lllyasviel / lllyasviel/stable-diffusion-webui-forge

### Is there an existing issue for this?

Open
#3,149 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
13k
Forks
1.7k
PR merge metrics
No merged PRs in 30d

Description

### Is there an existing issue for this?

- [X] I have searched the existing issues and checked the recent builds/commits

### What happened?

I have confirmed on my end that a seed saved to a file that has a random roll is wrong.
Updated photos as accidently put a nsfw one up second image that was a totally different girl, hair colour etc.
Initially I couldn't regenerate a friends images and dug deeper and found that the seed saved to the image from a random roll are wrong.

Irrelevant test as sample was off
I generated this image a few hours ago so no updates in the meantime since.
![03877-615058593-full body photo of one 29 year old woman, cat ears, pale skin, homeless in a fantasy land, upper body, dark hair, detailed skin,](https://user-images.githubusercontent.com/597280/218583581-6ea09783-45bf-49f0-b8f2-7ae15996dbd3.png)
![Screenshot 2023-02-14 055434](https://user-images.githubusercontent.com/597280/218583654-95317b71-f73c-49b3-a0cc-6f4eec6afabd.jpg)

Copied the data meta using Diffusion Toolkit and tried to regenerate the image right now as I type this and got this one below instead.
Any attempt using the settings in the above image will just render this bottom one.
So it appears the seed that saved from a random roll is bugged.

![03891-615058593-full body photo of one 29 year old woman, cat ears, pale skin, homeless in a fantasy land, upper body, dark hair, detailed skin,](https://user-images.githubusercontent.com/597280/218583498-1251f38a-c032-4607-b18b-7e72a0331e6b.png)
![Screenshot 2023-02-14 060820](https://user-images.githubusercontent.com/597280/218585720-07acba35-562a-437b-af0b-09945936aa45.jpg)

I have checked in Diffusion Toolkit and rolled a image from one saved with a static seed its fine.

### Steps to reproduce the problem

Render image with a ransom seed
Try to recreate same image with the exact same settings prompts and seed

### What should have happened?

Same image should have been created but a different one is.

### Commit where the problem happens

commit: 3715ece0

### What platforms do you use to access the UI ?

Windows

### What browsers do you use to access the UI ?

Google Chrome

### Command Line Arguments

```Shell
xformers has no difference
@echo off

set PYTHON=
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS=--reinstall-torch --xformers --reinstall-xformers
set PYTORCH_CUDA_ALLOC_CONF=garbage_collection_threshold:0.6,max_split_size_mb:128
git pull

call webui.bat

Other bat I use
@echo off

set PYTHON=
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS=--reinstall-torch
set PYTORCH_CUDA_ALLOC_CONF=garbage_collection_threshold:0.6,max_split_size_mb:128
git pull

call webui.bat
```

### List of extensions

No

### Console logs

```Shell
Not relevant as its image generation
```

### Additional information

Tried between me and another person I couldn't render his images myself from random seed roll that was rendered in the file but his static one worked so I dug deeper and found this bug.

_Originally posted by @Revenger in https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/7808_

Contributor guide

No contributing guide indexed for this repository

Research direction

Start by reproducing the issue on commit 3715ece0 using a random seed and then a static seed, comparing the saved image metadata with the rendered result. The fix is complete when the exact settings, prompts, and saved random seed regenerate the same image.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.