Generate images for webpages using SDXL lighning / turbo

[iChristGit]

The Inference speed of those model is almost instant,
Adding a small image to webpages could add to the immersion with a small cost in time to generate, can this be done? maybe as an option?

[Zetaphor]

I've added image support in #4 using SearXNG, and pwcarney said they're working on SD integration in their fork.

[Sebby37]

This is something I've been wanting to do myself, but I don't know all that much about running SD beyond setting up a1111 and using a few different 1.5 checkpoints and loras. I might give it a shot myself at some point and see how well it works

[iChristGit]

This is something I've been wanting to do myself, but I don't know all that much about running SD beyond setting up a1111 and using a few different 1.5 checkpoints and loras. I might give it a shot myself at some point and see how well it works

For it to be viable you probably want the faster models that can generate with 2-4 steps.
Good luck!

[scalar27]

The lightning models are quite amazing for that.

[Sebby37]

For it to be viable you probably want the faster models that can generate with 2-4 steps. Good luck!

Thanks! I'll take a look at the lightning modes as suggested in this thread but I've also seen LCM loras that seem to do the same sort of thing, I'll try both and see how well each work when I get the chance