API use for lora additional networks

[twitset]

First of all thanks for the API. It looks promising.

Is it possible to use sd-webui-additional-networks script as part of the api.Txt2Image call to include LoRAs in the inference?

[lebenoa]

Sorry for the late reply. It was midnight - 1 AM at the place I live.

Anyway, I think this fork will work and I just found out that there's a field call alwayson_scripts now so I'll update the field soon

[twitset]

Thanks so much for your reply. That's awesome. Right now I am sending requests with fn_index:[] over the /api/predict endpoint, which doesn't feel right.

[lebenoa]

I managed to get it working but will you be conformable doing this?

IMO I don't really like it and I don't have an idea how to improve this. Do you have any ideas?

[twitset]

I can give it a shot. It's better than what I am doing. I am generating a text template for the request to /api/predict.

[twitset]

I guess the "go way" to get more non-generic will be to implement an API interface for the important scripts that are most used. I will check if I can manage to implement one for sd-webui-additional-networks

[twitset]

I have implemented controlnet and as soon as I figure out how to create pull requests, I can send them.

[lebenoa]

You can create a fork. Commit the changes to the fork and then create pull request.