This is an ethically trained AI model, with CC content, specifically designed for generating abstract art. Take 3 days on a single RTX 3090. Training from scratch, or almost, the tokenizer end up verry inconsistent the prompt don't actually change much of the model. It should be able to output black and white icons but doesn't works But for a prove of concept I hope you enjoyWhere used 569 images:<a target="_blank" rel="ugc" href="https://www.pexels.com/pt-br/@anniroenkae/">https://www.pexels.com/pt-br/@anniroenkae/</a> 30 images<a target="_blank" rel="ugc" href="https://opengameart.org/content/700-rpg-icons">https://opengameart.org/content/700-rpg-icons</a> 533 imagesregularized 6 geometric shapes imagesAll images generate with this model can be use by comercial and research, I am working on expanding this model to more concepts. Currently we have a datatset of around half million CC and CC0 images, we appreciate anyone how want to contribute whit images or help the datatset classification.This model was trained and developed tool place in Brazil, and comply with national copyright legislation <a target="_blank" rel="ugc" href="https://www.planalto.gov.br/ccivil_03/leis/l9610.htm">https://www.planalto.gov.br/ccivil_03/leis/l9610.htm</a> it might differ if from you country.

Files with SHA256: b469c879881a275e6843128a85ecfa76b52109a3ebfbe4fc5d6a6445d42d545e