Become a sponsor to HyperGAN
HyperGAN, a configurable generative AI framework
HyperGAN is a configurable AI framework around generative adversarial networks. It comes packaged with configurations that enable training generative models around collections of images, text and even video.
Built for researchers, developers and tinkerers.
This [person] does not exist
Trained on a collection of human faces, HyperGAN creates infinite more. Trained for 700k steps on a TI 1080
GANs allow creating new data from a dataset. This is useful for all sorts of applications but using GANs is challenging and unstable. HyperGAN aims to fix this.
Alignment
Trained on unsorted day and night images. Top is the input image and bottom is the aligned image generated from HyperGAN. Step 500k trained on TI 1080
Much more
Alignment, inpainting, audio generation, text generation, text to speech, text to image, text to video. HyperGAN can potentially do all this and more, with your help.
How-tos and documentation
We have created comprehensive guides for using and building with hypergan https://hypergan.gitbook.io/hypergan and theres more to come!
Support our Goals
We want to extend and improve HyperGAN and we need your help to do it. Here's what you can do:
- Support us with funding
- Contribute to HyperGAN
- Build something using HyperGAN
- Join the community
Featured work
-
HyperGAN/HyperGAN
Composable GAN framework with api and user interface
Python 1,193