StableMaterials: Enhancing Diversity in Material Generation via Semi-Supervised Learning

We provide additional generation results for both image and text prompting. We also include the full set of material maps from the comparison with MatFuse, MatGen, Material Palette and Substance 3D Sampler, as well as, additional comparison samples.
We also evaluate our different steps in the diffusion procedure with high resolution ablation examples, demonstrating the benefits of our approach.
Finally, we include material editing samples and the text prompts used to generate the unannotated textures.


All results are included in the attached folders and images can be seen in full resolution by inspecting the original image file or opening in a new tab.