extern/easydiffusion
1
0
Fork 1
mirror of https://github.com/easydiffusion/easydiffusion.git synced 2025-02-16 18:32:25 +01:00
Code Issues Packages Projects Releases Wiki Activity

Updated How to Use (markdown)

Manuel Warum 2022-10-09 08:35:55 +02:00
parent ef00922530
commit 755abc766d
1 changed files with 10 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
How-to-Use.md

@ -19,6 +19,16 @@ You can use Face Correction or Upscaling to improve the image further.
**Another tip:** Images with the same aspect ratio of your generated image work best. E.g. 1:1 if you're generating images sized 512x512.
## Samplers
Here is a rough overview of the behaviors of different sampler implementations:
![samplers](https://user-images.githubusercontent.com/683528/194741816-1291abdd-caac-4fc2-ac97-dd88513d8b6e.jpg)
Most samplers are deterministic. If re-run with the same seed and settings, it will generate the same image. You can also increase the step size to see if artifacts in the image clear up, but it may be a slightly different image.
Note that `euler_a` and `dpm2_a` are *not always* deterministic. They can create different images when run in different batch sizes.
## Problems? Troubleshooting
Please try the common [troubleshooting](Troubleshooting.md) steps. If that doesn't fix it, please ask on the [discord server](https://discord.com/invite/u9yhsFmEkB), or [file an issue](https://github.com/cmdr2/stable-diffusion-ui/issues).