Spaces:

---

stabilityai

/

stable-diffusion

like 11k

Running on CPU Upgrade

Aside from running it locally you mean?

Maybe.
How do I run it locally, though? Sorry if that's too big a question!

I am also curious how to run this locally. Anybody know? @jcasoya

@raidensreturn @Adeyboy
The model is right here and provides simple example code to run
https://huggingface.co/stabilityai/stable-diffusion-2-1

I would probably recommend the newer gen models but I believe you guys dont like the plastic look. With flux, it’s very easy to solve it by using a better sampler and settings. Sd3.5 should even have too much plastic look. Main reason why people use them is that they are 100x better at prompt following, can render sentences of text even, much more crisp, and much better humans. There are lots of creative finetunes as well. There are turbo variants too that allow you to generate in as little as 1step(same speed as sd1.5) with much better quality.

If you want to still use old models, I would recommend 1.5 since that is way more popular. It has much more finetunes, Lora’s, controlnets, ip adapters and other things too. It does not have the plastic look at all and should be considerably better then 2.1.

Thanks for that detailed reply, hopefully some people will be able to follow and make use of it.

I can understand the basic idea of what you're suggesting, but I just used to click straight into that old 2.1 demo we all miss, and use it.
I guess I should try and learn, but I'm afraid it's all just a bit over my head 😊