Avtrs

Avtrs is an avatar presentation tool that turns selfies into AI avatars with prompt-driven style control. Avtrs trains a personalised model from your selfies using Dreambooth and Stable Diffusion, then renders avatars from text prompts with the @me token.

Avtrs is a selfie-to-avatar generator built on Dreambooth and Stable Diffusion. Users upload a handful of photos of themselves or another person from different angles, wait while the system builds what the site calls a personal studio, and then render images by writing prompts that reference the trained subject with the @me token. The output covers a wide stylistic range: photorealistic close-ups, oil paintings, pop art, baroque portraits, fantasy character art, 3D octane renders and retro fashion looks. Each render keeps the face from the uploaded dataset while the prompt dictates wardrobe, lighting, era and art direction. The workflow is aimed at people who want a recognisable likeness inside generated art rather than a generic model face. That covers social media users producing profile pictures and shareable sets, creators who need consistent character art, and anyone experimenting with visual identities across styles. The prompt library and prompt guide lower the entry cost for users who have never written a Stable Diffusion prompt, while the underlying prompt structure still rewards people who know how to specify artists, lenses and palettes. In practice the tool is a three stage pipeline: upload photos, wait for training, then prompt for outputs. The free tier yields 25 avatars across 6 styles with watermarked images, advertisements and a wait of at least two hours. The premium set, listed at $1.99, returns 100 high quality avatars across 30 styles, starts immediately and removes ads and watermarks. A native iOS app is offered alongside the web dashboard and is positioned as faster and more feature complete, with push notifications and token prizes for sets. Among avatar generators, Avtrs sits closer to custom trained likeness tools than to template based cartoon avatar apps. The style count and the open prompt field make it more flexible than fixed style packs, but the dependence on a trained model per person means each new subject requires its own upload and training cycle. Sample prompts on the site demonstrate the range, and the FAQ and prompt guide cover the rest.