HunyuanCustom
Open-source custom videos with consistent subjects
6 followers
Open-source custom videos with consistent subjects
6 followers
HunyuanCustom by Tencent is an open-source AI for customized video generation. It ensures subject consistency across frames using text, image, audio, or video inputs. Code & models now available.
Flowtica Scribe
Hi everyone!
Tencent just open-sourced HunyuanCustom, a new AI tool that's all about creating videos with consistent characters. If you've ever wanted an AI to make a video of a specific person or object doing something new, this is one to check out.
Big deal here is subject consistency. You can give it an image of a character, then describe a new scene or action with text, and HunyuanCustom aims to generate a video where that same character appears, even in different settings or performing new actions. It's designed so characters don't "change faces" and objects don't drift.
It's built to handle audio and video inputs for more control, like making your character speak or editing them into existing footage. They've released the code and models, and there's a demo to try.