GenerationQ started as a weekend project soon after the Stable Diffusion release: I was inspired by the spirit of good will and the flurry of activity and wanted to contribute something from my particular skillset.
The problem I wanted to solve had 3 parts:
1) For people without a fast GPU (like myself) running SD takes several minutes for a single small image: to alleviate this annoyance, I wanted to be able to queue up a bunch of image generation tasks and have something process that queue automatically.
2) I wanted a pleasing to look at, enjoyable to interact with UI for creating / managing these generation tasks, reviewing results, and maintaining a "gallery". I also wanted a way to make quick doodles in-app for img2img.
3) There are a wide variety of slightly different scripts / CLI args for running these image generation models, even just for Stable Diffusion. So I wanted a UI design that could be stable across these variations and near-future developments. I think it'll be useful even for those actively testing new scripts/models.
NOTE: this is not yet fully functional for actual image generation: I did the parts that require specialized knowledge around UI/UX/JS-frontend-stuff/Electron since I imagine there are a lot of people with complementary skills who don't want to mess with that stuff but would like to hack on something with a nice UI already in place.
I have also never really done anything with open source, so this is partly an experiment for me to see what that's like... happy to receive any suggestions/advising from the community!
1 comment
[ 2.6 ms ] story [ 11.6 ms ] threadThe problem I wanted to solve had 3 parts:
1) For people without a fast GPU (like myself) running SD takes several minutes for a single small image: to alleviate this annoyance, I wanted to be able to queue up a bunch of image generation tasks and have something process that queue automatically.
2) I wanted a pleasing to look at, enjoyable to interact with UI for creating / managing these generation tasks, reviewing results, and maintaining a "gallery". I also wanted a way to make quick doodles in-app for img2img.
3) There are a wide variety of slightly different scripts / CLI args for running these image generation models, even just for Stable Diffusion. So I wanted a UI design that could be stable across these variations and near-future developments. I think it'll be useful even for those actively testing new scripts/models.
NOTE: this is not yet fully functional for actual image generation: I did the parts that require specialized knowledge around UI/UX/JS-frontend-stuff/Electron since I imagine there are a lot of people with complementary skills who don't want to mess with that stuff but would like to hack on something with a nice UI already in place.
I have also never really done anything with open source, so this is partly an experiment for me to see what that's like... happy to receive any suggestions/advising from the community!