AI Art > Software > Stable Diffusion >
https://github.com/easydiffusion/easydiffusion [ 1 ]
https://stable-diffusion-ui.github.io/
A vastly easier-to-use fork of Stable Diffusion.
Fucking amazing.
Unfortunately, I can’t actually get anything decent to render. It comes down to the complexity of the prompts. It doesn’t use the same prompts as Stable Diffusion, so it’s too annoying to copy-paste and even prove that it works to make anything decent at all; fail.
–
- 2023-410-19 – Switched to ComfyUI
- 2023-09-16 – I simply cannot get anything decent to render; fail.
2023-04-12 – 2.5 ∞
-
Has CPU-only support!
- And it works incredibly well!
-
Has outpainting, but needs a plugin for it.
-
Lacks Xformers support
-
Has subdirectories, but it would be nice to have a
.gitignore-type file.
The Generate tab:
-
Is incredibly slow when there is a lot of content.
-
Is not remembered if the page is refreshed!
-
Needs a feature to move individual items to the top/bottom.
Usage and troubleshooting ∞
Best budget video cards to use for Easy Diffusion ∞
I haven’t researched anything, but it’s reasonable to assume it’s the same for it’s parent project, see Stable Diffusion § Recommended budget video cards to use for Stable Diffusion
TODO – Prompt guide ∞
- https://github.com/easydiffusion/easydiffusion/wiki/Writing-prompts [ 7 ]
- https://openart.ai/promptbook
-
Incorrectly interpreted prompt weight inside brackets #799 describes the use of
:
- You have a limited amount of complexity.
- Separate most terms with commas (
,). -
If you start using a colon (
:) to weight (emphasize) a term, as withexample:1.3then you do not surround that item with spaces, e.g.foo example:1.3 bar.
Ideas:
WARNING cpu Warning … is not a value, are you missing a space? ∞
19:33:21.359 WARNING cpu Warning: '1.0),' is not a value, are you missing a space? prompt_parser.py:60
This is because a prompt with something like foo, (bar:1.0), baz is not valid. See #799. The solution is to remove the commas and do foo bar:1.0 baz instead.
Inpainting guide ∞
- https://houseofcat.io/guides/ml/stablediffusion/inpainting
- https://onceuponanalgorithm.org/using-inpaint-in-stable-diffusion-tutorial/
-
Stable Diffusion Inpaint Tutorial : A simple guide on Inpainting in Stable Diffusion
- by Grendar
–
- Work on one area at a time.
- Work on the most simple problems first.
- Focus and repeat efforts on a single location until satisfied in that spot, then move on.
-
Where you are painting is called a “mask”.
Adjust the denoise (e.g. “latent noise”) multiple times when doing the many steps of inpainting to adjust and re-adjust a small area which you are working on:
- Lowest to erase and re-create that area.
- Medium to tweak it.
-
Large to request variation.
unknown — Resolution settings, asterisks ∞
-
TODO
Alternatives ∞
Footnotes
- was https://github.com/cmdr2/stable-diffusion-ui [ ↩ ]
- was https://github.com/cmdr2/stable-diffusion-ui/wiki/UI-Plugins [ ↩ ]
- was https://github.com/cmdr2/stable-diffusion-ui/issues/360 [ ↩ ]
- was https://github.com/cmdr2/stable-diffusion-ui/issues/1160 [ ↩ ]
- was https://github.com/cmdr2/stable-diffusion-ui/issues/1151 [ ↩ ]
- was https://github.com/cmdr2/stable-diffusion-ui/issues/511 [ ↩ ]
- was https://github.com/cmdr2/stable-diffusion-ui/wiki/Writing-prompts [ ↩ ]

killing this
I’ll likely revisit this page to dump my offline notes in.