of subject and screen, and one global threshold cannot split them. Hair, fur, smoke and glass. They are semi-transparent by nature, and there is no matte refinement or per-pixel maturity here — just a distance and a threshold. Boiling edges. No temporal filter runs in this tool. Softening hides it; only an editor with frame-to-frame analysis removes it. A wrinkled or unevenly lit screen. The shading varies across the frame, which needs a threshold that varies too. One slider is global. The subject contains the key colour. A green shirt on a green screen is indistinguishable from the backdrop by construction. Heavily compressed source. Macroblocking sits exactly where the edge is, and no threshold lands cleanly on it. You want per-frame hand work. There is no brush and no rotoscope here. Colour is the only lever." }, { "heading": "Find out which one you have", "text": "Open the tool, drop in the clip or the photo, and change one control at a time — tolerance, then edge, then spill — watching the same patch of outline. The video walkthrough covers the controls in order. The fixtures and benchmark behind every number on this site are in the reference repository, with the write-up archived at doi.org/10.5281/zenodo.22916767. Open the green screen remover" }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) Chroma Key Benchmark Chroma Key vs Luma Key Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-file-size-limits", "title": "Green Screen File Size Limits — How Big a Clip Can You Key?", "description": "No upload means no file size cap: why this green screen remover takes a file of any size, what limits you instead — tab memory, the browser decoder, the 1200-frame ceiling and real-time recording — and how to get a long clip through.", "markdown": "/ai/pages/file-size-limits.md", "sections": [ { "heading": "Green screen file size limits — how big a file can you actually key?", "text": "Most online removers stop you at 100 MB, or 500 MB, or \"up to three minutes\". This one carries no number, because nothing is uploaded. Here is what that removes and what still stops you." }, { "heading": "The short answer", "text": "This page imposes no file size limit and no length limit; drop in a multi-gigabyte clip and it starts. What limits you sits on your side: the memory available to the tab, the decoder in your browser, and two ceilings in the export path — a 1200-frame cap on the PNG sequence, real-time recording on the WebM. An upload cap is a rule somebody wrote into a pricing page. These are properties of your machine, and you can work around most of them." }, { "heading": "Why \"no upload\" removes the cap", "text": "An upload-based remover has to receive, store, render and return your file. Every step costs the operator money per gigabyte, so the free tier gets a ceiling and the paid tier a larger one. The cap is about their bill, not your clip. Here the file never moves. The browser gets a local object URL, decodes frames with the Canvas API, keys each one and writes the result back out. No server in the loop means no cost to protect and nothing to meter. The no-upload page covers what that means for your footage; the privacy page says what is and is not collected." }, { "heading": "The four limits that replace it", "text": "Memory. A PNG sequence is assembled frame by frame in the tab's memory before the archive is written. Long clips at high resolution feel this first. The decoder. If your browser cannot decode the codec, the file is refused before keying starts, and the message names the readable formats. The 1200-frame ceiling. Only the sequence has it. It counts frames, not bytes: 1200 frames at any resolution. Real-time recording. Only the WebM has it. The export takes as long as the clip takes to play." }, { "heading": "The frame ceiling, in seconds", "text": "The sequence caps at 1200 frames and you pick the rate, so the ceiling in seconds moves with the slider: 30 fps — 40 seconds 24 fps — 50 seconds 12 fps, the default — 100 seconds 1 fps — 1200 seconds Go over and nothing is exported: the page refuses and names the rate that would fit. The projected frame count sits next to the slider before you start. Lower the rate, or trim." }, { "heading": "The preview is capped. The export is not.", "text": "What you watch while dragging sliders is scaled down: the long edge of the preview is held at 1280 pixels whatever the source is. That is what keeps a slider responsive on a 4K clip. The export ignores it entirely — the transparent PNG and every frame of the sequence are written at the source's own pixel dimensions. So a heavy file is not slower to tune than a light one; full resolution is paid for only at export. One consequence: the WebM rounds each dimension down to an even number, so an odd-sized source loses one row and one column there. The PNG exports do not." }, { "heading": "What \"no limit\" actually costs you", "text": "Nothing here is free; the costs moved off a price list and onto your hardware: Full resolution, every frame. Nothing is downscaled on the way out; a long 4K sequence is a lot of pixels to key and a lot of PNGs to hold. The archive is stored, not compressed. The ZIP is stored, not deflated on top of the PNG encoding the frames already went through, so the download lands near the sum of the frames. The WebM is recorded, not rendered. It captures at up to 30 fps while the clip plays, at 8 Mbps. A ten-minute clip is a ten-minute wait, and a frame dropped mid-recording is gone. The whole job runs on your CPU. No render machine sits behind the page: a slow computer is a slow export." }, { "heading": "Getting a long clip through", "text": "Cut it first. Trim to the part that needs the key; that solves the frame ceiling and the memory pressure at once. Lower the sequence rate. 12 fps is the default and 8 or 6 is usually invisible in a matte that will be composited anyway. That buys minutes. Take the WebM for long clips. It has no frame cap. It costs real time and a lossy pass, but length is not the problem there. Downscale the source before it arrives. If the delivery is 1080p, keying a 4K original only makes every step heavier for pixels nobody will see. Close other tabs. The memory ceiling belongs to the tab, and a browser with forty tabs open has less to give." }, { "heading": "Where the numbers on this site come from", "text": "Every keying measurement published here comes from a reference set of deterministic synthetic frames at 320×240 with hand-built masks, regenerated by a script so tables can be diffed when the algorithm changes. The CSVs live in the reference repository and the write-up is archived at doi.org/10.5281/zenodo.22916767. Those fixtures measure whether a key separates subject from backdrop. They say nothing about size, throughput or codec quality, and they are far smaller than real footage — a 1080p frame is 2,073,600 pixels against 76,800, twenty-seven times as many. So this page quotes no megabytes per second: we have not measured it, and a guess dressed as a number is worse than no number. The benchmark page says what those tables do and do not cover." }, { "heading": "When this does not work", "text": "Your clip has no readable duration. A live or fragmented stream is refused before it starts. Re-encode it as MP4 or WebM first. The file has no video track. An audio-only file, or a container your browser decodes but finds no picture in, is rejected with that stated. Your browser cannot decode the codec. ProRes, HEVC in some browsers and odd camera formats will fail; there is no server-side fallback. You want the sequence and the clip is long. 1200 frames is fixed. Lower the rate or trim; the ceiling is not adjustable. You are short on memory. A long high-resolution sequence can exhaust the tab, and nothing here streams to disk. You need it faster than the clip plays. The WebM cannot be rushed, it records in real time. The sequence is not time-bound but is capped by frames. You expected the export to be downscaled. There is no output resolution picker: you get your source size and the file size that comes with it. You want a smaller file out than in. A stored-mode ZIP of full-resolution PNGs is the heavy path by design, the WebM is light and lossy, and there is no third option." }, { "heading": "Try it on the file you actually have", "text": "Open the tool, drop the clip in and read the frame count next to the rate slider before you commit — it is the one number that decides whether the sequence runs. The PNG sequence vs WebM page covers which export to pick; the export settings page goes through what each keeps. Open the green screen remover" }, { "heading": "Related video pages", "text": "Green Screen Remover (home) Remove Green Screen from Video Transparent Background Video Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/transparent-png-alpha-explained", "title": "Transparent PNG Alpha — Why Keyed Edges Go White or Dark", "description": "What the alpha channel in a transparent PNG really stores: straight alpha, 8-bit coverage, why the same keyed edge looks white on a dark page and dark on a white one, and what this browser keyer writes into the file.", "markdown": "/ai/pages/transparent-png-alpha-explained.md", "sections": [ { "heading": "Transparent PNG alpha — what the fourth channel really stores", "text": "A transparent PNG is not a picture with holes in it. Every pixel still carries three colour bytes; a fourth says how much of what sits behind it shows through. Most edge trouble blamed on “bad transparency” is that byte meeting a background it never matched." }, { "heading": "The fourth byte", "text": "A transparent PNG uses colour type 6: an R, G, B triple plus one alpha sample per pixel, at 8 or 16 bits. Per the PNG specification, 0 is full transparency and 2bitdepth−1 full opacity, intermediate values compositing against the background. The spec is explicit about colour too: “PNG does not use premultiplied alpha.” That matters: a pixel at 30% opacity stores its full colour, not 30% of it. Straight alpha, also called unassociated. One decision in a 1996 format is behind most of this page." }, { "heading": "Where the alpha in this file comes from", "text": "The keyer converts each pixel to YCbCr, measures its distance from the key colour in the Cb/Cr plane (luminance for a neutral backdrop) and turns that distance into alpha: 0 at or below the lower threshold, 1 at or above the upper one, a straight line between. The lower threshold is (tolerance / 100) × 118 for a colour key, × 150 for a neutral one: default tolerance 25 opens a band 29.5 units wide on green, 37.5 on a white wall. A positive edge widens the ramp by (edge / 100) × 46 units for colour, 60 for neutral; a negative edge pushes the threshold inward instead. At the default edge of 0 the two thresholds coincide and the matte is hard: alpha is only ever 0 or 255, and a default export has no partially transparent pixels at all. Fractional alpha exists only where you asked for a soft edge — at edge +100 with tolerance 25 on green the ramp runs 29.5 to 75.5 units, and that span is the whole soft edge. Alpha is a threshold result, not a fade. In the shadow sweep, a backdrop at 65% brightness sits 29.3412 units away and lands on alpha 0 at tolerance 25; the same backdrop at 60% sits 33.5337 away and lands on 255, fully opaque and not keyed at all. Raise tolerance to 50 and both go to 0." }, { "heading": "Why one file looks white here and dark there", "text": "Compositing is one line of arithmetic: out = C × α + B × (1 − α), where C is the colour stored in the pixel and B is whatever is behind it. At α = 0.1, ninety per cent of what you see is the background. Nothing in the PNG records what that background should be — the file is a promise, not a finished picture. So one export can look fine on a checkerboard, pale on white and grimy on near-black, with no change to the file. The preview on the tool has Checker, Dark and Light buttons for exactly this: flip the frame between backings and