johannakarras/DreamPose
A research project that generates fashion videos from single images by fine-tuning Stable Diffusion on pose-conditioned inputs.
Not currently ranked — collecting fresh signals.
star history
DreamPose fine-tunes Stable Diffusion to generate coherent videos of fashion models from a single reference image and a target pose sequence. It uses DensePose to extract pose information from target frames and conditions the diffusion model on these poses during fine-tuning on the UBC Fashion Dataset. The system allows both using pretrained models for generation and fine-tuning on custom subjects.
Frequently asked
- What is johannakarras/DreamPose?
- A research project that generates fashion videos from single images by fine-tuning Stable Diffusion on pose-conditioned inputs.
- Is DreamPose open source?
- Yes — johannakarras/DreamPose is open source, released under the MIT license.
- What language is DreamPose written in?
- johannakarras/DreamPose is primarily written in Python.
- How popular is DreamPose?
- johannakarras/DreamPose has 1k stars on GitHub.
- Where can I find DreamPose?
- johannakarras/DreamPose is on GitHub at https://github.com/johannakarras/DreamPose.