@honk_honk Could you tell, what you mean with "Try it with instruct-pix2pix!"? Cause i also have no good pics with inpaint

Me too, there is some sound effect in the photo. And when I try with Pix2Pix, with or without inpainting, it shows me another image.

I had same problem: same girl, similar environment, similar clothes. After much trial and error, I realised you have to keep the lora weight VERY low, no more than 0.4, and as low as 0.1 if you want your character in a particular setting / outfit. Not sure if you need to do this to compensate for a low lora weight, but you can increase the weight of the prompts. Once you work this out, it's a great lora!

likely overcooked

I had same problem: same girl, similar environment, similar clothes. After much trial and error, I realised you have to keep the lora weight VERY low, no more than 0.4, and as low as 0.1 if you want your character in a particular setting / outfit. Not sure if you need to do this to compensate for a low lora weight, but you can increase the weight of the prompts. Once you work this out, it's a great lora!

In your prompt, enclose the keyword you want to strengthen in parentheses. Then add a colon followed by a number. 1 is full, normal strength (like, what it would be without the parens and the number). If you want to weaken the keyword, make the number a number between 0 and 1, e.g. 0.5 or 0.8. If you want to strengthen the keyword, make the number more than 1, e.g. 1.2 or 1.5. That's it. Or, if you're using Automatic11111 for your GUI, I believe you can just highlight your keyword then hold down CTRL and click the up arrow on your keyboard. It'll automatically surround the word with parentheses and add the number. Every click of the up arrow bumps the number up by .1 and every click of the down arrow drops it by .1. Hope that helped!

more cum (literally)

@honk_honk very nice thank you

Be sure to download an image associated with the Lora and rename it accordingly. Put them in specific folders in your Lora folder, eg. Liquids, Clothing, Panties, Ass Poses... Works for me.

@12foot yeah there is an auto-image downloder for A1111, but still its still too many loras. :D i got terabytes of models and loras lol

Hey man, if you happen to use ComfyUI, the Pygossss custom node pack has a Lora Loader that can show you the trigger words. Just a tip if you join us in Comfy land.

When i download the lora i copy the name, i download the first image and paste the name, i create a text file, give it the same name, then i copy paste the Trigger and words under it, the sample prompt and negative, and the whole page of txt by the creator and then save it. In windows if im looking for a certain style i can tell from its image and read the text file if it does what i want.

Just an FIY: If you guys are using SwarmUI, they recently added a new feature to "add civitAI thumbnail" by adding civitAI link to models.

Use CivitaiHelper extension for Automatic1111

Did you manage to complete this problem?

@qingmeizhujiu No.. that's why i wishing there was a way, maybe creator might know or if someone could make a plugin add on.

Funny you should mention, the creator actually left this post a while back outlining a procedure to accomplish exactly what you're describing: https://civitai.com/posts/92585