All products

Create half-body VTube Studio avatars with AI

Prepare consistent character layers and expressions for a half-body runtime avatar, with a separate review inside VTube Studio.

Create half-body VTube Studio avatars with AI from a character illustration and a staged set of expressions. The live2d-halfbody skill prepares variants, separates layers and builds a runtime model while asking for visual approval along the way.

The documented workflow has been checked on macOS; Windows remains unverified. It needs local tools, separately obtained Cubism Core and a compatible image tool. Online layer separation uploads the artwork with your permission; local separation and layered PSD input are alternatives. The resulting moc3 loads in VTube Studio but is not an editable cmo3 Cubism Editor project. More detailed editor work requires importing and rigging the layers again. Camera-triggered hand gestures have a known limitation, so review the model and use the defined hotkeys rather than assuming universal tracking.

What it does

  • Prepare a character for a half-body avatar.
  • Check expression variants against the reference.
  • Separate hair, eye and hand layers.
  • Review the moc3 inside VTube Studio.
  • Use defined hotkeys for supported gestures.

Install and use

  1. Provide the guide and character artwork, prepare local dependencies and obtain Cubism Core under its own terms.
  2. Approve variants and choose consented online separation, local separation or a layered PSD.
  3. Build the model, open it in VTube Studio and review expressions and hotkeys.

FAQ

Source

Made by bellisji233. Original skill: live2d-halfbody, published on GitHub under the MIT licence.

View on GitHub