Can't read this article, it wants me to pay. Are you using Khoya_ss? I am not aware of it having an
Can't read this article, it wants me to pay. Are you using Khoya_ss? I am not aware of it having an API.
Cyou must have hit your limit. i'm using whatever is in the "RunPod Stable Diffusion v1.5+v2" template on runpod
Ci'm new to this space and don't really know what khoya_ss is
AThats not Khoya_ss @felixsanz , unless you installed it afterwards, but the template does not include it by default, it includes the Automatic1111 Web UI by default.
Cit's exposing port 7861, not sure if that's a default khoya_ss port
AIts actually better to edit
/workspace/stable-diffusion-webui/webui-user.sh and add your launch parameters in there and use /workspace/stable-diffusion-webui/webui.sh -f to start the WebUI rather than running launch.py directly, otherwise you will not get the benefit of loading libtcmalloc.so, and you will run into memory leaks without it.Cweb-user.sh or webui-user.sh cant find web-user.sh
AAnd if you want to use the API, then add
--api in /workspace/stable-diffusion-webui/web-user.sh and get the public IP like I mentioned instead of using --share. That is what I am doing to use the API, I am not using --share.Asorry my bad
webui-user.sh.A@ciphermonk which APIs are you interested in using? Standard Web UI stuff or Dreambooth?
Cjust standard webui stuff
CI've already built a model, but want to access it via api instead of webui
Ci've tried this method before but when I click the api button at the button of gradio I get this
ADid you upgrade the A1111 Web UI? They removed it in a previous commit and it was added back again.
Do this from Terminal to update it:
Then you should have the API docs working correctly again.
Do this from Terminal to update it:
Then you should have the API docs working correctly again.
AIn the previous commit you could either have a Web UI or api docs but not both, but fortunately its fixed now so you can have both again.
Ci'm using 5ab7f213bec2f816f9c5644becb32eb72c8ffb89, is this one outdated?
Ctrying now
AYes, that commit is from 2nd May, v1.3.0 is from 27 May.
Av1.3.0 is currently the exact same commit as the latest
master branch.AI prefer working off releases rather than branches or commits as much as possible.
Cthat worked!!
Cthank you so muchj
LI've been trying again for hours and can't resolve this error message, does anyone know how to fix it?
Lraise UnidentifiedImageError(msg)
PIL.UnidentifiedImageError: cannot identify image file '/var/folders/y8/f2_8t01n5x73vb4rxlkb2x8m0000gn/T/frame_0001h2a7uh9b.png'
PIL.UnidentifiedImageError: cannot identify image file '/var/folders/y8/f2_8t01n5x73vb4rxlkb2x8m0000gn/T/frame_0001h2a7uh9b.png'
LError in Multi-frame Video rendering script.
AWhat were you trying to do when you got that error? Seems like it may be a corrupt image file.
LThis error occurs after loading a guide frame in the Multiframe Video Rendering Script
after loading the guide frame and pressing the
Generate button
after loading the guide frame and pressing the
Generate button
L
LResolved. Updated stable diffusion to the latest version.
Cseems to be a running theme, we need to just using most up to date versions
C@ashleyk does txt2img api require a cpt file specified? looks like I can run the request without it
LYou're right, I missed something easy and wasted hours.
I added this to the last line of webui.bat.
git pull
I added this to the last line of webui.bat.
git pull
CI think it may be the set options api
AIt will use the default ckpt. Runpod has 2 default CKPT.
AYes thats how you set your model
Cthank you
ZHey everyone, trying to get roop to work in linux, when i run python run.py --gpu , i get File "inswapper_128.onnx" does not exist"
Zi've already installed onnxruntime and onnxruntime-gpu
GThe old stable diffusion video instructions get this error now:
"Detected that PyTorch and torchvision were compiled with different CUDA versions. PyTorch has CUDA Version=11.7 and torchvision has CUDA Version=11.8. Please reinstall the torchvision that matches your PyTorch install."
Where do we get the matching torchvision?
"Detected that PyTorch and torchvision were compiled with different CUDA versions. PyTorch has CUDA Version=11.7 and torchvision has CUDA Version=11.8. Please reinstall the torchvision that matches your PyTorch install."
Where do we get the matching torchvision?
ATorch 1.13.1 is not available for 118, but 117 works on 118. Torch 2 is available for 118 but Dreambooth still does not work with Torch version 2.
AI assume you are using the RunPod 8.0.1 image?
GBut that is exactly how I got the error. In the past I used that exact code "pip3..." and Dreambooth extension worked in SD.
GYes exactly
AI recommend using the 6.0.1 image. 8.0.1 still seems buggy, Jupyter Lab doesn't work properly with the 8.0.1 image.
GHow do we find that? What is the template called?
A8.0.1 fixed a lot of bugs with 8.0.0 but still seems buggy.
GI'm using the one called "RunPod Stable Diffusion v1.5+v2"
Aimage name:
templae id:
runpod/stable-diffusion:web-automatic-6.0.1templae id:
runpod-stableAThat should still be 6.0.1 and not 8.0.1.
AIt was 8.0.0 but they rolled it back because it was too confusing.