watch the outline. A fringe on both is colour in the pixel; a fringe on one is the backing showing through coverage you did not mean to create. That is arithmetic, not a measurement: everything here is measured on synthetic 320×240 fixtures against a binary ground truth." }, { "heading": "Why a clean matte can still show a white rim", "text": "Four things produce the white or dark rim people report; only one is in the file. Straight alpha read as premultiplied. A compositor expecting premultiplied data computes C + (1 − α) × B instead of C × α + (1 − α) × B — brighter than intended, a pale halo. The reverse mistake darkens it. The PNG is unambiguous; the reader is not. Flattening on white. Convert to JPEG, “save for web”, most thumbnailers and plenty of upload pipelines fill transparency with white first. A pixel at α = 0.1 then picks up ninety per cent white — a fringe that was never in the PNG. A soft pixel keeps its old colour. This keyer writes alpha, not a replacement colour: the source RGB goes back unchanged unless you move spill, which only pulls the dominant channel toward the brighter of the other two. A feathered pixel at low alpha stays green-tinted — dull over dark, pale over light. A fully transparent pixel still carries three colour bytes. Software that ignores alpha, or a JPEG screenshot, displays them: usually green, sometimes black. Neither is what the file means." }, { "heading": "What actually gets written", "text": "Export makes a canvas at the source’s full pixel size, clears it, draws the frame, keys it in place and encodes once as PNG. Nothing is composited onto white and nothing is stamped on. Each pixel keeps the source RGB — spill-adjusted if you asked — with alpha as round(α × 255), one of 256 levels. The same pixels go into a PNG sequence’s frames; the transparent WebM path is a different route with different losses, covered in PNG sequence vs transparent WebM and transparent WebM explained." }, { "heading": "Checking a file before you hand it over", "text": "Put the result on Dark, then Light, and watch the outline rather than the middle of the subject. Visible on both means colour: move spill, or trim with a negative edge. Visible on one means coverage: lower the feather. If a destination shows it identically on black and white, its alpha was flattened on the way in. Keep the PNG as master. Once a copy has been through a JPEG or a web export, the fourth channel is gone." }, { "heading": "When this does not work", "text": "The subject is genuinely semi-transparent. Hair, smoke, glass and motion blur give pixels that are partly subject, partly backdrop, and alpha has one number per pixel: it cannot say which is which. See which edge problem you have before reaching for a slider. You need a soft edge that also changes colour. Feathering gives coverage without a new colour, so soft edges keep their green cast. Spill suppression is the only colour lever, and no sweep measures how far it goes. The destination ignores or flattens alpha. Nothing in your control; the file is correct and the pipeline is not. The source was already heavily compressed. The edge pixels’ colour is unreliable before the key runs, and straight alpha faithfully preserves unreliable colour. Backdrop brightness crosses the threshold mid-frame. A shadowed screen can sit at alpha 255 on one side and 0 on the other — the byte flips, it does not fade. Light the screen rather than chasing it with tolerance. You want 16-bit alpha. The canvas here is 8 bits per channel, so the file is 8-bit; the format allows more, this pipeline does not produce it. You need per-pixel hand work. There is no brush and no matte painting. Alpha comes from colour distance and nothing else. You are judging the result from a thumbnail. Thumbnailers composite, usually on white, at a scale where a two-pixel rim becomes a border." }, { "heading": "Where the numbers come from", "text": "Thresholds and the feather and trim scales come from the keyer in the reference repository; the alpha column above is from its shadow sweep on a synthetic 320×240 fixture, not real footage. The write-up is archived at doi.org/10.5281/zenodo.22916767; the format rules are the PNG specification. For a photo, the photo walkthrough covers the same export. Open the green screen remover" }, { "heading": "Related image pages", "text": "Green Screen Remover (home) Keying Hair and Transparent Edges Blue Screen Remover Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-for-zoom-and-meetings", "title": "Green Screen for Zoom and Meetings — What a Real Key Adds", "description": "A meeting app replaces your background live; a keyer gives you a transparent file you can reuse. What each one actually does, why a meeting recording cannot be keyed back, and how a desk setup fails — with the synthetic-fixture measurements behind it.", "markdown": "/ai/pages/zoom-and-meetings.md", "sections": [ { "heading": "Green screen for Zoom and meetings — what a real key adds", "text": "Every meeting app now ships its own background replacement, and it runs while you are on the call. Producing a transparent asset you can reuse is a different job. This page is about the seam between the two. It does not benchmark any meeting app: we have not measured one, and nothing below is a claim about how any of them behave." }, { "heading": "Two jobs that look alike", "text": "A meeting app's background feature reads the frame from your camera, decides which pixels are you, and sends the composited result into the call. Nothing is written to disk. The only artefact is the call itself. A keyer like this one takes a file and writes a new one: a transparent PNG, a ZIP of PNG frames, or a transparent WebM. Nothing is live, and what you get is a reusable asset. So: if all you need is a tidy background on this afternoon's call, the built-in feature is the right tool. If you need the subject with the backdrop actually gone, to put over something else more than once, you need the second job — and footage recorded separately." }, { "heading": "The recording you get back from a call is already baked", "text": "This is the one that catches people. A recording from a meeting app is a video of the composited call: the background was flattened into every frame before the file was written, and MP4 has no alpha channel to carry a matte anyway. No keyer can separate a subject from a backdrop merged on someone else's machine. To get a keyable version of yourself, record locally from the camera, before the meeting app touches it. That file is what you drop into the tool. Which formats keep transparency at all is on the transparent background video page." }, { "heading": "Shooting at a desk is harder than shooting in a studio", "text": "Desk setups fail predictably: the screen is lit by the same lamp as the subject, so it is never even; it is narrower than the frame, so the border of the shot is room; and a laptop webcam sits close and low, so the subject casts onto it. All three map onto measurements we do have. These come from synthetic 320×240 fixtures scored against a binary ground truth — read them as what a clean setup can do, not a prediction for your webcam." }, { "heading": "A shadow on the screen is the common failure", "text": "In the shadow sweep, a backdrop at 65% brightness sits 29.3412 units from the key colour in the Cb/Cr plane and keys clean at the default tolerance of 25. The same screen at 60% sits 33.5337 away and comes out at alpha 255 — fully opaque, not keyed at all. That step is a cliff, not a gradient. On the shaded fixture the default tolerance leaves 21,201 of 55,459 backdrop pixels behind: IoU 0.617718. At tolerance 35 it leaves 12,720; at 40 every backdrop pixel goes, IoU 1.000000. At 90 it takes 21,341 subject pixels with it, IoU 0.722122. The fix is real and it has a ceiling — which is what the tolerance page is about." }, { "heading": "A screen that does not fill the frame breaks the auto pick", "text": "The key colour is guessed, not asked for: it is the median of the outer 5% band around the frame. That median shrugs off a subject touching one edge, but assumes most of the border is backdrop. Hang a screen that stops short of the frame and the median is your wall. Shade the screen and the pick drifts: on the shaded fixture the detected colour is #00a23b where the real screen is #00b140, a maximum channel delta of 15. You can always set the colour by hand. If the backdrop has almost no colour in it — a white wall, which is what most home offices actually have — the keyer falls back to a luminance key. The band widens to 37.5 units at the default tolerance instead of 29.5, and so does what gets caught. The honest margin on a white wall is on the white background page." }, { "heading": "What you get out of the tool", "text": "Three exports, all built in your browser with no upload: a transparent PNG at the source resolution, a ZIP of PNG frames, and a transparent WebM. The sequence is capped at 1,200 frames; the WebM records in real time — a 30-second clip takes 30 seconds. Picking between them is on the PNG sequence vs WebM page; resolution and frame rate are on export settings. No watermark, because there is no server: the file never leaves the tab. That matters more than usual here, since footage shot for work is often internal. More on the no-watermark and no-upload pages." }, { "heading": "When this does not work", "text": "A recording that came out of a meeting app. The background was baked in. There is nothing to key. A live call you want to look better right now. This tool works on files. It is not a camera effect and it does not create a virtual camera device. A screen that does not reach the frame edge. The auto pick reads the border: either fill the frame or set the colour by hand. A shadow across the screen, at default settings. Below about 60% brightness the default tolerance leaves solid pixels. Raise tolerance — and stop before it eats the subject. Hair, smoke, glass — anything genuinely semi-transparent. At the default edge of 0 the two thresholds coincide, so alpha is only ever 0 or 255: no partially transparent pixels describe a soft edge unless you dial one in. We have no measurement of how that holds up on real hair. Noise that changes frame to frame. Every frame is keyed on its own, with no temporal or spatial filtering, so noise can make an edge boil between frames. Not measured: the fixtures are single frames. A long recording. The sequence is capped at 1,200 frames, the WebM records in real time, and everything happens in one tab's memory. Nothing checks the file size on the way in, so nothing warns you before the browser runs out of room. Anything compressed. Every number here comes from uncompressed synthetic fixtures. A webcam file or a re-encoded MP4 discards colour detail before the keyer sees it. Not measured, not guessed." }, { "heading": "Where the numbers come from", "text": "The IoU figures come from the synthetic-fixture benchmark behind this site: 320×240 frames, a binary ground truth, Cb/Cr distance keying against two naive baselines. On flat green the shipped default scores 1.000000; on the shaded fixture 0.617718 at tolerance 25, 1.000000 at 40. Data and code are in the reference repository; the write-up is on Zenodo under DOI 10.5281/zenodo.22916767. Fixture, not footage. For the shooting side — what to light, how big a screen to hang, why wrinkles ruin the matte — start with the lighting guide and the DIY screen page." }, { "heading": "Related video pages", "text": "Green Screen Remover (home) Green Screen File Size Limits Remove Green Screen from Video Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/keying-hair-and-transparent-edges", "title": "Keying Hair and Transparent Edges — What a Chroma Key Can and Cannot Do", "description": "Flyaway hair, smoke, glass and motion blur: why semi-transparent edges break a colour key, what the edge slider actually changes, and the cases where no slider gets you a clean