You can see several butterfly shaped clusters emerge over the video. I think that means that some of the meaning from the training data was successfully imbued into motionnet, as SDXL was only invoked on frame 1. I wonder what will happen when trained on a bigger video corpus and when the SDXL piece of the pipeline is fleshed out!

#AI #python #torch #video #videocodec #codec #artificialintelligence #aivideo #generativeai #generativevideo #generativecodec