VTuber commissions are a trash fire. Most people doing "commissions" are just reskinning assets.
When I did it, I would write these long process posts explaining every step, what I was doing, and how it was done.
Sometimes, a client would want secrecy, so you can't always do that. Either way, it's essential to let the person who's paying for the work know what's going on with it, how the process works, and what you're bringing in from the outside vs what you needed to sculpt or hard sculpt by hand.
When I used a stock Unity environment (environments and stages were my thing), I would say so, explain how I customized it, and link to the original asset.
And what's exciting there is that when you do customizations, it makes it so much cooler when you can show how you improved something or added value to a stock asset.
Same deal with outsourcing. I really enjoyed hyping up my VTuber fashion consultant, who used serious tooling to design my outfits.
This girl could make anything, but fitting and positioning, and then mapping the work onto the character was my job.
In any craft art, the art doesn't necessarily live in the sum of the parts so much as how you make the art come together.
Even if the art is made entirely by one person, entirely by hand, it's still kinda scammy to leave paying clients in the dark about how you make the magic happen.
If you're worried that what you do can be copied just by explaining it, you're not doing anything worth protecting in the first place.
But the money isn't there.
Typically, VTubers are normal people with normal jobs and limited budgets, unless you're talking about some of the bigger ones with infrastructure and studio money.
So your average commission runs $200–$500 and takes around 30 hours, or just under a week, to do by hand, with or without stock assets.
You're not necessarily saving time by using stock assets, especially if you're working in 3D.
And half the time, you end up doing cheap commissions at a loss for friends who want to pay $40 for a finished talker, VRChat avi, or game character.
There's really no sustainable path to doing this kind of art commercially, or for any reason other than the love of doing it.
At least by hand.
AI makes 2D characters a lot simpler to create. You can use ChatGPT to create model sheets that translate into Live2D models, but more importantly, it can help with the mesh, too.
Have you noticed how, in 2025, the overwhelming majority of people working in Cubism suddenly got a lot better at what they do, without a major upgrade to the software itself?
Yes, Cubism changed the mocap AI. They moved to MediaPipe from the older pipeline. But that's not even the thing I'm looking at, or what I look for there.
They're all using Claude and staying very quiet about it.
But, by the same token, it's also really obvious when people aren't.
3D is different.
At least today, 3D vtubering isn't really affected by AI in the same way. Even if you could generate the character completely, or used AI-generated assets and textures, you would still have to do most of the rigging manually, especially for the blendshapes and ARKit stuff.
Auto-rigs have gotten better, but there's still a lot of handwork involved.
The real problem with the whole thing is that you have to draw this wildly nuanced line between what is and isn't AI when you do the VTuber thing at all. And watching the mental gymnastics involved is frankly quite entertaining.
The VTubers are good kids.
Although, I do wish more of them understood what art is in a more mature way than they do.