hoelee/stable-diffusion-webui
1
0
Fork 0
mirror of https://github.com/AUTOMATIC1111/stable-diffusion-webui.git synced 2025-08-10 10:50:09 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

changes for inpainting for #35

Browse Source 
support for --medvram
attempt to support share
This commit is contained in:
AUTOMATIC
2022-09-01 11:41:42 +03:00
parent 3e4103541c
commit e1648fc1d1
2 changed files with 76 additions and 53 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
README.md
View File

@@ -71,10 +71,10 @@ Run the command to start web ui:
python stable-diffusion-webui/webui.py
```
If you have a 4GB video card, run the command with `--lowvram` argument:
If you have a 4GB video card, run the command with either `--lowvram` or `--medvram` argument:
```
python stable-diffusion-webui/webui.py --lowvram
python stable-diffusion-webui/webui.py --medvram
```
After a while, you will get a message like this:
@@ -280,17 +280,18 @@ print("Seed was: " + str(processed.seed))
display(processed.images, processed.seed, processed.info)
```
### `--lowvram`
### 4GB videocard support
Optimizations for GPUs with low VRAM. This should make it possible to generate 512x512 images on videocards with 4GB memory.
The original idea of those optimizations is by basujindal: https://github.com/basujindal/stable-diffusion. Model is separated into modules,
and only one module is kept in GPU memory; when another module needs to run, the previous is removed from GPU memory.
It should be obvious but the nature of those optimizations makes the processing run slower -- about 10 times slower
`--lowvram` is a reimplementation of optimization idea from by [basujindal](https://github.com/basujindal/stable-diffusion).
Model is separated into modules, and only one module is kept in GPU memory; when another module needs to run, the previous
is removed from GPU memory. The nature of this optimization makes the processing run slower -- about 10 times slower
compared to normal operation on my RTX 3090.
This is an independent implementation that does not require any modification to original Stable Diffusion code, and
with all code concenrated in one place rather than scattered around the program.
`--medvram` is another optimization that should reduce VRAM usage significantly by not peocessing conditional and
unconditional denoising in a same batch.
This implementation of optimization does not require any modification to original Stable Diffusion code.
### Inpainting
In img2img tab, draw a mask over a part of image, and that part will be in-painted.