Vimi

Vimi is a controllable AI character video generation product built on SenseTime's SenseNova large model, first unveiled at the 2024 World Artificial Intelligence Conference. By uploading a photo, users can drive the generation of a single-shot character video up to one minute using existing video, animation, voice, or text. It provides precise facial expression control, automatically generates hair, clothing, and background changes, and supports half-body motion generation, maintaining image stability without degradation over time.

Its main strengths include better controllability over expressions and motions than many existing tools, eliminating the need for repetitive random attempts. It can produce videos up to one minute with consistent high quality. The product is designed for consumer use, especially targeting female users, with features like digital avatars and emoji packs. However, it is still in the early stage, with only limited media reviews and beta sign-up; no official pricing or launch date has been published. There are also ethical concerns regarding potential misuse for creating deepfake content, and no specific safeguards have been disclosed.