The Omnihuman model from Bytedance can take an image and animate it. It came with a technical paper, and specifically, they described their “omni-conditions” approach, which allows them to scale up the training data by mixing different “conditions”
Link to seaweed model which was used as a base model that was post-trained: https://arxiv.org/pdf/2501.083…
Listen to this episode with a 7-day free trial
Subscribe to Startup Monologues with @dianasaurbytes to listen to this post and get 7 days of free access to the full post archives.
