I had problems until I switched to the Euler A sampler. It took me a bit to realize that PonyXL and most derived models strongly prefer Euler A for some reason. If you use other samplers, I find it often produces really messy images.

It took me a long time to figure this out because for some other unknown reason (the wonders of machine learning), Autismmix tends to be more tolerant of other samplers, so it might not show up as an issue until you start adding other style LoRAs to it.

Might not be your issue, but hopefully helps someone.

Pony models always has a problem with faces ((

Have you tried running the lora at lower strengths? Something like <lora:incase_style_v2_ponyxl_ilff:0.7>?

@Wahn129180 yes, it doesn't help

There is a setting called 'restore faces'. Try turning it off.

That's calling up the lora to actually do its work, and yes it does work in Automatic 1111 if you place the lora in the lora folder first. Inside the <> is the name of the lora file, and the strength it is run at (1 in this case). For example, I run the lora at 70% strength using "<lora:incase_style_v2_ponyxl_ilff:0.7>"

Incase is an artist

Who is "his style"?

It's probably because it looks good + works well across multiple styles and other loras. Do you know of any other loras with similar range that actually look good? ...I don't mean to be dick that's an actual question I'd like to know, I agree this one is way overused but that could be for lack of better options at the moment especially for hardcore scenes.

I'm pretty new to this stuff but it seems you basically need to chuck in a particular artist style lora like this to get anything good and this one tends to be quite easy to get good results with.

@onelongwinter0451859 Uuh, what? There are quite a few style loras not directly based on a single artist and as long as you're not using pure Pony, you don't need one at all.

Same here. It definitely affects my pictures, but what is the intention, exactly? What is it trying to do? No explanation here.

Apparently it facilitates underage images, which it's not nice to talk about it, if that's the case. The thing is, there isn't any explanation about what this lora does, and no matter how hard I try in the prompt, even with tags like "25 years old", the woman always looks like a 14yo girl. So, if nobody is talking about the purpose, maybe that's because one can't talk about it. IF that's really the case, I'll not condemn anyone. If you keep it in your mind, then you're not commiting any crime, but keep it clean and legal in the real world.

@anthonirox what the fuck you are talking about ? i don't even understand your reply , it's for me or you speak to who !? and i'm still have no idea what this lora do

@amazingbeauty I'm sorry. If you didn't understand what I said, I hope your parents don't let you use forks and knives.

yeah I'm sick of this grossly homogenized lip filler instagram whore aesthetic. It just screams smoothbrain on all fronts

While I think the style is nice, I agree that some variety is cool too. I've been inpainting my pony stuff with flux models and I think the results have been pretty decent.

It's a great style when blended with something else but it's so boring by itself

It's kinda you see a female character generated by a Pony checkpoint and you already know this lora was used without checking the used resources hahaha 🤣

Or maybe different have different preferences to you and you can't accept that. Grow up lol

You just told people how to use this resource to generate underage content offline.

@CabbageBatman if they wanted to do that they would not need me to tell them, there are almost no limits offline...

Dude you don't need this lora to generate an image that looks underage. You just prompt 16yo and you get it.

@anthonirox that is true but that gets filtered by the online generator or on upload. The problem is when you specify a legal age and it still turns out looking underage.

@Kaleidia Absolutely. I did download this lora just to try it out and even with "25 years old" in the prompt, the girl always looks like a 14yo. Maybe that's the purpose of this lora, since there's no explanation anywhere about what it does.

@anthonirox it is the style of an internet web comic artist who is doing a fantasy comic with sex scenes and some very nsfw side stories.

Why? I'm curios why this lora is so popular.

@anthonirox  The artist InCase has long been known for stunning artwork and their Um, porn, secondary. This Lora gets very close to that "realistic drawing" style that doesn't go uncanny valley. To use it just call it in your prompts, it just helps push the art style in a specific direction.

Incase is an adult artist, and this is a LoRA that resembles his artstyle.

@magma_dan Ok, thank you. But why one doesn't make this clear in the about session? I mean, I bet there are at least 8 billion people in the world that don't know anything about this guy. It's not like he is Picasso or something.

@anthonirox Maybe its to avoid having the model taken down by the artist. I could be wrong but since they never admit that it was trained on incase's art they could just say it's a coincidence that the model mimics their style and is named after them. It might sound silly but I think this has happened a few times before. I've seen loras trained on famous people that got taken down pretty quickly and the ones that stay up longest don't describe much of anything about who it was trained on.

@peenusbutter it's a reasonable assumption. Thank you.

Probably because your canvas dimensions are different. Pretty sure the original canvas size is half of the example image. I got similar results when I tried with those.

Only Adetailer. There are quite a few errors in this Lora.

the art style, literally everything on the picture, the thickness of the lines, the rendering of the texture of the colors, and most notably the dimensions and ratios of the face that are unique to inCase's style