How AI avatars work
Dreambooth
Hello Avatar uses Dreambooth, a technique that fine-tunes Stable Diffusion on your photos. The model learns your face and body, then generates new images of you in any scenario you describe.
Training
Training involves diffusing your images through filters to create realistic outputs. The process typically takes about 20 minutes and produces a private model unique to your studio.
Generation
Once trained, you describe any style or scenario in natural language. The model generates 4K avatars that match your appearance while following your prompt.