[Stable Diffusion] Prompt Sharing and Learning Thread

[Mr-Fox]

My bad, was using x/y/z script, didn't know about the other one.
I don't even have x/y script there

Might it be something wrong with the checkpoints?
I just tested the other 5 checkpoints and see no such issue

Impossible for anyone to know. Go to Extensions tab and press "check for updates" then "apply and restart".
Now all extensions are updated. It's a good idea to do this with regular intervals. If SD is acting strange it could be a sign that it's time to do a full reboot. Meaning closing the cmd window and do a git pull, then update the extensions.
If it still is acting strange then it's reinstall that is next step by doing a git clone. Then git pull and updating extensions.
If it still is not resolving the issue, the next step is to post a message in issues on github and ask on Stable Diffusion Reddit thread. You can also ask the authors of the checkpoints on civitai.
I hope it helped...

 

[Mr-Fox]

My random seed (-1) is not working now, no matter what I do
It generates the image and then the second image +1 to the seed, no matter of the settings

I've pressed Unload SD checkpoint to free VRAM and renamed some checkpoints before doing it.

what's wrong with it?

View attachment 2615087 View attachment 2615088

You have set batch count 2.. If you set it to 1 it will work like normal again.

 

[Jimwalrus]

You have set batch count 2.. If you set it to 1 it will work like normal again.

That's what I said. Only I used a lot of unnecessary words!

 

[Nano999]

I thought that was how batches have always worked? Certainly that's how it's always worked for me i.e. the first image is a random seed, every subsequent image in the same batch is +1 to the seed.

Oh yeah, you are right, I become paranoid with this weird checkpoint bug

Impossible for anyone to know. Go to Extensions tab and press "check for updates" then "apply and restart".
Now all extensions are updated. It's a good idea to do this with regular intervals. If SD is acting strange it could be a sign that it's time to do a full reboot. Meaning closing the cmd window and do a git pull, then update the extensions.
If it still is acting strange then it's reinstall that is next step by doing a git clone. Then git pull and updating extensions.
If it still is not resolving the issue, the next step is to post a message in issues on github and ask on Stable Diffusion Reddit thread. You can also ask the authors of the checkpoints on civitai.
I hope it helped...

Git pull, git clone?

 

[Jimwalrus]

Oh yeah, you are right, I become paranoid with this weird checkpoint bug




Git pull, git clone?

Or just whack the words "git pull" (no quotes) at the start of your Webui.bat file and restart. That does most of the git updating stuff for you. Whether you like it or not sometimes...

 

[Nano999]

Or just whack the words "git pull" (no quotes) at the start of your Webui.bat file and restart. That does most of the git updating stuff for you. Whether you like it or not sometimes...

It was there before:

Echo off

set PYTHON=
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS=--no-half-vae
git pull
call webui.bat

 

[Mr-Fox]

It was there before:

Echo off

set PYTHON=
set GIT=
set VENV_DIR=
set COMMANDLINE_ARGS=--no-half-vae
git pull
call webui.bat

Git clone is how you install SD Auto1111.. By typing cmd in the root folder, meaning inside a folder named Stable Diffusion and in the file adress field type "cmd" then paste: git clone

You must be registered to see the links

. and run it.

 

[Nano999]

Very weird stuff is going on, I will ask the creator about this.

 

[sharlotte]

Nano999 , if you are using xyz, below the section where you enter the prompts, there is a tick box to 'keep -1 for seeds'. That could help you out maintaining random seeds on your creations.

 

[Nano999]

Btw is there a setting allowing to Unload SD checkpoint to free VRAM each time you change checkpoints?
I don't know how it works when you have like 50 checkpoints in XYZ, will it unload weights by itself once VRAM is full or no?
Because the loading time of every new checkpoint is increased and I wanted to unload it every time checkpoint changes.

 

[devilkkw]

in Settings,Stable diffusion put these value a 0:

Checkpoints to cache in RAM

VAE Checkpoints to cache in RAM

 

[Nano999]

in Settings,Stable diffusion put these value a 0:

Checkpoints to cache in RAM

VAE Checkpoints to cache in RAM

Hm, they were set to 0
Then I don't understand.

e.g. check point #1 need 15 sec to load, checkpoint #2 loading time will increase to 60 sec, checkpoint #3 160 sec.
but if I restart SD checkpoint #3 will load in 6 seconds. So why we have this load time increase? And can it be decreased?

 

[devilkkw]

this is really strange. i get difference in loading time only when checkpoint is new and sd have to calculate hash.
Otherwise all models is loaded in about 15 sec.
These issue maybe come for some extension.
I've removed for example an extension for making 360 image( don't remember correct name), because it affected all sd work: from time for loading model to max size for generation.
Check if your extension is up to date, or try to move in other folder to run without it and check time loading.
If you have same issue, problem is something other, but if issue vanish is surely some extension.

 

[Mr-Fox]

If a checkpoint is new and SD needs to calculate hash and "stuff" it's not strange if it would be a longer load time and then after a shorter load time imo. I think it's just normal and you didn't notice before.

 

[Nano999]

If a checkpoint is new and SD needs to calculate hash and "stuff" it's not strange if it would be a longer load time and then after a shorter load time imo. I think it's just normal and you didn't notice before.

How to save the hash? I mean it feels like it deletes the hash every time on exit or smth, so once I open SD again it processing checkpoints for some time, like on the pic below. 1st checkpoint is fast, all the next will be with increased processing time, 60, 120, 180 sec or smth

 

[Mr-Fox]

I have not observed this, though I haven't looked. Sorry I don't know enough to be able to help. Hopefully some one else knows more and have an idea.

 

[Nano999]

I tried to use

You must be registered to see the links

, but when on img2img tab in the script field I did not find R-ESRGAN 4x+ upscaler to check

But in the settings it should be there, it's checkmarked:

So what's the deal?

 

[Mr-Fox]

I tried to use

You must be registered to see the links

, but when on img2img tab in the script field I did not find R-ESRGAN 4x+ upscaler to check

But in the settings it should be there, it's checkmarked:

So what's the deal?

Did you restart ?

 

[Sharinel]

How to save the hash? I mean it feels like it deletes the hash every time on exit or smth, so once I open SD again it processing checkpoints for some time, like on the pic below. 1st checkpoint is fast, all the next will be with increased processing time, 60, 120, 180 sec or smth

Just in case you think you are going mad, this happens to me as well. And from what I can see the 1st checkpoint isn't fast, it is incorporated in the startup of SD so you don't really notice it.

 

[Nano999]

Did you restart ?

Yea
Also it dissapeared from txt2img, but I can load it from PNG info page to txt2img, like with NMKD example here:
(I noticed it before, but did not pay any attention as I alway use PNG info to load settings)