Main Page
Welcome to the Diffusion Matrix wiki!
Diffusion Matrix is the project I've wanted since beginning my journey with AI art: it's a gallery of generated prompts by various models so you can compare and contrast the output of models and select the best one for your project.
If you check out models on Civit.ai or HuggingFace, you'll see sample images curated by the developers to showcase the best case output for each project... useful and beautiful! But once you start experimenting with AI art, you realize just how much work goes into each of those images... information that doesn't tell you what your starting points will look like before refinement.
Diffusion Matrix is a project to compare models and prompts in a selection matrix, where you can see the strengths and weaknesses of various models and prompts. These outputs are initial outputs without refinement; as the artist, it's your job to take the initial outputs and massage them into the final products.
With models weighing in in gigabytes, memory consumption in the tends of gigabytes, and processing time that can reach up into the many minute mark, knowing what you can expect for first drafts would be useful.
I hope you find this project useful, I'm building this for me and for my own experience, but of course I hope it can be a resource for you too!
Click on any image and see the exact prompts and sampler settings used. All images generated in ComfyUI; download the ifull-sized mages and drag them into a ComfyUI window to see the workflow and details.