matte.", "markdown": "/ai/pages/hair-and-transparent-edges.md", "sections": [ { "heading": "Keying hair and transparent edges: what a chroma key can and cannot do", "text": "Flyaway hair, smoke, a glass, a hand moving fast — things that are half backdrop and half subject. Here is what happens to them, what the edge slider changes, and where the honest answer is \"not with this tool\"." }, { "heading": "Why anything semi-transparent is hard", "text": "A chroma key makes one decision per pixel: measure the distance from the key colour in the Cb/Cr plane, compare it to a threshold, write an alpha. A strand of hair over green is not \"70% hair\" — it is one colour at one distance. Past the threshold it is opaque, inside it clear. Nothing in the image carries coverage. So semi-transparency has to be recovered rather than measured. A pixel that is half dark hair over bright green and a pixel that is entirely mid-green sit at two different distances, and the maths cannot tell which is which." }, { "heading": "What this keyer writes by default: a binary matte", "text": "Start here, because it decides what the rest of the page means. The edge slider starts at 0. Feather is then zero, the two thresholds are equal, and the loop sets hard = hi - lo < 1e-6. Every alpha written is 0 or 255, and the value reaching the file is alpha >= 1 ? a : Math.round(alpha * a). At default settings there are no semi-transparent pixels at all. Hair comes out as a hard silhouette: each strand survives whole or vanishes whole. That is not an export bug and not the PNG dropping data — the keyer never computed an intermediate value." }, { "heading": "What the edge slider buys you", "text": "Two things, depending on direction. Positive edge is feathering. At edge 100 with a colour key the ramp spans 46 units of Cb/Cr distance (60 for a neutral backdrop such as a white wall). With the default tolerance of 25 the lower threshold is 29.5, so the ramp runs 29.5 to 75.5: closer is fully clear, further is fully opaque, and only pixels inside that band get partial alpha. That band is the only place a soft edge can exist. Strands blended with the screen can land in the band and come out soft. But the band is narrow, defined by colour distance rather than coverage, and it moves the whole edge — it does not detect hair. Push it far enough and you feather areas that are not edges at all. Negative edge is trimming, the opposite trade. It raises the lower threshold so more pixels go fully clear. At edge −100 it lifts it by 62 units, and on the spill-green fixture that eats 6,902 of the subject's 21,341 pixels: IoU drops from 1.000000 to 0.889322. That is the price of killing a fringe — real subject goes with it." }, { "heading": "Smoke, glass and motion blur", "text": "Same mechanism, worse odds. Smoke and glass refract and accumulate rather than blend. A pixel of smoke is green tinted toward grey; a pixel of glass is green shifted, often with the backdrop distorted through it. Neither sits reliably in a narrow band around the key, so some lands in the ramp and some does not: patches that key out next to patches that stay solid grey. No slider fixes that, because the distance is not consistent across the area. Motion blur is the same problem in time. Every frame is keyed independently, with no temporal or spatial filtering, so a soft edge along a moving hand is re-decided each frame." }, { "heading": "What spill suppression does here, and what it does not", "text": "Spill suppression changes colour, never alpha. It runs on pixels above zero alpha and pulls the key channel toward the brighter of the other two — for a green key, green toward max(red, blue) by the slider fraction, 50 by default. On hair carrying green bounce that genuinely helps. It cannot make hair transparent, because it never touches the matte. A strand the key called opaque stays opaque, just less green. It also skips pixels already cleared, so it does nothing on the background side of an edge." }, { "heading": "When this does not work", "text": "Flyaway hair at default settings. The matte is binary, so strands are present or gone. You need the edge slider above zero before any partial alpha exists. Hair against a shadowed screen. The backdrop itself is failing to key, and no edge setting recovers it. On shadow-green, tolerance 25 leaves 21,201 of 55,459 background pixels behind (IoU 0.617718); it takes tolerance 40 to reach 1.000000. Backlit or strongly rim-lit hair. A bright halo between the strands and the screen shifts the colour of every edge pixel, differently on each strand. Smoke, fog, haze or fire. Density varies across the frame, so no single threshold band follows it. Glass and clear plastic. A window or a glass refracts the backdrop instead of blending with it, moving those pixels to a colour the key was never pointed at. Fast movement. Each frame is keyed alone with no temporal filtering, so the blend along a moving edge is re-decided every frame. Anything near the key colour. A green shirt, a green reflection in glass, green-tinted hair dye: on flat-green, raising tolerance far enough to chase them turns the whole subject transparent at once — at tolerance 90 all 21,341 subject pixels are gone and IoU falls to 0.722122. Detail thinner than one pixel. The camera has already blended it into its neighbours. No keyer separates what the sensor never resolved." }, { "heading": "What to do instead", "text": "Most of this is won on the shoot day, not at the slider. Light the subject separately from the backdrop, put distance between them so bounce falls off, add a hair or rim light on the far side so strands separate by brightness rather than hue, and avoid backlighting that pushes screen colour into the hair. If a shot needs smoke or glass in frame, plan for it not to key cleanly. For the few shots that need a hand-built matte — a hero product frame someone will see full screen — the right tool is a rotoscope brush in a real editor. This tool is a colour key and does not pretend otherwise. If you want a soft edge for compositing, the transparent PNG export carries 8-bit alpha, but at default settings there are only two values in it. The tolerance page covers where the thresholds sit, and the artifact guide covers telling a soft edge from a fringe. For footage, the export settings page explains what each format keeps." }, { "heading": "Where the numbers come from", "text": "Every figure here comes from the synthetic 320×240 fixtures in the chroma-key-reference repository, also published as a preprint (DOI 10.5281/zenodo.22916767). They are flat rectangles of solid colour scored against a binary ground-truth matte by mean IoU, which measures how well a backdrop separates from a solid subject. They contain no hair, no smoke, no glass, no motion blur and no soft edges, so they say nothing about how those come out. Every claim on this page about hair, smoke and glass is mechanism read off the algorithm, not a measurement. We did not test those cases and will not quote a number we did not produce." }, { "heading": "Related image pages", "text": "Green Screen Remover (home) Blue Screen Remover Remove a Green Screen Without Uploading Anything Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/chroma-key-vs-luma-key", "title": "Chroma Key vs Luma Key — How This Tool Decides Which One to Use", "description": "Chroma key or luma key: the one line that picks between them, why a white wall falls back to brightness, what changes when it flips, and where neither one separates anything.", "markdown": "/ai/pages/chroma-key-vs-luma-key.md", "sections": [ { "heading": "Chroma key vs luma key: which one this tool uses, and when it switches", "text": "A chroma key measures colour. A luma key measures brightness. This keyer uses one or the other depending on the colour you key out, without telling you which. Here is the rule, what changes when it flips, and the shots where neither one separates anything." }, { "heading": "The one line that decides it", "text": "Every parameter starts with one test on the key colour: neutral = Math.sqrt(ku*ku + kv*kv) < 7. If the colour you asked it to remove has essentially no colour in it, there is nothing to measure in the Cb/Cr plane, so the pixel loop switches to luminance — the only thing that can tell a white wall from a white shirt. There is no toggle. The mode follows the key colour, and the presets show how wide the gap is. Green #00b140 measures Cb −26.6337 / Cr −79.3117, magnitude 83.6643. Blue #0047bb is 69.9803 / −44.9322, magnitude 83.1633. A white wall #f2f2f2 is 0.0000 / 0.0000, magnitude 0.0000, flagged neutral — and so are black and mid grey. Broadcast green and blue sit further out again at 136.1412 and 129.1750. Nothing lands anywhere near the threshold of 7: a backdrop is either clearly coloured or clearly neutral." }, { "heading": "What each one measures", "text": "Chroma. For each pixel: du = Cb(pixel) - Cb(key), dv = Cr(pixel) - Cr(key), d = sqrt(du*du + dv*dv). Under the threshold it is background, over it subject. Hue and saturation decide it. The practical upside: anything genuinely colourless — a grey laptop, a black cable, white text — sits at the origin of that plane, far from a green or blue key, and is safe. Luma. d = |luma(pixel) - luma(key)|, where luma is 0.299R + 0.587G + 0.114B. Hue stops mattering entirely. The distance is absolute, so a pixel brighter than the key by the same amount counts just as much as one darker. Key a white wall at luma 242 with the default tolerance of 25 and the radius is 37.5: everything from 204.5 upward is treated as backdrop, including a shirt brighter than the wall behind it." }, { "heading": "What changes when it flips", "text": "Three constants change together. The threshold scale goes from 118 to 150, feathering from 46 to 60 and trimming from 62 to 70 — luminance spans a wider range than chroma distance. At the default tolerance of 25 that puts a colour key's threshold at 29.5 and a neutral key's at 37.5. A number you tuned on a green screen does not transfer to a white wall. The other thing to know: the test runs on the key colour, not on your footage. Point it at a warm off-white wall and it may have just enough chroma to fail the neutral test, so you get a colour key with a very small chroma magnitude — a tight, fragile threshold that follows tint instead of brightness. That is the mode you did not want, and there is no way to ask for the other one." }, { "heading": "What the neutral fixture shows", "text": "The neutral-white fixture is a flat #f2f2f2 field with a #9a9a9a rectangle on it — the two differ by 88 luma and by no chroma at all. At the shipped defaults it scores IoU 1.000000: all 49,579 backdrop pixels cleared, none of the 27,221 subject pixels touched. Automatic key detection read #f2f2f2 exactly, with a maximum channel delta of 0. Sweeping tolerance across it gives a cliff, not a curve. Every value from 0 to 55 stays at 1.000000. At 60 it drops to 0.645560 and stays there through 100. That is arithmetic: the radius is (tolerance/100) * 150, so tolerance 55 gives 82.5, under the 88-luma gap, and tolerance 60 gives 90, past it. One step over the margin removes the entire subject, and no further setting brings any of it back. For contrast, a green-dominant baseline scored 0.000000 on the same fixture even when handed the best possible threshold. That is the whole argument for the fallback: a colour rule has nothing to grab on a white wall." }, { "heading": "When this does not work", "text": "A subject close to the backdrop in brightness. On a neutral backdrop, brightness is the only separation you have. This fixture had 88 luma of margin; a pale object on a pale wall has far less, and one tolerance step past the margin took out all 27,221 subject pixels at once. Same brightness, different colour. A luma key cannot see colour. A grey prop on a grey backdrop is not recoverable at any slider position. Anything brighter than the wall. The distance is absolute, so on a wall keyed at 242 a highlight reaching 250 