Changelog

🔥 Anisora V3 Released Under Apache 2.0

Anisora V3 is now available under the permissive Apache 2.0 license, bringing powerful new capabilities to video generation.

• Supports arbitrary-frame inference for flexible video generation
• Enables character 3D video generation with natural motion
• Video style transfer capabilities for artistic transformations
• Multimodal guidance for enhanced control
• Ultra-Low-Resolution Video Super-Resolution
• Generates 5-second 360p video clips in under 8 seconds

🔥 AniMe Research Paper Submitted

We have submitted our work on agent-related research AniMe to arXiv. Stay tuned for further updates!

Anisora V2 Released

Anisora V2 weights are now licensed under Apache 2.0 and publicly available for download on ModelScope and Hugging Face.

Anisora V1 Bug Fix

Fixed Anisora V1 inference bug that was causing video artifacts.

🔥 Anisora V3 Preview Update

Anisora V3 Preview is updated, with new progress to be shared at Siggraph on July 11th.

🔥 Fully Open Source

All our work is now open source. Check it out!

Paper Accepted by IJCAI25

Our paper has been accepted by IJCAI25. Camera ready version has been updated.

Project Launch

Submitted our paper to arXiv and released the project with evaluation benchmark.