cjwbw / diffae

Image Manipulatinon with Diffusion Autoencoders

  • Public
  • 17.1K runs
  • T4
  • GitHub
  • Paper
  • License

Input

image
*file

Input image for face manipulation. Image will be aligned and cropped, output aligned and manipulated images.

string

Choose manipulation direction.

Default: "Bangs"

number
(minimum: -0.5, maximum: 0.5)

When set too strong it would result in artifact as it could dominate the original image information.

Default: 0.3

integer

Number of step for generation.

Default: 100

integer

An enumeration.

Default: 200

Output

image
image
Generated in

Run time and cost

This model costs approximately $0.0069 to run on Replicate, or 144 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia T4 GPU hardware. Predictions typically complete within 31 seconds.

Readme

This is a cog implementation of face manipulation from https://github.com/phizaz/diffae

Official implementation of Diffusion Autoencoders

A CVPR 2022 (ORAL) paper (paper, site, 5-min video):

@inproceedings{preechakul2021diffusion,
      title={Diffusion Autoencoders: Toward a Meaningful and Decodable Representation}, 
      author={Preechakul, Konpat and Chatthee, Nattanat and Wizadwongsa, Suttisak and Suwajanakorn, Supasorn},
      booktitle={IEEE Conference on Computer Vision and Pattern Recognition (CVPR)}, 
      year={2022},
}
Original in imgs directory
Aligned with align.py
Using manipulate.ipynb