Wonderscapes #143

In the infinite abyss of wondering there are some random moments of beauty which will stay in memory and at the end they will remind us that exploration and efforts were not in vain.

This artwork is rendered in real-time using custom webStyleGAN model (~25mb) which was trained on a dataset of about 5000 wonderscapes generated by DALL-E from OpenAI.

In addition to the main generative model, there is a CNN classifier model that filters the rendered images by predicting similarity to landscapes (this may take some time with each generation).

On top of the image generated by the neural network, there is a processing script that converts the value of each pixel into a colored ASCII character, the final result is a 64x64 vector grid with 4096 text symbols.

Controls:

• Press [s] to save the high resolution image