🔧🔗https://github.com/FoundationVision/OmniTokenizer OmniTokenizer: one model and one weight for image-video joint tokenization.
