Lora not working correctly

[hadynpark]

Having strange issues with using lora:add_detail:1.4 at the end of every prompt in the list making my images fried.
Using it only for the first one just meant it didn't work after a while, found out I can basically resolve it by only using it every other prompt.
I havent tested but I presume this is an issue with any lora in a morph.

[agentd00nut]

The quick fix is to only define your lora's in the "second" prompt. so instead of

red square <some_lora:1>
blue triangle <some_lora:1>

just give it

red square
blue triange <some_lora:1>

The "every other" prompt thing is likely more related to this issue #6

The script is leveraging the composable diffusion "AND" syntax; With some quick playing around it seems like having the lora defined only in the B prompt or after the weight of the B prompt means the lora's apply correctly.

Otherwise it seems like StableDiffusionProcessing gets "confused" or just ignores the double declared loras as invalid.

A nicer fix would probably involve just taking another input box where you define your loras and the script can append that text to the end of every generated prompt. For now just try setting the lora in the second prompt and see if that works for you.