You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Could just be me, I usually can make a model in about 2min doing a 1600x1024 image.
But if I add a control net 600x800. It can take 10min or more. Each time I make a new one the time gets longer. And then once done, I have to turn the program off because if I try to make normal images again, it can take 20min or more per image.
I'm trying to make avatars using SDXL, maybe its the model itself? Can it not handle the XL? I am using the right control net, using it as an open pose face only, with openposesdxl v1 but the model doesn't seem to matter. I tried it a while ago when I was new and it took a long time then too. I can't imagine what a larger image would do.
I'm using a 4070 card, windows 10, vivaldi. I can't guess why it takes so long, its a small image. Not sure if this counts as a bug or if that's just a downfall of a model.
Describe the bug
A clear and concise description of what the bug is.
To Reproduce
Steps to reproduce the behavior:
Go to '...'
Click on '....'
Scroll down to '....'
See error
Expected behavior
A clear and concise description of what you expected to happen.
Screenshots
If applicable, add screenshots to help explain your problem.
Desktop (please complete the following information):
OS:
Browser:
Version:
Smartphone (please complete the following information):
Device:
OS:
Browser
Version
Additional context
Add any other context about the problem here.
The text was updated successfully, but these errors were encountered:
The author has moved the software to maintenance only, as they are a single developer that has had to focus on other projects as priority. The ED discord is fairly active with volunteers helping other users as well as maintaining and supporting their community plugins.
As for your issue, without a rundown of your os, system ram, gpu, gpu ram,
My guess is you are running out of faster gpu memory and the system is forced to use slower system ram or swap file on disk.
SDXL is very memory intensive and adding controlnet will increase your gpu memory needs. If it overflows into slower memory you will see an exponential increase in delay like you are experiencing now.
You should start by going in to ED settings and setting the "GPU Memory Usage" setting to LOW.
You can also improve performance a lot by un-checking "Show a Live Preview" which can speed up generation quite a bit.
i've tried both sdxl and 1.5. I can do the 1.5 pretty well but right after, its super slow and I have to shut the diffuser down and open again or everything is slow.
I can try the low, I haven't done that one yet. Everything else runs pretty well. I always have the live off, its way too slow. And mostly noise until its an image.
Could just be me, I usually can make a model in about 2min doing a 1600x1024 image.
But if I add a control net 600x800. It can take 10min or more. Each time I make a new one the time gets longer. And then once done, I have to turn the program off because if I try to make normal images again, it can take 20min or more per image.
I'm trying to make avatars using SDXL, maybe its the model itself? Can it not handle the XL? I am using the right control net, using it as an open pose face only, with openposesdxl v1 but the model doesn't seem to matter. I tried it a while ago when I was new and it took a long time then too. I can't imagine what a larger image would do.
I'm using a 4070 card, windows 10, vivaldi. I can't guess why it takes so long, its a small image. Not sure if this counts as a bug or if that's just a downfall of a model.
Describe the bug
A clear and concise description of what the bug is.
To Reproduce
Steps to reproduce the behavior:
Expected behavior
A clear and concise description of what you expected to happen.
Screenshots
If applicable, add screenshots to help explain your problem.
Desktop (please complete the following information):
Smartphone (please complete the following information):
Additional context
Add any other context about the problem here.
The text was updated successfully, but these errors were encountered: