SD_API 2.0 pre-release version
Same version from my previous post if you have already downloaded it!
will likely update the operator slightly for the 2.0 official version, but this is very close with a few known bugs.
Controlnet outputs currently do not get sent back and saved per frame - I will be fixing this soon.
Dreamstudio is not currently working. I will need to add this back in for 2.0 or create a separate dedicated operator for it. If anyone is especially interested in this, let me know.
Comments (10)
Appreciate this great project！！When I'm running in real-time, there are too many images in my output folder. Is there a way to automatically delete the oldest sequence images in real-time?Additionally, it seems that ContralNet has not taken effect yet~
You can create a new output folder! I usually do one per TD project for my setups. If you have more info on the controlnet bug, or can send a screenshot on discord, id be happy to take a look
Hey, great project you’ve setup. Unfortunately I’m having trouble getting this working on Mac M1 (Silicon) I have done the manual install, also downloaded Automatic 1111 and I’m up to date. I have a connection, however the Web UI won’t launch. On the settings page, on status, I have had these errors: sDapi: Error occurred: 404 Client Error SDapi: Failed to make GET request to sd-models
Hey Andy. For Mac, the launch/restart/shutdown of a1111 from TD is still not functional, but the rest of the operator should work once you start a1111 with --api command adding to your COMMANDLINE_ARGS
Dreamstudio support would be nice. Is there any possibility to connect automatic1111 from colab?
Yes this works in an updated version I will post soon. It requires the --share to be added to the launch command.
Hi there, unfortunately I had to put some arguments for running stable diffusion, because I had a AMD gpu. After starting SD_API, these arguments are overwritten (I think the "Webui" is starting). Is there a possibility to start the "webui-user" batch from the SD_API? Thanks in advance.
If you click the "More Automatic1111 Options" pulse in the a1111 settings page, then you should see more options to add other arguments to the TD launch file. I think that should solve that for you!
Hello, I am trying to run webui through this plugin within exactly the same webui folder that works with SD_API 1.22, but its so strange that it keeps showing these two errors, 1. C:\Stable_Diffusion\sd-webui-aki\venv\lib\site-packages\torchaudio\backend\utils.py:67: UserWarning: No audio backend is available. warnings.warn("No audio backend is available."), 2.launch.py: error: unrecognized arguments: --api-server-stop. and i desperately want to have a solution to this, wondering if you have a thought on these.
Not sure on the audio backend. That might be something to do with an extension that you have installed possibly? And the --api-server-stop is something I added so that users can restart/kill a1111 from the TD operator. I'm guessing that you might need to update your a1111 for that to work. You can launch from the stable-diffusion-webui with --api and the operator should work as normal.