Open a tab, let it see your hand, and draw. Online drawing has meant dragging a mouse across a desk for twenty-five years; the camera you already own turns out to be a better input than the one you were using.
Free · no account · no download
There is no project to create, no canvas size to choose and no tool to pick before you can make a mark. The only step that takes any real time is the one where the tool measures your hand, and it happens once.
No download, no extension, no account screen. The canvas is the page.
The browser asks once. The feed is read inside the page to find your hand.
Two seconds with an open hand, one with a pinch. It is learning the gap between your fingers, not applying an average.
Thumb to index puts the pen down and your fingertip is the nib. Open the pinch and it lifts.
Thirty seconds is the realistic figure for a first visit, and most of it is the browser's permission prompt and the second you spend reading what it says. On a second visit the permission is already granted, so it is closer to ten.
The measurement is worth understanding, because it is the only part with no equivalent in mouse-era online drawing. The distance between a person's thumb and index finger when they pinch is not the same from person to person, so a fixed threshold makes the pen twitchy for some hands and unresponsive for others. Holding the two poses lets the tool place the threshold inside the gap it just measured on you — which is why the pen goes down when you pinch and not slightly before or after.
None of this makes a mouse bad at what it is for. It makes the point that a mouse was never designed for drawing — it was designed for pointing, and drawing with it is something people learned to tolerate.
The posture difference is the one nobody expects and the one people notice first. Drawing with a mouse folds you inward — wrist anchored, elbow tucked, eyes down at a surface eighteen inches away. Drawing in the air does the opposite: the arm moves from the shoulder for anything long, the hand stays up in front of you, and your head comes off the desk. It is a worse instrument for pixel-level precision and a noticeably better one for gesture, scale and anything you intend to film.
The honest limit is accuracy at small sizes. A hand held in mid-air is never perfectly still, which is why a low-pass filter runs underneath at three different strengths depending on how fast you are moving — light while you are idle, heavier while you drift, almost off when you move quickly so fast strokes do not lag behind you. That smoothing is what makes air-drawn lines look deliberate rather than nervous, and it is also why very small detail work is still easier with a stylus on glass.
It is also worth saying who this is not for. If you are retouching a photograph, laying out a UI to the pixel, or inking line art that has to hold up at 300% zoom, a stylus on glass is the right instrument and this is not trying to replace it. The people it fits are the ones drawing to explain, to demonstrate, to film, or simply to make something in the next ten minutes without buying anything first.
What you get in exchange is that nothing sits between the intention and the mark. There is no device to pick up, no surface to find room for on the desk, and no mapping to learn between where your hand is and where the cursor went — your fingertip is the nib, so the line appears where you are pointing.
A camera-driven canvas would be a novelty if the tools behind it were thin. These are the ones that matter once the first five minutes are over.
Marker, Neon, Velocity, Rainbow, Mono and Sparks. Velocity is the expressive one — stroke width follows how fast your hand is moving.
Neon, Sunset, Ocean, Y2K, Monochrome and Forest, eight colours each, with an HSV wheel behind them when none of the forty-eight is the one you meant.
Draw roughly, hold the pinch still, and the stroke becomes a clean line, circle, ellipse, arc, triangle, rectangle or polygon.
Strokes are stored as data rather than pixels, so you can grab a finished line and pull it into a different shape.
Pan with a three-finger pinch, zoom with thumb and middle. There is no edge to reach.
Undo is the first item in the session menu — raise index, middle and ring together to open it.
A snapshot writes a full-resolution PNG. A recording captures the whole session as video — MP4 where the browser can manage it and WebM otherwise — for up to sixty minutes. The timelapse export at 2×, 3× or 5× is the one most people actually post, because the process is more interesting than the result almost every time.
Every export carries a small getairva.com mark in the corner, on images and video alike, and videos end with a short card. It is small on purpose, and there is no paid tier that removes it — the free version is the whole version.
What there is not: cloud storage, project files or accounts. What you make lives in the tab until you export it, which is worth knowing before you spend an hour on something. There are also no layers and no collaboration mode — two people cannot draw on one canvas from two machines. None of that is coming as a paid upgrade; it simply is not built, and a page that implied otherwise would be selling you a different product.
The one requirement worth stating plainly is light. The camera has to be able to see your hand, and the classic failure is a bright window directly behind you, which turns your hand into a silhouette — and a silhouette has no knuckles to track. Facing the light rather than sitting in front of it fixes almost every bad first session, and costs nothing to try.
The hand tracking runs inside your browser tab — a self-hosted MediaPipe model reading twenty-one points on your hand about thirty times a second. Your camera feed never leaves your device: not uploaded, not recorded, not sent to a server. There is no account system anywhere in the product, so there is nothing for a drawing to be attached to and nobody to attach it to.
Still stuck? Open the canvas and try it — the thing answers faster than reading about it does.
Try it FreeNo. There is no sign-up, no trial and no account system in the product at all — you open the tab and the canvas is there. Nothing you draw is tied to an identity because there are no identities.
Yes. There are two canvas modes, portrait and landscape, and portrait exists because that is how phones are held. You need a camera and a reasonably modern browser; everything else is the same as on a laptop.
The measuring takes about three seconds — two holding an open hand, one holding a pinch. Getting the pinch to feel natural takes a few minutes rather than a few seconds, and the usual mistake is squeezing hard when a light touch of thumb to index is all it reads.
Both. Erase is a mode: raise index and middle to open the tools menu, pinch ERASE, then pinch over the strokes you want gone. Undo is the first item in the session menu, which you open by raising index, middle and ring together.
Completely — no account, no trial, no paywall, no in-app purchases. Exports carry a small getairva.com corner mark, on images and video alike, and there is no paid tier that removes it.
A camera and support for WebAssembly. If either is missing the app says which one instead of failing quietly. No plugin, no extension, no download.
Full-resolution PNG for a still, and video for the whole session — MP4 where the browser can manage it, WebM otherwise, up to sixty minutes. There is also a timelapse export at 2×, 3× or 5×.
Yes. Draw the shape roughly and then hold the pinch still: it snaps to a line, circle, ellipse, arc, triangle, rectangle or polygon. If it cannot tell what you meant it leaves your stroke exactly as you drew it.
About thirty seconds from opening the page to your first line, and nothing is asked of you on the way.