Wan2.2-Fun-Inp is the inpainting version of the video model from Alibaba’s PAI team that lets you guide how videos play out using start and end frames. You drop in the first and last image and it fills in the rest, giving you more control over the video’s flow.
You can do text-to-video or image-to-video with start and end frames. It’s set up for 512×512, 768×768, and 1024×1024. It’s built to make smooth moves between your chosen frames.
It's built on Wan2.2 and produces movie-like quality. It’s under Apache 2.0 so you can use it for business too.
Alibaba-PAI vs Alibaba.
Alibaba Cloud is the cloud side of Alibaba Group. They run stuff like virtual servers, storage, databases and AI tools. Alibaba’s PAI is one of those tools. It’s a team inside Alibaba Cloud that builds AI and machine learning systems.
They’re not the same group, just under the same umbrella. PAI handles the base AI stuff. Another group builds public chatbots like Tongyi Qianwen, but they’re now part of a different team.
So when folks say “Alibaba,” they might mean the parent company or Alibaba Cloud. “Alibaba-PAI” points to the smaller AI team inside Alibaba Cloud working on the tech behind it all.
If you'd like to access this model, you can explore the following possibilities:
Workflow
How to use ComfyUI to complete the Wan2.2 Fun Inp start-end frame video generation example
