Is it possible to build an API for an automatic1111 extension to be used through Runpod serverless?
I want to use the faceswaplab extension for automatic1111 as a serverless endpoint on Runpod.
I managed to find this github repository which features reActor (a similar faceswap extension) but not faceswaplab https://github.com/ashleykleynhans/runpod-worker-a1111
Is it possible build faceswaplab's features as an API to be used through Runpod's serverless endpoints? If yes, any pointers on how I can do so?
GitHub
GitHub - ashleykleynhans/runpod-worker-a1111: RunPod Serverless Wor...
RunPod Serverless Worker for the Automatic1111 Stable Diffusion API - GitHub - ashleykleynhans/runpod-worker-a1111: RunPod Serverless Worker for the Automatic1111 Stable Diffusion API
6 Replies
I'm not sure if @.ashleyk is still around, but he might be able to give you a direction on how you can do this
@NERDDISCO Thanks for the idea. I dm-ed him asking for some pointers.
Just substitute the extension installation commands in the install script / installation instructions, I only support ReActor, not faceswaplab
faceswaplab only has 526 stars on Github, ReActor has over 1300, so ReActor is clearly more popular/better
Also ReActor is actively maintained, las commit was 2 days ago, last faceswaplab commit was 3 months ago
There is also no documentation on how to use faceswaplab with A1111
alwayson_scripts so you are pretty much on your own.@.ashleyk thank you very much!
ReActor is a great extension and it's what I used in the beginning. However after I discovered faceswaplab, I stopped using reActor altogether because faceswaplab just provides much better results.
The issue with reActor is it only takes a single photo. Faceswaplab allows you to build a face model using as many pictures as you want. The face model blends all the images you uploaded for a more accurate result.
ReActor is indeed more actively maintained and more popular by stars on Github, but from my experience, faceswaplab is simply the superior face swap extension as of right now. Hopefully ReActor adds the ability to use multiple photos instead of just 1 photo.
Faceswaplab actually does have API with typing support in its features, I just realized. https://glucauze.github.io/sd-webui-faceswaplab/features/
FaceSwap Lab
Features
FaceSwapLab is an extension for Stable Diffusion that simplifies face-swapping. Some key functions of FaceSwapLab include the ability to reuse faces via checkpoints, batch process images, sort faces based on size or gender, and support for vladmantic.
Hey bro! Could you please tell if I can run a stable diffusion model in the serverless solution and get a 10 or less seconds as response? 😃