[Issue]: Prompt matrix script problem when using DPM++ SDE Sampler.

[chmatse]

Issue Description

When Using the Script Prompt matrix with the Sampler DPM++ SDE and a defined Seed like 3106912320 you get the an error Message below the picture output window:

SeedSequence expects int or sequence of ints for entropy not 3106912320.0

This does not happen with e.g. Euler aor DPM++ 2M Karras for example.

To reproduce:
+Prompt A Vase of Flowers|by Brian Stelfreeze|by Leonid Afremov|by Rob Gonsalves
Steps 50
Sampler DPM++ SDE
CFG scale 7
Seed 3106912320
Size 512x512

I have found a Bug report on Automatic1111 about the same Issue:
AUTOMATIC1111/stable-diffusion-webui#7989

And two weeks ago a user named grumpyland hast posted a solution:
AUTOMATIC1111/stable-diffusion-webui#7989 (comment)

I have changed the sampling.py file as considered and afterwards it worked for me. with DPM++ SDE.

Is there any chance you could implement this small bugfix into smapling.py? Or do you see any reason why the adjustment shouldn't be made as a local modification?

Version Platform Description

23:24:06-812126 INFO Starting SD.Next
23:24:06-812126 INFO Python 3.10.11 on Windows
23:24:07-112755 INFO Version: 1921504 Fri May 12 15:58:00 2023 -0400
23:24:07-498152 INFO Setting environment tuning
23:24:07-498152 INFO nVidia CUDA toolkit detected
23:24:09-349969 INFO Torch 2.0.0+cu118
23:24:09-365590 INFO Torch backend: nVidia CUDA 11.8 cuDNN 8700
23:24:09-365590 INFO Torch detected GPU: NVIDIA GeForce GTX 1080 Ti VRAM 11264 Arch (6, 1) Cores 28
[...]
23:24:18-930569 INFO Model created from config: D:\AI\sdvlad\configs\v1-inference.yaml
23:24:20-972211 INFO Applying scaled dot product cross attention optimization
23:24:20-997893 INFO Embeddings loaded: 0 []
23:24:20-997893 INFO Model loaded in 2.7s (load=0.2s create=0.4s apply=0.8s vae=0.6s move=0.6s)
23:24:21-176267 INFO Model load finished: {'ram': {'used': 4.88, 'total': 63.97}, 'gpu': {'used': 3.07, 'total': 11.0}, 'retries': 0, 'oom': 0}
23:24:21-546084 INFO Startup time: 12.1s (torch=3.8s gradio=1.6s libraries=0.9s codeformer=0.1s opts onchange=0.4s ui=1.5s start=0.1s scripts app_started_callback=0.1s
checkpoint=3.4s)
23:24:21-556155 INFO Memory used: 0.91 total: 63.97

[vladmandic]

that should be reported and merged in https://github.com/crowsonkb/k-diffusion
if i make changes downstream, then i'll loose ability to quickly sync-up with anything new.
i don't know why @grumpyland didn't create a pr there (i just checked, didn't see it)?

[grumpyland]

I only intended to create a quick and dirty fix for whoever was having the problem, as I had come across the same issue. I did not investigate further to see if this is the best solution, and thus did not create a pr for it there.

[vladmandic]

Owner of k-diffusers is pretty open, I suggest to create a PR amd see what she says.