luma is inside the default radius and gets keyed out along with the wall. Shadow or uneven light on a neutral backdrop. Not measured. Our brightness sweep covers a green screen only, with no numbers for a wall carrying a gradient. Treat it as untested. You cannot choose the mode. There is no switch. The key colour decides, and changing it changes what gets removed. Off-white backdrops land in the wrong mode. A cream or warm-grey wall can fail the neutral test and be keyed by colour instead, with the tighter 29.5 threshold and a matte driven by tint. Automatic key detection can flip it for you. It takes the median of the outer 5% border of the frame. If the wall does not reach the edge of the shot, the median reads whatever else is there, and both the key colour and the mode change with it. Every number here is synthetic: 320×240 flat rectangles of solid colour scored against a binary ground truth by mean IoU. No real walls, no fabric, no clothing, no shadows on neutral backdrops. Everything said above about those is mechanism, not a measurement." }, { "heading": "Which one you are getting", "text": "Green or blue screen, you get a chroma key — that is the default and it is what the tool is built for. White, black or grey backdrop, it switches to brightness on its own. Neither needs a decision from you. What decides the outcome is the separation: hue distance on a colour screen, brightness distance on a neutral one, and brightness is the easier of the two to lose. If you are working without a screen at all, the white background page covers what that fallback is actually good for. The tolerance page explains where the threshold sits in both modes, the blue vs green page covers picking a coloured screen, and the artifact guide helps when the result is close but not clean. Start from the tool if you just want to drop a file in." }, { "heading": "Where the numbers come from", "text": "Every figure on this page comes from the synthetic 320×240 fixtures in the chroma-key-reference repository, also published as a preprint (DOI 10.5281/zenodo.22916767). They are flat rectangles of solid colour scored against a binary ground-truth matte by mean IoU, and the threshold arithmetic is read straight from the keyer source. They are not real photographs of real walls. The neutral-white result describes one grey rectangle on one flat grey field. Claims about clothing, shadows on neutral backdrops and uneven room light are mechanism, not measurement — we did not test those." }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) Green Screen Lighting Guide DIY Green Screen at Home Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/about", "title": "About — Green Screen Remover", "description": "Who runs Green Screen Remover, what the tool actually does, why it is free, how it is built, and what it deliberately does not do.", "markdown": "/ai/pages/about.md", "sections": [ { "heading": "About", "text": "Green Screen Remover is a single-purpose web tool: drop in a video or a photo shot against a green or blue screen, and get the subject back with the backdrop removed — as a transparent PNG, a PNG sequence or a transparent WebM." }, { "heading": "Who runs this site", "text": "This site is run under the name Green Screen Remover. It is an independent, single-page tool site — not a studio, an agency or a media company. What it publishes is the tool itself and the handful of pages that explain it." }, { "heading": "What the tool actually does", "text": "The whole job happens in your browser tab. When you drop a file in, the page: reads the backdrop colour from the border of the frame, so you rarely have to pick it by hand; converts every pixel to YCbCr and measures the distance from the key colour in the Cb/Cr plane only — which is why a shadow falling across the screen still keys out; switches to a luminance key when the backdrop has almost no colour in it, the one case a chroma key cannot separate a white wall from a white shirt; pulls the green light that bounced off the screen back out of hair, skin and light clothing; hands you the result as a transparent PNG, a ZIP of PNG frames, or a transparent WebM recorded from the keyed canvas." }, { "heading": "Why it is free", "text": "There is no server doing the work, so there is no per-file cost to recover and nothing to charge per export. The site carries one disclosed affiliate link instead — the disclosure is on every page, in the footer, and next to the link itself. That is the entire business model: your file never leaves your machine, and the page is sustained by that single link, not by ads or subscriptions." }, { "heading": "What it deliberately does not do", "text": "It does not upload your file. There is no server-side processing and no render queue. It does not ask for an account, an email address or a payment. It does not stamp a watermark on the result. It does not pretend to have a manual brush or a rotoscope tool. Colour is the only lever this key has, and the page says so rather than showing you a button that does nothing. It does not keep your file. Close the tab and it is gone." }, { "heading": "How it is built", "text": "One HTML file, one stylesheet, one script. No framework, no build step, no CDN and no web fonts. The tool works offline once the page has loaded, which is the honest test of whether the processing really is local." }, { "heading": "Contact", "text": "Questions, bug reports and corrections go to contact@greenscreenremover.net. The contact page says what to include so a report can actually be acted on. Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/privacy", "title": "Privacy Policy — Green Screen Remover", "description": "How Green Screen Remover handles your files and your data: local processing, no upload, no account, no advertising, and what the one disclosed affiliate link means for third-party tracking.", "markdown": "/ai/pages/privacy.md", "sections": [ { "heading": "Privacy Policy", "text": "This policy explains what happens to your files and your data when you use Green Screen Remover at greenscreenremover.net. Last updated: 17 September 2026" }, { "heading": "The short version", "text": "Files you open in the tool are processed inside your browser. They are never uploaded to us or to anyone else. We do not ask you to create an account, and we do not collect your name, email address or payment details. This site has no advertising, no analytics, and no tracking pixels. It contains one disclosed affiliate link, and the destination site is run by someone we are not part of." }, { "heading": "What the tool does with your files", "text": "When you drop a video or an image onto the tool, the file is read and decoded by your own browser using the Canvas API. The keying, the spill removal and the exports all run on your device. The file is not sent to a server, there is no upload step and there is no render queue. Because nothing is uploaded, nothing is stored on our side. When you close or reload the tab, the file is gone from the page. If you download a result, it is saved by your browser to the location you choose, on your own machine. A practical consequence worth knowing: the tool keeps working if you go offline after the page has loaded. That is the honest test of whether the processing really is local." }, { "heading": "What we do not collect", "text": "No account, no sign-up, no email address, no password. No copies of your files, your frames or your exports. No names, addresses or payment information — there is nothing to pay for." }, { "heading": "Cookies and local storage", "text": "The tool itself does not use cookies. The only cookies you might encounter are the ones set by the destination site after you click the affiliate link — those are governed by that site’s privacy policy, not ours. You can block or delete cookies in your browser settings at any time. Blocking them does not stop the tool from working, because the tool does not depend on them." }, { "heading": "Advertising and affiliate links", "text": "This site has no advertising. We do not run ad networks, analytics, or tracking pixels, and nothing on this page sets a cookie on your behalf. What this page does have is one affiliate link. When you click it, you leave this site and go to a third-party page operated by someone we are not part of. That third-party site may use cookies or similar technologies to record that the visit came from here, and may collect information under its own privacy policy. We do not receive that information in a form that identifies you, and we do not control what the third party collects. You can read the disclosure next to the link itself, and you can read the general one in the footer. Blocking or clearing third-party cookies in your browser also works, and the tool keeps working either way." }, { "heading": "Server logs", "text": "This site is served by Cloudflare Pages. Like any web host, Cloudflare processes the technical data needed to deliver a page and to protect against abuse — the IP address the request came from, the browser's user agent, the URL requested and the time of the request. This is ordinary web-server logging. We do not use it to identify individual visitors, and we do not combine it with anything else." }, { "heading": "Analytics", "text": "This site runs no analytics script of its own. We do not build visitor profiles, and we do not track you across other websites." }, { "heading": "Children", "text": "This site is a general-purpose tool and is not directed at children under 13. We do not knowingly collect personal information from children. If you believe a child has sent us personal information, write to the address below and we will delete it." }, { "heading": "Changes to this policy", "text": "If what this site does with data changes, this page changes with it. The date at the top of the page is the date of the most recent revision." }, { "heading": "Contact", "text": "Questions about this policy, or a request to see or delete anything we hold about you, go to contact@greenscreenremover.net. Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/contact", "title": "Contact — Green Screen Remover", "description": "How to reach Green Screen Remover: the contact address, what to include in a bug report, and what we cannot help with.", "markdown": "/ai/pages/contact.md", "sections": [ { "heading": "Contact", "text": "One address, read by the people who build the tool. contact@greenscreenremover.net There is no phone line, no live chat and no support desk — this is a small site, not a call centre." }, { "heading": "What to write about", "text": "A file that will not key. Tell us the format and roughly what it is — a phone clip, a screen recording, a still from a camera. A file that will not open at all. Say which browser and version you are on, and what the page said when you dropped the file in. A result that looks wrong. Spill left on hair, a halo on the outline, part of the subject vanishing. Say which controls you moved and roughly where you left them. A correction. If something on this site is inaccurate or misleading, we would rather fix it than leave it up." }, { "heading": "What makes a report actionable", "text": "Which browser and version you are using. The kind of file — container and codec if you know them, or just \"MP4 from an iPhone\". What you expected and what you got instead. A screenshot of the page if the problem is visual. Please do not send the media file itself unless it is small and you are happy to share it. Because the tool never uploads anything, we have no copy of your file and cannot look it up." }, { "heading": "What we cannot help with", "text": "Recovering a file. Nothing you open in the tool is stored anywhere, so there is nothing on our side to recover. Editing your video. This is a keying tool, not an editor. It removes a backdrop; it does not cut, grade or stabilise footage. Ads, sponsorship or partnership pitches. This page is a free tool with one disclosed affiliate link — there is no ad slot, no sponsorship tier and no partnership programme." }, { "heading": "Privacy", "text": "Messages you send us are used to answer you and nothing else. What the site itself does with data is set out in the privacy policy. Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-camera-settings", "title": "Green Screen Camera Settings: Shutter, Frame Rate and ISO | Green Screen Remover", "description": "Choose camera settings for cleaner green screen footage: control motion blur, exposure, ISO, focus and white balance, then test the actual key before recording.", "markdown": "/ai/pages/green-screen-camera-settings.md", "sections": [ { "heading": "Green Screen Camera Settings: Shutter, Frame Rate and ISO", "text": "A clean key starts with the recorded image. Shutter speed affects the blur around moving hands, exposure affects the colour information available to the keyer, and automatic adjustments can make an otherwise usable edge change throughout a clip. Choose settings by testing your subject in motion under the lights you will actually use." }, { "heading": "Start with a stable recording", "text": "Set up the screen and subject lighting before choosing camera settings. Keep the backdrop evenly lit, avoid deep folds, and move the subject away from it when space allows. Camera controls cannot remove a cast shadow or stop green light bouncing onto clothing. The green screen lighting guide explains those setup decisions. Frame the intended action, including raised hands and any props. Record a short rehearsal and inspect the original file. Watch for exposure changes when a pale hand enters the frame, focus shifting to the backdrop, and colour changes as the subject moves. Where your camera allows it, lock exposure and white balance after setting them. Lock focus only when the subject stays within the focused area." }, { "heading": "Shutter speed: balance blur against the finished motion", "text": "A longer exposure records more movement within each frame; a shorter exposure reduces that blur. Nikon explains this relationship in its guide to camera settings for video. For keying, a moving edge matters because it can contain a mixture of subject and backdrop colour. Removing green from that mixture may leave a thin, weak or discoloured outline. Try a shorter exposure if fingers or hair become broad translucent streaks. Then check playback: very crisp individual frames can produce motion that looks abrupt. Judge the keyed result at its intended viewing speed as well as while paused. There is no shutter setting that simultaneously removes all blur and preserves every possible motion style. Shortening the exposure also lets less light reach the sensor. Compensate with suitable lighting or another exposure control, then test again. Do not solve blur by accepting a severely underexposed subject. Check the lights for visible bands or flicker after changing shutter speed; some artificial lighting interacts poorly with particular camera timings." }, { "heading": "Frame rate: choose for the actual delivery", "text": "Frame rate describes how often frames are recorded. Shutter speed describes the exposure within a frame. Raising frame rate does not automatically make a blurred edge sharp. Select the recording rate for the intended playback and editing workflow, then assess shutter speed separately. If you plan slow motion, check how your editor interprets the recording before shooting the whole scene. Extra frames are useful only when the workflow can use them. Also inspect whether changing recording mode alters the crop, available exposure controls or image detail on your camera. Do not assume every mode produces the same image apart from motion timing." }, { "heading": "ISO and exposure: protect useful colour information", "text": "Give the camera enough light to record the subject and backdrop cleanly. Trying to brighten a dark recording later can reveal noise and weak colour separation. Random colour variation near an edge can become changing transparency when a keyer applies the same colour threshold across successive frames. Use an ISO setting that produces a usable image on your camera under your lighting. Avoid treating the lowest available setting as an absolute requirement: a dark result is still a problem. Improve illumination where possible, and compare actual recordings rather than relying on the brightness of the camera display. Stanford's exposure teaching tool explains the tradeoffs among exposure time, aperture and ISO. Check bright clothing, skin highlights and the backdrop for lost detail. A green screen does not need to glow. Overexposing it can destroy useful colour distinctions and increase unwanted reflected light. Keep exposure stable throughout the action so the key does not have to follow a changing recording." }, { "heading": "Focus, aperture and white balance", "text": "Focus on the subject and choose enough depth of field for their movement. A shallow focused area can make a face sharp while hands or hair move out of focus. Those soft boundaries also mix subject and screen colour. Test the nearest and farthest positions the subject will reach; if focus is manual, mark a usable standing area. Set white balance for the lighting and keep it consistent during the take. Automatic colour shifts can change the screen colour after you have adjusted the key. Avoid changing creative filters between test footage and the final recording. Keep the original camera file for processing instead of passing it through a messaging app that may recompress it." }, { "heading": "Test the key before the full take", "text": "Open the rehearsal in the Green Screen Remover. Check the selected backdrop colour, then adjust tolerance cautiously. Inspect hair, fingers, clothing and props against both light and dark replacement backgrounds. Play the movement through: a still frame can hide an edge that flickers. If a shorter exposure improves a moving hand but makes the image noisy, revisit the lighting. If the whole screen clears only after clothing disappears, revisit colour separation. Change the camera or setup variable responsible for the defect, record another rehearsal, and compare it using the same key settings." }, { "heading": "When this does not work", "text": "Camera settings cannot separate a green garment from a similar green backdrop using colour alone. They cannot reconstruct detail already lost to clipping, missed focus, compression or motion blur. Transparent glass, smoke and fine flyaway hair may remain difficult because their recorded pixels combine foreground and background. For those shots, consider a different backdrop, a simpler movement, a reshoot, or an editor with masking and rotoscoping. The chroma key reference repository documents the underlying colour-key approach. Its synthetic fixtures are not evidence for a universal camera preset or a guarantee about your footage." }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) Green screen lighting guide DIY green screen at home Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-for-product-photos", "title": "Green Screen for Product Photos: From Cutout to White Background | Green Screen Remover", "description": "How to shoot and key product photos, preserve real product colours, check transparent PNG edges, and prepare a white background without losing important details.", "markdown": "/ai/pages/green-screen-for-product-photos.md", "sections": [ { "heading": "Green Screen for Product Photos: From Cutout to White Background", "text": "A green backdrop can make an opaque product easier to separate from its surroundings. It does not automatically produce a finished white-background catalogue image. The useful workflow is to photograph a clean outline, remove the screen, inspect the transparent cutout, then place it on white in an image editor." }, { "heading": "Choose the backdrop around the product", "text": "Start by looking at the object rather than the background. A matte cardboard box, a dark shoe or an opaque accessory can be a reasonable candidate when its colours differ from the screen. A green label, leaf pattern or translucent green cap creates a conflict: colour keying cannot know which green pixels belong to merchandise. For a green product, consider a blue backdrop only if blue is absent from the important details. Check the smallest printed marks and reflections as well as the main surface. If neither colour separates cleanly, photograph against a neutral background and use a workflow with a manual mask. Changing the background is often more useful than forcing a difficult key." }, { "heading": "Build separation before opening the tool", "text": "Use a smooth backdrop that fills the space behind the whole product. Remove folds that create deep shadows. Keep the object away from the backdrop where practical, and light the product and screen independently. The goal is an evenly coloured screen without sacrificing the lighting that describes the product's shape. A green tabletop can help expose the outline around a base, but it also reflects green light onto the underside. Inspect pale soles, white packaging and polished edges before continuing. Neutral cards placed outside the frame can help control what a reflective surface sees. If the green reflection remains prominent, change the setup instead of relying entirely on spill removal. Keep the camera stable and focus on the details customers need to see. Check small text, stitching and the far edge of the object at full size. Avoid clipped highlights that erase surface detail. Save the original photograph so that a later crop, colour correction or masking decision does not depend on an already processed cutout." }, { "heading": "Key the photograph in a deliberate order", "text": "Open the Green Screen Remover and load the photograph. Inspect the chosen key colour before adjusting the result. If the automatic choice does not represent the backdrop, use the colour picker or Pick from media on an unobstructed part of the screen. Do not sample a reflection on the product. Adjust Tolerance until the backdrop clears, watching the product at the same time. Stop if a coloured logo, a thin strap or a pale edge starts disappearing. Raising tolerance to remove every dark fold can cost real detail. An uneven screen may require a better photograph or a manual mask elsewhere. Next inspect Edge and Spill removal. Use edge adjustments conservatively around seams, bristles and narrow protrusions. Spill removal addresses a green cast, but it can also change colours you intended to preserve. Compare with the original, especially around printed artwork. A clean silhouette is not enough if the edit misrepresents the product." }, { "heading": "Check the cutout before making it white", "text": "Switch among the checkerboard, dark and light preview backgrounds. The checkerboard helps reveal missing pieces and leftover backdrop. A dark background exposes pale halos that can be hard to see on white. A light background makes dark contamination easier to notice. Also inspect holes inside handles and gaps between separate parts. Review both at full size and at the intended display size. A close inspection can reveal damaged lettering, while a smaller view shows whether the outline looks natural in a listing. If a rim disappears only when you shrink the image heavily, treat that as an unresolved edge problem rather than proof the cutout is clean." }, { "heading": "Export transparency, then compose the final image", "text": "Download the PNG and keep it as a reusable cutout. PNG can store an alpha channel describing transparency; this is documented in the W3C PNG specification. Transparency is different from white: a transparent area takes on the background beneath it when composited. In an image editor, place the PNG above a solid white layer. Set the canvas, crop and product position for the destination, then inspect the composite before exporting the delivery file. The remover's light preview is