Hi Community,

IC-GAN a new AI generative picture engine from Facebook.

A new Colab appeared from Facebook

Traditional GAN models can only generate images of zebras in grasslands because their training data consists solely of those types of pictures. When you try and train a GAN using urban areas, such as New York City or Los Angeles, it will fail since there’s not enough examples from which the model can learn what features define these landscapes – like buildings or trees. The IC-GAN model can generate novel combinations of data by including objects that are not usually seen in standard datasets. For example, the research shows how it created an image of cows walking across the sand – this is just one way they choose from many possibilities. source

https://github.com/facebookresearch/ic_gan?

In short, you can use prompts and init pictures like other ClIP-GAN tools, and also specify an index of image class will drive the picture creation. The engine will alo produce set of pictures with closer similarities than other engines. In certain case, you may think that the AI draw the same elements but from different point of view which is quite new.

I feel this will be a new tool for further explorations. My first step did not show stunning results but at least it works under a free colab.

One tip, if you use init images, store them in the icgan directory not at the content root.

Let's explore this tool, it can be quite surprising.