Kunlun Wanwei officially released the SkyReels-A3 model, bringing the magic of making digital humans “speak.”

Kunlun Wanwei released the SkyReels-A3 model, an audio-driven portrait video generation model based on the “DiT video diffusion model + frame interpolation model + reinforcement learning for motion optimization + controllable camera movement.” Users only need to upload a portrait image and audio to generate naturally synchronized videos. The model supports features such as bringing photos to life, creating new videos, and editing dialogue.