for checking the key; it does not add a permanent white background to the downloaded cutout. Follow the destination's current image requirements rather than assuming one export suits every marketplace. Consider the base of the object carefully. Removing the screen may remove or weaken the original contact shadow, leaving the product looking suspended. Preserve the original as a reference and handle any replacement shadow in the compositing editor. Keep it consistent with the lighting and avoid covering a damaged outline with a heavy artificial shadow." }, { "heading": "When this does not work", "text": "Clear glass, transparent packaging and glossy metal can carry the backdrop inside the product's visible surface. Removing that green does not reconstruct the appearance the object would have against white. Fine mesh and translucent fabric mix foreground and background too. These subjects often need controlled photography and detailed masking or compositing beyond a global colour key. The same limitation applies when the screen colour is part of the product, when blur merges the outline with the backdrop, or when highlights have lost important detail. Stop when further adjustment improves the background by damaging the merchandise. For a white object already photographed on white, switching to a brightness key cannot reliably distinguish equally bright subject and background. Reshoot or choose a manual selection workflow. For reproducible algorithm background, see the chroma-key reference preprint. Its synthetic fixtures are not product photographs; they cannot establish how a particular catalogue shot will turn out." }, { "heading": "Related image pages", "text": "Green Screen Remover (home) Remove Green Background from Photo — Export a Transparent PNG Transparent PNG Alpha — Why Keyed Edges Go White or Dark Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/why-green-screen-edges-look-jagged", "title": "Why Green Screen Edges Look Jagged: Pixels, Resolution and Feathering", "description": "Diagnose jagged green screen edges, separate scaling artifacts from a hard matte, and adjust tolerance and feathering without erasing useful detail.", "markdown": "/ai/pages/why-green-screen-edges-look-jagged.md", "sections": [ { "heading": "Why Green Screen Edges Look Jagged: Pixels, Resolution and Feathering", "text": "A rough outline can come from the original picture, the transparency mask, or the way the result is displayed. Find which stage introduces the steps before trying to hide them with a softer edge." }, { "heading": "Start with the source, not the slider", "text": "Open the original file and inspect a diagonal shoulder, a curved object, and a strand of hair. Compare those same areas in the keyed result at the same display size. If the original already has blocky contours, the keyer is working with damaged or limited information. If the source looks smooth but the cutout looks like a staircase, the transparency decision deserves closer attention. Also check whether the problem exists only in an enlarged preview. A raster picture is a grid of samples. Enlarging it can make that grid obvious, particularly along diagonals. Judge the result at its intended delivery size as well as close up; a highly magnified inspection is useful for diagnosis but can exaggerate defects that are invisible in normal use." }, { "heading": "Why a hard matte makes steps visible", "text": "A matte describes how transparent each part of an image should be. A hard decision makes a pixel either background or foreground. Along a diagonal boundary, those decisions follow the rectangular pixel grid. Partially transparent boundary pixels can make the transition look smoother when composited, because they allow the new background to contribute to the visible edge. That does not mean every outline needs a broad blur. A solid product edge and a motion-blurred hand contain different information. Smoothing should preserve the appearance of the photographed boundary. An opaque object that develops a translucent rim has usually been pushed too far, even if the staircase becomes less obvious." }, { "heading": "Resolution and scaling change the diagnosis", "text": "A small subject within a large frame still has limited samples describing its outline. Cropping tightly afterward does not create detail around the subject. For future shoots, frame the subject so it occupies useful image area while keeping the required movement inside the screen. Preserve the original file through the keying stage instead of repeatedly resizing and saving intermediate copies. Display software also matters. The MDN image-rendering reference documents how browser scaling algorithms can smooth an enlarged image or emphasize its pixel structure. If a cutout looks stepped only on a particular webpage, compare the downloaded file in another viewer before changing the matte. A display choice can make an otherwise acceptable export look rough." }, { "heading": "Adjust tolerance before adding softness", "text": "In the Green Screen Remover, begin by checking that the intended backdrop colour is selected. Move Tolerance gradually until the screen clears, watching both the empty backdrop and the subject. Stop if clothing, hair, or product details begin disappearing. More aggressive background removal can make an outline look thinner and more broken rather than cleaner. Then try moving Edge toward its positive side to soften the transition. This tool uses a range of colour distances to assign intermediate transparency; it does not redraw the contour or reconstruct missing hair. Moving Edge toward its negative side trims more of the colour boundary. That can remove a contaminated fringe, but it can also remove useful detail. Compare changes against the original instead of assuming that a cleaner silhouette is always more accurate. Keep Spill removal conceptually separate. It reduces unwanted backdrop-colour contamination in visible pixels. It is useful for a green cast, but it does not supply missing spatial detail. A green rim and a stepped silhouette may occur together, so inspect colour and contour separately as you adjust them." }, { "heading": "Use a repeatable export check", "text": "Choose a representative frame with both a solid contour and fine detail. Keep that frame visible while making the initial adjustments. Export a transparent PNG for a still-image check, and place it over a light background, a dark background, and the intended destination background. Inspect the contour at the intended size. Look for missing pieces, an unexpectedly transparent rim, and retained green patches rather than judging only the checkerboard preview. For video, inspect moving sections as well as the still frame. If the outline changes unpredictably, check the source for noise, shifting exposure, and uneven screen lighting. Compare the final destination with the exported file. If the defect appears only after resizing or conversion elsewhere, investigate that stage before repeating the key. A useful comparison changes one control at a time and keeps the background and viewing size consistent. Otherwise, a darker backdrop or a smaller preview can disguise an edge problem without improving the file." }, { "heading": "When this does not work", "text": "Feathering cannot recover a contour that was lost through poor focus, heavy compression, or insufficient source detail. It can make a damaged edge less conspicuous, but excessive softness trades a staircase for a halo. Reshooting or returning to a better original is often the practical fix. Colour-based keying also struggles when the subject and backdrop share similar colours. A green label, translucent glass, or reflected screen can be classified with the background. A global tolerance setting cannot reliably distinguish objects by meaning. Such shots may need a different backdrop, controlled reflections, or an editor with local masks and manual correction. Motion blur and fine hair contain mixtures of foreground and background. Simple colour-distance feathering is not a full reconstruction of those mixtures. If a result needs frame-specific repair or precise product outlines, use a workflow with that control. The chroma-key reference repository provides implementation context and synthetic test fixtures; these are not a guarantee of performance on your footage." }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) Chroma key artifacts — what each edge problem actually is Green screen tolerance, explained Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-vs-ai-background-removal", "title": "Green Screen vs AI Background Removal: Which Workflow Fits?", "description": "Choose between chroma key and AI background removal by looking at your footage, subject, edge detail and export needs, with practical tests and honest limits.", "markdown": "/ai/pages/green-screen-vs-ai-background-removal.md", "sections": [ { "heading": "Green Screen vs AI Background Removal: Which Workflow Fits?", "text": "Choose the method around the shot you can make and the details you must preserve. A controlled backdrop gives chroma key a colour to remove. AI background removal can separate a subject without that backdrop, but its result depends on the model and the scene." }, { "heading": "What each method uses to separate the subject", "text": "A chroma key compares pixel colours with a chosen backdrop colour and turns matching areas transparent. It does not need to recognise a person, a chair or a product. That makes colour separation useful for varied subjects, provided the screen colour stays distinct from everything you want to keep. Green clothing against a green screen creates an immediate conflict. AI background removal is a broad label. Many systems use learned segmentation to predict which parts of an image belong to a subject. Others add matting or interactive guidance. Google's MediaPipe image segmentation documentation shows why model choice matters: different models support different categories. A person segmenter should not be assumed to preserve every object that person holds. A segmentation confidence mask also does not automatically describe physical transparency." }, { "heading": "When a green screen is a useful starting point", "text": "Consider chroma key when you control the recording space and can keep the background evenly lit. Repeated presenter recordings, demonstrations and opaque product shots can benefit from a backdrop that stays consistent between takes. Check costumes, labels, props and reflections before choosing green or blue. The Blue screen vs green screen explains the colour conflicts to watch for. The useful control happens before recording. Smooth visible folds, separate the subject from the screen where space allows, and adjust the screen light independently of the subject light. Record a sample with the intended movement. If the key removes part of a sleeve, changing the sleeve or screen may be more effective than pushing tolerance until the background disappears." }, { "heading": "When AI removal is worth testing", "text": "If the footage already exists without a coloured backdrop, segmentation may be a practical route to a cutout. It can also suit a location where hanging and lighting a screen would interfere with the shot. Test the actual subject and background together: a clear portrait does not establish that the same workflow will preserve a bicycle, an open umbrella or an intricate product. Check what the tool actually exports. A preview with a replacement background does not prove that you can download a reusable transparent file. Also check whether processing runs locally or sends media to a service. Neither chroma key nor AI determines that deployment choice; privacy depends on the particular implementation and its documented handling of files." }, { "heading": "Use a short test that exposes the difficult parts", "text": "Use representative material before committing to a workflow. For planned filming, shoot tests of the same action under each intended setup. For existing footage, use the same source wherever the methods accept it. A colour key on a cluttered room is not a meaningful test of a properly prepared green screen. Include loose hair, hands crossing the body, gaps between fingers and any props that must remain visible. Inspect the cutout over a light background, a dark background and the intended final scene. Look for holes, coloured fringes and missing detail. Play video continuously. Pause around fast movement and occlusions, then watch again for flicker or boundaries that change shape. Open the exported file in the destination editor. Confirm that transparency survives and the result still looks acceptable at its final size. Judge the defects that matter to the job. A missing product handle may be unacceptable even if the face looks excellent. A soft edge may disappear at the final display size, while a flickering hand remains distracting. Keep the original source so you can change workflows without relying on an already altered export." }, { "heading": "When this does not work", "text": "A colour key cannot reliably distinguish the screen from foreground pixels with the same colour. Reflected green light can contaminate pale surfaces and edges. Heavy compression, deep screen shadows and motion blur can make foreground and background colours overlap. Increasing tolerance can remove more backdrop while also erasing real subject detail; feathering cannot recover detail that was never separated. AI removal can fail when the subject falls outside what the model handles well, when foreground and background are ambiguous, or when an object disappears behind another. Watch thin structures and objects detached from the body. These are reasons to inspect your result, not claims that every model fails on every such shot. Glass, smoke, veils and fine hair require special care with either approach. A partly transparent pixel mixes foreground and background information. Simply labelling it as subject or background does not reconstruct the original mixture. Difficult shots may need guided matting, masks, rotoscoping or a reshoot. If exact edge colour and transparency are essential, budget time for that work before promising a finished composite." }, { "heading": "How this site's tool fits", "text": "The Green Screen Remover uses colour and brightness rules rather than a learned subject model. Use it to test footage with a separable backdrop; do not expect it to recognise a person against an arbitrary room. Its reference repository documents the keying core and synthetic test fixtures. Those fixtures are not real footage or an AI comparison, and they do not establish a universal winner. Choose the workflow whose test preserves your required details and reaches your destination correctly. If you can improve the shoot, do that first. If you cannot, choose a removal method with enough correction controls for the problems your actual footage reveals." }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) Green screen lighting guide Blue screen vs green screen Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-wrinkles-and-shadows", "title": "Green Screen Wrinkles and Shadows: Fix the Backdrop Before Keying", "description": "Fix green screen wrinkles and shadows before recording, diagnose leftover patches, and adjust the key without damaging your subject, with practical limits.", "markdown": "/ai/pages/green-screen-wrinkles-and-shadows.md", "sections": [ { "heading": "Green Screen Wrinkles and Shadows: Fix the Backdrop Before Keying", "text": "A green backdrop can look acceptable in the room yet leave dark streaks and cloudy patches after keying. Start by checking the recorded image: folds change how light reaches the cloth, while a subject or object can cast a separate shadow. Fixing that physical cause usually gives you more room to adjust the key without damaging the subject." }, { "heading": "Why a wrinkle becomes a leftover patch", "text": "A colour key removes pixels that resemble the selected background colour. A folded screen presents surfaces at different angles to the light. Raised areas can catch bright highlights while recessed areas stay dark. The camera records a range of colours instead of a consistent backdrop, especially where a crease is deeply shaded or light reflects unevenly from the fabric. A shadow does not automatically make a screen impossible to key. A modest brightness difference may still leave a recognisable green background. The problem is whether the recorded background remains distinguishable from the subject. Deep shadows, noise and reflections make that separation harder. Raising tolerance to remove every remaining patch can also remove similarly coloured parts of clothing, hair or props." }, { "heading": "Diagnose the cause before moving a slider", "text": "Look at the original image alongside the keyed result. Follow the shape of each leftover patch. Narrow streaks aligned with folds suggest the cloth and lighting need attention. A silhouette that follows the performer suggests a cast shadow. A broad dark corner suggests uneven illumination or an obstruction. These are clues to investigate, rather than a reliable diagnosis from the matte alone. With the camera framing fixed, compare the screen with and without the subject in position. If the troublesome patch appears only when the subject returns, move the subject or change the direction of the subject light. If it stays in the empty shot, inspect the screen surface and backdrop lighting. Check the whole area used by moving hands and props, not just the space around a stationary face." }, { "heading": "Flatten the backdrop without creating new problems", "text": "Unfold the material fully and support it securely. Apply gentle, even tension along the edges rather than pulling one corner hard enough to create diagonal ridges. Keep clamps and supports outside the recorded area. If the fabric needs washing, steaming or ironing, follow its care instructions; heat or moisture can damage materials and coatings. Test an inconspicuous area when the care guidance permits it. You do not need to eliminate every tiny texture. Prioritise deep creases and folds that produce visible bands in the camera image. A smooth surface lit badly can still fail, and stretching cloth cannot fix a shadow cast by the performer. Recheck the framing after adjusting the screen so an uncovered wall or support does not enter the shot." }, { "heading": "Light the screen and subject with separate goals", "text": "Aim for a backdrop that looks reasonably consistent across the recorded area. Change the position and angle of available lights while watching the camera image. Softer illumination can reduce harsh shadow boundaries, but a light placed too close or aimed poorly can still produce a bright patch. Inspect the result rather than assuming a particular fixture or lighting diagram guarantees success. Give the subject enough separation from the screen to reduce cast shadows and green light bouncing onto edges. There is no universal distance that suits every room, lens and light. Move the subject forward when space permits, then check whether the framing still fits the backdrop. Reposition the subject light if it continues to throw a strong silhouette onto the cloth. See the Green Screen Lighting Guide for the wider setup. Set exposure and white balance deliberately and keep them consistent during the take when the camera allows it. Avoid a bright screen with washed-out areas and a dark subject that needs heavy correction later. Record a test with the actual costume and movements: an empty backdrop cannot reveal green reflections on glasses or a shadow made by an outstretched arm." }, { "heading": "Adjust the key after improving the recording", "text": "Open the test in the Green Screen Remover. Check the chosen key colour against an ordinary, representative area of the backdrop. A bright highlight or the darkest fold is a poor starting point for the rest of the screen. Increase tolerance gradually while watching both the leftover background and the subject. Stop if the subject starts developing holes or losing useful edge detail. Use edge adjustment for the boundary, and spill adjustment for unwanted screen colour on the subject. Neither is a substitute for clearing a large background patch. The official OBS Chroma Key Filter documentation likewise distinguishes colour similarity, edge smoothness and spill reduction. Its controls belong to OBS; their values should not be copied into another keyer as a preset." }, { "heading": "When this does not work", "text": "If a dark fold and an important part of the subject have overlapping recorded colours, a single colour threshold may not separate them cleanly. More tolerance can trade a background patch for a missing sleeve. Extreme underexposure, clipped highlights, translucent material and green reflections can also leave information that simple keying cannot reconstruct. Reshooting is often the clearest solution when the recording setup is still available. For irreplaceable footage, an editor with masks or rotoscoping may help isolate problem regions. That requires additional work and does not guarantee recovery of fine transparent edges. The chroma-key reference archive provides reproducible synthetic fixtures, not proof that a particular real recording can be repaired. This guide makes no measured performance claim for wrinkles or shadows." }, { "heading": "Check the result in motion", "text": "Review the keyed test over contrasting backgrounds and over the intended replacement. Watch for remaining patches, flickering boundaries and lost subject detail throughout the movement. If a fix only works on a paused frame, return to the lighting or accept that the shot needs more selective editing. Keep the original recording so you can revisit the key without working from an already altered export." }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) Green Screen Lighting Guide ? Light the Backdrop So the Key Comes Out Clean DIY Green Screen at Home ? Materials, Size and Wrinkles That Actually Key Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-clothing-colors-to-avoid", "title": "Green Screen Clothing: Colors and Materials to Avoid", "description": "Choose clothing for green screen filming: avoid color overlap, test reflective fabrics and fine patterns, and learn when wardrobe changes cannot save a key.", "markdown": "/ai/pages/green-screen-clothing-colors-to-avoid.md", "sections": [ { "heading": "Green Screen Clothing: Colors and Materials to Avoid", "text": "Avoid clothing that records as the same color as your backdrop. For a green screen, that usually means leaving green garments out of the shot, but the camera image matters more than a color name on a label. Reflections, translucent fabric and small patterns can create problems even when the main outfit looks different from the screen." }, { "heading": "Why a green sleeve can disappear", "text": "A chroma key separates pixels by their recorded color. It does not understand that a green area belongs to a jacket while another belongs to the backdrop. If both fall within the color range being removed, both can become transparent. A logo, shoelace or decorative stitch can therefore develop a hole even when most of the outfit survives. The official OBS Chroma Key Filter documentation explains that increasing similarity removes colors more aggressively and may remove content you want to keep. That principle helps explain wardrobe failures across color keyers; OBS settings are not transferable presets for this website. Making a key more selective can preserve the shirt while leaving parts of the screen behind." }, { "heading": "Check the whole outfit for green", "text": "Inspect sleeves, collars, trousers, shoes and accessories, including details exposed only during movement. Green prints on a dark shirt still count. Yellow-green, turquoise or muted olive may be