Make a Streaming-Ready VTuber From a Single Picture Without Rigging
This studio builds a complete VTuber model from a single character image and needs no Live2D or mesh rigging. Generate the character with AI or upload a sprite of your own. The studio then guides you through generating matching animation clips: idle, talking, intro and outro, camera-triggered expressions such as happy, sad, angry, surprised, head turns and winks, and hotkey moves such as dance, BRB and nod.
Once the clips are in, your webcam drives the model live. When you talk, the speaking clip plays. When you smile, the happy clip plays. When you turn your head, the model turns. Micro-motion head tracking keeps it alive between expressions. There is nothing to install: you paste one URL into OBS as a browser source, or open it full screen. A cue board lets you trigger animations from your phone or any other device.
It was built in collaboration with the VTuber Leaflit and is meant for new streamers who do not want to commission art and rigging, and for existing ones who want a second model fast. Every clip also exports as chroma-keyed transparent WebM or GIF, and the same saved models power the waiting scenes in Stream Kit and the bodies of Living Avatars.
Open AI VTuber Studio by Leaflit on Reality Fabricator
What you can do
- Generate the character with AI or upload your own sprite
- Guided clip generation: idle, talking, intro, outro, expressions and hotkey moves
- Webcam face tracking switches clips automatically as you talk, smile or turn
- One URL pasted into OBS as a browser source, no downloads
- A cue board triggers animations from a phone or any other device
- Chroma-keyed transparent WebM and GIF export of every clip
How it works
- Name your character and generate its image with AI, or upload a sprite.
- Generate each animation clip from a pre-written prompt, pick the take you like, and set the loop point.
- Cut out the background with chroma key and save the clip to its slot in the model.
- Add camera reactions and hotkey moves, then copy the OBS URL and go live with your webcam.
Frequently asked questions
Do I need to rig anything?
No. The model is a set of animated clips, one per state, and the studio switches between them from your webcam. There is no Live2D, no mesh and no rigging software. A classic rigged option exists separately in the Live2D Studio.
How does the face tracking work?
The live overlay reads your webcam. Speaking plays the talking clip, smiling plays happy, and head turns turn the model. Expression sensitivity is adjustable, and if there is no camera the microphone alone drives the talking state.
How do I use it in OBS?
Each saved model has a permanent overlay URL. Add it in OBS as a browser source. Hotkeys trigger moves such as dance, BRB or nod, and the cue board can fire the same animations from another device without touching the streaming machine.
Can I make more than one model?
Yes. Start Fresh clears the studio for a new character. Older models stay live at their own OBS URLs, and their clips and settings remain saved on the server, so switching characters does not break a scene you already set up.
What do I get as files?
Every clip exports as a transparent chroma-keyed WebM or GIF, and you can download the whole model as a zip pack. The same saved models can be used as waiting-scene actors in Stream Kit and as bodies for Living Avatars.