Stable Diffusion just generates images so they are doing something on top to convert to SVG. Hard to tell how well the svg-ification is of the resulting image.
Is this doing something more than just generating bitmap images with stable diffusion and then running some vectorizer on top? Maybe just adding "vector art" to the prompts or is there more going on?
tell us a little more about the tech please OP? why did you decide to work on this? what needed to be done to wrangle SD to produce vectors? what were some surprising findings?
Did you just extract the foreground, posterize, then turn each layer into a vector path?
If so, the SVGification process is interesting, but this has basically nothing to do with Stable Diffusion; you could just as easily have done this with photos.
9 comments
[ 3.3 ms ] story [ 33.0 ms ] threadIs this doing something more than just generating bitmap images with stable diffusion and then running some vectorizer on top? Maybe just adding "vector art" to the prompts or is there more going on?
If so, the SVGification process is interesting, but this has basically nothing to do with Stable Diffusion; you could just as easily have done this with photos.