troublesome if their recorded colors approach the screen color or if a broad tolerance is needed. They are candidates for testing, rather than a universal list of forbidden shades. Choose an alternative whose visible colors stay distinct from the backdrop under the actual lights. Plain burgundy, purple or blue fabric can be useful starting points for a green screen, but no named color guarantees success. If you switch to a blue screen, blue clothing becomes the obvious concern instead. Plan the wardrobe and backdrop together; see Blue Screen vs Green Screen before changing the setup." }, { "heading": "Prefer matte, opaque fabric when you have a choice", "text": "Shiny fabric can reflect the green surroundings. Satin, sequins, metallic decoration and glossy accessories may pick up green patches that move as the performer turns. The key sees the reflected color, not the material underneath it. Test the garment from the side as well as straight on, since a reflection may appear only at a particular angle. Matte fabric is often easier to manage because it produces fewer sharp reflections, but it can still receive green light bouncing off the screen. Pale clothing makes a color cast especially noticeable. White is not automatically unusable, and black is not automatically safe: inspect what the camera records. Separate the subject from the backdrop where space allows and control reflected light before trying to neutralize everything with spill removal. Lace, mesh and sheer layers introduce another problem. Some background remains visible through them, so preserving their appearance requires partially transparent edges and interior detail. A simple color key may leave tint, erase threads or flatten the material's texture. If those details define the costume, test them early enough to change the fabric or choose a different compositing workflow." }, { "heading": "Fine stripes are a camera problem as well", "text": "Small repeating patterns can produce moir? or unstable detail when the camera samples them. Scaling and compression can further change how a pattern looks. This is different from directly keying out a green shirt: the pattern may already look wrong in the original recording before any background removal. Do not try to repair a camera sampling problem solely by changing tolerance. Watch a recorded test at its intended display size. Turn, lean forward and move through the planned framing. If stripes shimmer or crawl, try a plainer garment or a larger pattern, then record again. Avoid assuming that a clean still photograph proves the same shirt will remain stable throughout a moving shot." }, { "heading": "Run a wardrobe test before the full take", "text": "Use the intended camera, lighting, framing and outfit. Include gestures that reveal cuffs, jacket linings and accessories. Keep exposure and white balance consistent when the camera permits it. Review the original recording first, looking for green reflections, blocked-up dark areas and unstable fabric detail. Fix visible recording problems while the subject and equipment are still available. Open the sample in the Green Screen Remover and inspect both the backdrop and clothing as you adjust tolerance. Start with enough separation to clear the ordinary screen areas, then watch for disappearing garment details. Use edge adjustment sparingly around the silhouette. Spill removal addresses unwanted color cast; it cannot restore a section that the transparency calculation has already removed. Inspect the result over a light background, a dark background and the intended replacement. Watch the full movement instead of approving a paused frame. Look for holes inside the outfit, flickering seams and green highlights that travel across shiny fabric. Keep the original sample so you can compare a wardrobe or lighting change against the same starting material." }, { "heading": "When this does not work", "text": "If essential clothing and the backdrop share indistinguishable recorded colors, a single global color key cannot reliably tell them apart. Lowering tolerance may leave the background; raising it may remove the garment. Strong spill, transparent fabric and blurred moving edges can also mix subject and screen information. A different shirt helps future takes, but it does not reconstruct missing information in an existing recording. For footage you cannot reshoot, an editor with masks or rotoscoping may let you protect selected regions. Expect additional work, especially where the garment crosses the silhouette. The chroma-key reference archive provides reproducible synthetic fixtures, not wardrobe trials on real footage. This guide claims no measured success rate for any color or fabric. A test of your own shot remains the deciding evidence." }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) Blue screen vs green screen Green screen spill Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] }, { "path": "/green-screen-distance-from-subject", "title": "Green Screen Distance from Subject: Spill, Shadows and Focus", "description": "Find a workable subject-to-green-screen distance by testing spill, shadows, framing and focus, with practical steps for small rooms and honest limits.", "markdown": "/ai/pages/green-screen-distance-from-subject.md", "sections": [ { "heading": "Green Screen Distance from Subject: Spill, Shadows and Focus", "text": "Place the subject far enough from the green screen to control reflected green light and keep troublesome shadows away from the visible backdrop. Then check that the screen still covers the shot and that the subject stays sharp. There is no universal distance that solves all of these problems: screen size, lighting, framing and movement determine what works in your room." }, { "heading": "Start with the shot, then choose the gap", "text": "Decide what the camera must see before moving furniture. A seated portrait needs a different area of green than a standing presenter with outstretched arms. Rehearse the widest gesture and the closest approach to the camera. The backdrop must cover the visible area behind the entire movement, including hair and hands, rather than just the resting pose. Distinguish the subject-to-screen gap from the camera-to-subject distance. Moving the subject toward the camera changes framing unless you also adjust the camera or lens. Moving the screen backward changes its coverage in the image. After either change, inspect the actual recording frame rather than judging the setup from beside the camera." }, { "heading": "Use separation to reduce green spill", "text": "The green surface reflects light into the room. Some can reach skin, pale clothing and hair, producing a green cast even where the subject should remain opaque. Increasing separation can help reduce this contamination, but its effect depends on the size and brightness of the screen and on nearby reflective surfaces. Distance is a useful adjustment, not a guarantee. Look at the original camera image before applying a key. Inspect the shoulders, underside of the chin, loose hair and shiny accessories. Move the subject away from the screen while keeping the subject lighting consistent, then compare recorded samples. If the green tint remains, check whether a bright screen, reflective wall or glossy garment is contributing. See the Green Screen Spill Suppression guide for the distinction between color cleanup and transparency. Do not simply make the backdrop as bright as possible. Aim for a usable, reasonably even recorded green while exposing the subject appropriately. Extra screen brightness can increase reflected light, and clipped recording channels lose information. Adjust the screen and subject lights separately where the setup allows it." }, { "heading": "Check where the shadows actually land", "text": "A subject close to the backdrop can cast a conspicuous shadow across the green. Moving forward changes the shadow's position and softness, but the result also depends on the light's direction and apparent size. Separation alone does not ensure that the shadow leaves the shot. Watch the backdrop while the performer turns and gestures. Try changing the subject light's angle or controlling where it falls, then record again. Keep the screen lighting even without letting the performer block it. If a dark patch follows an arm, diagnose it as a lighting and placement problem before widening the key. A setting that clears the shadow may also remove similar colors in the subject. The chroma-key reference repository includes a shadow-distance table based on synthetic fixtures, not real footage. It models how changing backdrop brightness changes color distance; it does not measure physical subject spacing. It therefore cannot justify a recommended gap in feet or meters. Use it to understand the keyer's boundary, not as a room-layout formula." }, { "heading": "Balance background softness with subject focus", "text": "Separating the subject from the backdrop can make it easier to render the screen out of focus, depending on the lens, aperture and camera position. Mild background softness may make small surface details less prominent. It cannot remove broad lighting gradients, eliminate spill already falling on the subject or restore information missing from a dark recording. Focus on the subject and test the intended movement. Opening the aperture to soften the backdrop also reduces the range that appears acceptably sharp. If a presenter leans outside that range, their blurred silhouette becomes a keying problem. Favor a reliably sharp subject over an attractively blurred screen. Review recorded motion rather than trusting a small camera preview." }, { "heading": "A practical placement test for a small room", "text": "Begin with the largest usable gap that still permits safe movement, proper framing and full backdrop coverage. Record a short rehearsal with the final clothing and lights. Examine the unkeyed recording for green reflections, moving shadows and soft subject edges. Change a single placement or lighting factor, record another sample and compare the same gesture. Open the samples in the Green Screen Remover. Adjust tolerance while watching both the background and the subject. The official OBS Chroma Key Filter documentation likewise warns that more aggressive color similarity can remove content you want to keep. Its controls are not interchangeable presets for this tool; the shared lesson is to inspect what disappears. Compare the keyed result over light, dark and intended replacement backgrounds. Check moving hair, shoulders and hands, plus holes inside clothing. If more separation improves spill but exposes the edge of the screen, reconsider framing or backdrop position. In a cramped room, a tighter shot can be a more useful compromise than forcing a wide shot into inadequate coverage." }, { "heading": "When this does not work", "text": "More distance cannot separate clothing that records as the same color as the screen. It also cannot recover a crisp outline from severe motion blur or restore clipped detail. Glass, shiny surfaces and translucent fabric may continue to reflect or transmit green even after the subject moves forward. Those areas can require masking, specialist compositing or a reshoot. If no position provides both backdrop coverage and manageable lighting, simplify the shot, use a larger suitable backdrop or change the recording location. Keep the original files and approve an actual keyed rehearsal before the full session. Choose the placement that survives the intended movement; a clean still frame is not enough evidence." }, { "heading": "Related guides & setup pages", "text": "Green Screen Remover (home) Green Screen Lighting Guide Green Screen Camera Settings: Shutter, Frame Rate and ISO Some links on this site are affiliate links. If you buy through one, we may earn a commission — the price you pay stays the same. Green Screen Remover · chroma key for video and images · runs entirely in your browser." } ] } ] }