[Bug]: PLMS Sampler distortion at the last step

Is there an existing issue for this?

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

What happened?

After updating SD from v1.2.1 > 1.6.1 the sampler PLMS started to malfunction. First it looks like a VAE issue, but i tried about 3 different VAEs but no one changed anything at all to solve this issue. The whole Picture breaks with distortion.

Steps to reproduce the problem

Update to v1.6.1/ Use a newer version as v1.2.1
Load your old prompt and seed you used into the txt2img from png info
Generate
Look how ugly it is

What should have happened?

Update to v1.6.1
Load your old prompt and seed you used, into the txt2img from png info
Generate
Enjoy generating

Sysinfo

sysinfo-2023-12-06-21-11.txt

What browsers do you use to access the UI ?

Mozilla Firefox

Console logs

################################################################
Install script for stable-diffusion + Web UI
Tested on Debian 11 (Bullseye)
################################################################

################################################################
Running on thefrieber user
################################################################

################################################################
Repo already cloned, using it as install directory
################################################################

################################################################
Create and activate python venv
################################################################

################################################################
Launching launch.py...
################################################################
Using TCMalloc: libtcmalloc_minimal.so.4
Python 3.10.12 (main, Nov 20 2023, 15:14:05) [GCC 11.4.0]
Version: v1.6.1
Commit hash: 4afaaf8a020c1df457bcf7250cb1c7f609699fa7
Installing requirements
If submitting an issue on github, please provide the full startup log for debugging purposes.

Initializing Dreambooth
Dreambooth revision: c2a5617c587b812b5a408143ddfb18fc49234edf
Successfully installed accelerate-0.19.0 fastapi-0.94.1
[+] xformers version 0.0.20 installed.
[+] torch version 2.0.1+cu118 installed.
[+] torchvision version 0.15.2+cu118 installed.
[+] accelerate version 0.19.0 installed.
[+] diffusers version 0.16.1 installed.
[+] transformers version 4.30.2 installed.
[+] bitsandbytes version 0.35.4 installed.
Launching Web UI with arguments: --xformers --no-half-vae
Loading weights [3eede11bf8] from /home/thefrieber/Desktop/AUTO1111/stable-diffusion-webui/models/Stable-diffusion/fakePvcStyle_11.safetensors
Creating model from config: /home/thefrieber/Desktop/AUTO1111/stable-diffusion-webui/configs/v1-inference.yaml
Couldn't find VAE named anything-v4.0.vae.pt; using None instead
Model loaded in 1.3s (load weights from disk: 0.1s, create model: 0.5s, apply weights to model: 0.5s).
Applying attention optimization: xformers... done.
/home/thefrieber/Desktop/AUTO1111/stable-diffusion-webui/extensions/sd_dreambooth_extension/scripts/main.py:257: GradioDeprecationWarning: The `style` method is deprecated. Please set these arguments in the constructor instead.
  with gr.Row().style(equal_height=False):
/home/thefrieber/Desktop/AUTO1111/stable-diffusion-webui/extensions/sd_dreambooth_extension/scripts/main.py:1047: GradioDeprecationWarning: The `style` method is deprecated. Please set these arguments in the constructor instead.
  db_gallery = gr.Gallery(
/home/thefrieber/Desktop/AUTO1111/stable-diffusion-webui/extensions/sd_dreambooth_extension/scripts/main.py:1047: GradioDeprecationWarning: The 'grid' parameter will be deprecated. Please use 'columns' in the constructor instead.
  db_gallery = gr.Gallery(
Running on local URL:  http://127.0.0.1:7860

To create a public link, set `share=True` in `launch()`.
Startup time: 24.2s (prepare environment: 14.5s, import torch: 2.2s, import gradio: 0.8s, setup paths: 1.8s, initialize shared: 0.2s, other imports: 0.4s, load scripts: 2.3s, create ui: 0.7s, gradio launch: 1.2s).
100%|███████████████████████████████████████████| 24/24 [00:01<00:00, 13.51it/s]
100%|███████████████████████████████████████████| 18/18 [00:02<00:00,  7.38it/s]
Total progress:  93%|█████████████████████████▏ | 42/45 [00:05<00:00,  7.96it/s]
100%|███████████████████████████████████████████| 24/24 [00:01<00:00, 15.63it/s]
100%|███████████████████████████████████████████| 18/18 [00:02<00:00,  7.81it/s]
Total progress:  93%|█████████████████████████▏ | 42/45 [00:04<00:00,  9.18it/s]
100%|███████████████████████████████████████████| 24/24 [00:01<00:00, 15.54it/s]
100%|███████████████████████████████████████████| 18/18 [00:02<00:00,  7.77it/s]
Total progress:  93%|█████████████████████████▏ | 42/45 [00:04<00:00,  9.19it/s]
^CInterrupted with signal 2 in <frame at 0x564c6c76fd50, file '/usr/lib/python3.10/threading.py', line 324, code wait>

Additional information

-Used with and without "Use old emphasis implementation. Can be useful to reproduce old seeds." -Used different VAEs -Generated with a rx580 8gb gpu(old, ROCm) now generating with rtx 4070 12gb edition(CUDA, still the same, new version malfunctions) -Tried to play around with clip skip 1 and 2 -Looked if Face restoration was the issue

AUTOMATIC1111 / stable-diffusion-webui