konpatp/diffae
A deep learning model combining diffusion probabilistic models with autoencoders for semantic image generation and manipulation.
Not currently ranked — collecting fresh signals.
star history
This is the official implementation of a CVPR 2022 paper on Diffusion Autoencoders (DiffAE). The model uses a convolutional encoder to learn semantic latent representations paired with a diffusion-based decoder for high-quality image synthesis. It supports unconditional generation, semantic manipulation, interpolation, and autoencoding of images. The project includes Jupyter notebooks for various tasks and web demos via Replicate.
Frequently asked
- What is konpatp/diffae?
- A deep learning model combining diffusion probabilistic models with autoencoders for semantic image generation and manipulation.
- Is diffae open source?
- Yes — konpatp/diffae is open source, released under the MIT license.
- What language is diffae written in?
- konpatp/diffae is primarily written in Jupyter Notebook.
- How popular is diffae?
- konpatp/diffae has 968 stars on GitHub.
- Where can I find diffae?
- konpatp/diffae is on GitHub at https://github.com/konpatp/diffae.