Andrew Ryzhkov

No, current code doesn't have such an option. Although you can merge a LoRA into a model.

Try to increase the strength parameter. If it's too tiny the model won't change the initial image much.

I'm really digging this project. It's pretty interesting how the timing and memory usage don't really change based on the precision - unlike llama.cpp where speed scales linearly with precision...

> Can Flux.1 be supported? Better switch to https://github.com/xhinker/sd_embed It isn't that flexible and convenient but supports SD3 and FLUX.1.

@bonlime Agree. I wouldn't say FLUX.1 generates worse results with sd_embed, but it definitely quite different. It should be a bug. Compel at least gives exactly the same images. But...

You would need to convert it first to diffusers. For example: https://gitlab.com/-/snippets/2505243

Yes, it is twice as small as the 16-bit model, but the quality is terrible, I must admit.

> What if we convert the model ourselves? Yes, you can. I did it for SD2.1 and it works fine. You can even convert it with dynamic axes to set...

> > Yes, you can. I did it for SD2.1 and it works fine. > > Mind telling us how to do that? Sure. Mentioned this already somewhere here. It's...

@arisha07 Yes, it's the same. You can change the resolution since SD2.1 native resolution is 768x768. It could be even higher but requires more memory. The only difference with SD1.4...