How to become a VTuber.
Becoming a VTuber is three decisions and a weekend: what kit you need, where the avatar comes from, and where you stream. Only one of those is specific to VTubing, and it is the one this guide spends the most time on.
Nothing here assumes you can draw, and nothing here assumes a budget. The four routes to an avatar are laid out cheapest first, with what each costs in time and money. One of them is ours and it is the last one.
New to the word itself? What is a VTuber covers what the setup is and how it differs from a PNGTuber, in about two minutes.
What you actually need
Less than the equipment lists suggest. If you already stream, you are probably one webcam away.
- Required
A PC that runs your game
Tracking and rendering an avatar costs a few percent of a modern machine. If it already streams a game without dropping frames, it will do this too.
- Required
A webcam
Any webcam that produces a clear, evenly lit picture of your face. Tracking reads shape rather than detail, so resolution matters far less than lighting.
- Matters most
A microphone
The one piece worth spending on. An audience forgives a plain avatar and leaves over bad audio, and this is as true for VTubers as for anyone else on stream.
- Nice to have
One soft light
Aimed at your face, not the wall behind you. It does more for tracking accuracy than any upgrade to the camera, and it costs less.
No capture card, no green screen, no second PC. A VTuber setup asks less of a room than a face-cam setup does, which is one of the reasons people move to it.
How to make a VTuber avatar: the four routes
No best one. There is the one that matches whether you want to draw, whether you want to wait, and whether the avatar has to be unmistakably yours.
- 01
Draw and rig it yourself
Free, plus weeksWeeks to monthsWhat is good about itTotal control, no cost, and a skill you keep. If you already draw, the art is the easy half and the rigging tools are free to learn.
What it costs youRigging is a discipline, not a step. Most first models get abandoned at the point where the mouth works and nothing else does.
A drawing app with layers, and rigging software.
- 02
Buy a pre-made model
Tens to low hundredsSame dayWhat is good about itIt is already rigged and already works. The fastest honest route to a tracked avatar, and cheaper than commissioning by an order of magnitude.
What it costs youOther people are using it. Most licences allow some edits and forbid resale, so read what you are allowed to change before you get attached.
A marketplace, and the tracking software the model was built for.
- 03
Commission a custom model
Hundreds to thousandsWeeks to months, in a queueWhat is good about itExactly the character you described, rigged by someone who does it professionally. If this avatar is going to be your face for years, it is the version that holds up.
What it costs youThe price and the wait are both real, and they are usually two separate people — an artist and a rigger. Get the commercial terms in writing.
A brief, references, a budget, and patience.
- 04
Generate the character
Credits, or a one-off purchaseMinutesWhat is good about itNo drawing and no rigging: the character and its expression sheet are drawn for you, and tracking picks which face shows. Calibration is once and applies to every avatar afterwards.
What it costs youIt is 2D and expression-driven rather than a fully articulated rig, so it will not do everything a commissioned Live2D model does. And you did not draw it.
Windows, a webcam, and about five minutes.
If what you meant was making a model
"How to make a VTuber model" is a different question from the one above, and it has a longer answer. A model is a drawing cut into separate layers — hair, each eye, the mouth, the body — and then rigged: every layer wired to a control so that a number between zero and one opens a jaw or turns a head. Tracking produces those numbers; the rig is what turns them into movement.
The software for this is mostly Live2D, and the work is a discipline rather than a step. Expect several weeks to get competent, and expect the first model to be abandoned somewhere around the point where the mouth works and nothing else does. That is normal, and it is why a rigging commission costs what it costs — you are buying weeks of somebody's practice, not an afternoon.
It is worth it if you want a specific character to move in a specific way, and if the avatar is going to be your face for years. Most people asking the question do not want that yet. They want to be on stream as somebody other than themselves, which is a smaller problem.
GetGuise does not rig models, and it will not turn a drawing you own into one. It works the other way around: the characters arrive already finished, each with a sheet of expressions rather than a rig, and your face picks which one shows. That is a real trade — a rigged model does things an expression sheet cannot — and it is the trade route four above is describing.
Five steps to a first stream
- 01
Settle the character before the software
Who they are, what they wear, what their face does when they are surprised. Every route above is cheaper and faster when this is decided first, and every one of them is expensive to change afterwards.
- 02
Pick the platform before the schedule
Twitch and YouTube reward different things — one live and habitual, the other searchable and permanent. Nothing about being a VTuber changes that decision, but making it first decides whether you are producing clips or hours.
- 03
Get the avatar working before you announce anything
Whichever route you took. A debut date set before the model is in hand is the most common reason a first stream gets postponed twice.
- 04
Calibrate in your actual streaming light
Tracking tuned in daylight behaves differently under a ring light at midnight. Do it where you stream, in the light you stream in.
- 05
Capture it, then watch a VOD
Window Capture or Browser Source, cropped and placed. Then record five minutes and watch it back — size, framing and dead-eye tracking are all obvious in a VOD and invisible in a preview.
The setup page covers steps four and five with the app and the capture dialog on screen.
Three things that go wrong
Rigging before the art is final
Changing the drawing after it is rigged means redoing the rig. Lock the art, then wire it.
The avatar is too small to read
At the size most people put a VTuber in a corner, subtle expressions disappear. Bigger, or broader faces — usually both.
The capture goes blank
A Window Capture of a window that something else covered. Always-on-top on the source window, or a Browser Source instead.
Deciding a rig is more than you want to take on is a legitimate answer. Avatars that arrive finished react with your face and have no modelling step, and the PNGTuber guide covers the still-image version of the same job.
Route four, if you want it.
A character and its expressions, drawn for you and ready to install. Free to download, and about five minutes from there to on screen.