No watermark on the free export.Transparent WebM, GIF or green screen, straight out — free exports go up to 1920 px.
Automatic matting removed the green screen from the equation, but it didn't remove physics. The model behind VideoBGNinja (Robust Video Matting) predicts, pixel by pixel and frame by frame, what belongs to you and what belongs to the background. The easier you make that decision at recording time, the cleaner your edges come out. These seven habits cost nothing and make the difference between "pretty good" and "did you rotoscope that?".
You don't need good equipment for this. Looking at the 1,536 clips our GPU processed across every tool on this site in the 40 days before this was written, the average upload was 1.72 megapixels and the largest was 8.29 — for comparison, a single 4K frame is about 8.3 megapixels, so almost nobody is sending that much. Most uploads (930 of them) sat in the 1-to-2-megapixel range: think a basic phone-camera window or a cropped webcam frame, not a cinema shot. A further 503 landed between 2 and 4 megapixels, and only 85 reached 8 to 16.
That matches the advice below: contrast, light direction and a steady camera matter far more to a clean cutout than resolution does. The model processes at up to 1920px on the long side anyway, so filming in 4K mostly costs you upload time, not edge quality. If your footage looks like everyone else's — a phone held at arm's length in a normal room — it's already the kind of input this tool sees every day.
This is the single biggest factor. Dark hoodie against a dark bookshelf, or blonde hair against a beige wall, gives the model almost nothing to separate. Aim for a clear tonal difference: dark clothes in front of a light wall, or the reverse. You don't need a special colour, just difference.
Face a window or put a lamp behind the camera pointing at you. Soft, frontal light keeps your whole outline visible. Harsh light from one side plunges half of you into shadow, and a shadowed shoulder against a dark room is a coin-flip for any matting model.
A bright window or lamp behind you is the classic cutout-killer. Backlight blooms around your head and shoulders, your camera crushes you into a silhouette, and the glow becomes a halo that belongs half to you and half to the background. If you can't avoid the window, close the curtains or film facing it instead.
Tripod, stack of books, phone leaning against a mug: anything stable. RVM tracks you through time with a frame-to-frame memory; a static background lets that memory lock on and keeps edges rock-steady. Handheld shake forces the model to re-find you every frame, which shows up as edge shimmer.
The model cuts out people: all the people it sees. A housemate crossing the background will be faithfully cut out with you. For a predictable result: one person, nobody walking behind you, and no large posters or screens showing people.
Solid colours that differ from your background are ideal. Tricky items: see-through and lacy fabrics (the model must guess partial transparency), very fine mesh, extremely reflective jackets, and (obviously) anything the same colour as the wall behind you. Unlike a green screen, though, green clothes are totally fine.
Film at 1080p if you can, hold the framing so you're a reasonable size in frame, and keep enough light that your phone doesn't crank up ISO (noise blurs your edges before the tool ever sees them). Fast, whip-like movements cause motion blur: where your arm is smeared across ten pixels, no algorithm can draw a crisp line.
The seven habits get you most of the way; these details squeeze out the last bit of edge quality and cover the questions the tips raise.
Your original sound is preserved in the transparent WebM and the green-screen MP4, so a clean recording saves you re-syncing later. Get the mic close — a phone at arm's length in a quiet room is fine, but across a room it picks up echo. Turn off a noisy fan or AC while you record; matting cleans up your background, not your sound. If it's already recorded and the voice still sounds distant or roomy, CleanGrab's speech enhancer can pull it forward afterwards, free.
Honesty beats marketing, so here is the real limitation list:
The tool is free and takes about a minute per clip, so treat the first upload as a lighting test: shoot 10 seconds, upload, look at your hairline and shoulders in the preview. Edges wobbling? Move the lamp, step away from the wall, swap the hoodie. A second take costs you 90 seconds and usually lands the clean cutout. Then use it as a transparent OBS overlay or as green-screen footage in any editor.
✂ Test your setup: upload a clip freeNo. A window in front of you or a normal desk lamp behind the camera is enough. What matters is direction (frontal), softness and avoiding backlight, not wattage.
No, the tool works against normal rooms and outdoor scenes. Plain and static is simply easier, so edges get cleaner. Avoid backgrounds full of moving people.
1080p is the sweet spot: processing runs at up to 1920px anyway, so 4K adds upload time but little edge quality. Anything from 720p up works.
Fine hair is the hardest part of any matting (automatic or chroma key). More contrast between hair and background, and more light, sharpen it noticeably.
Different tool for that: PicReviver's photo enhancer sharpens a soft photo and doubles its resolution, the still-image equivalent of getting more detail in frame here.
More guides: green screen vs AI matting · OBS overlay guide · the full video background remover guide