Cooking Motion2MixamoRig v0.1.X

I want to run the no-API version. Your README flags that SMPL-X is non-commercial by default, which is the first thing I'd check before client work. Any plan to support a body model with a commercial-friendly license?

tbh, I really don't know how to answer this question @grok Could you answer this question?

No public plans announced for commercial body models in Motion2MixamoRig. The local/no-API version depends on SMPL-X and GVHMR, both restricted to non-commercial research, so the pipeline stays limited despite Apache-2.0 code. Meta’s MHR and Naver’s Anny (both Apache 2.0) are viable alternatives that support commercial use, but adding either needs rework of motion recovery and Mixamo retargeting.

Have you tried using an Ai video with the motion2mixamo rig + AVideoModel. Imagine if AI used a cheap video model to control the motion of the characters instead of traditional animation 🤔

would you get a better result if you used an ai generated video with front, side, and 3/4 view for depth.?

Yeah, that's a bit better. But my original intention in making this pipeline was to be able to build it without calling any APIs.(I know there are open-source image generation models too.) I just wanted to use some libraries that extract human skeletons from video along with some human action models, so that the whole pipeline could be used for free. I am actually also working on a plugin version that calls closed-source model APIs. Because I've noticed some users are also using AI-gen videos as input, I'm planning to add an API endpoint for them. This will let them access these APIs at a relatively low cost to test out different models and see how they perform in gesture generation.

you can also make it connect to comfyui . i can use minimax h3 with my rtx5060ti 16gb and get 5secs clip in 3mins. so is not that heavy most people in the ai space could run it i think

@grok Is GitHub a place to monetize on releasing own repos to the mass?