Hey, try this:
Don't use the same prompt for both positive and negative. The metadata for the negative in the sample images was probably not loaded correctly by Civitai.
Try another Illustrious checkpoint instead of the base model.
Also, this is an artist style LoRA, not a character LoRA. We don't know if any character names were included in the training.

@bertrand_krenshaw I see, the doubled prompt in pos and neg indeed is a problem and without this I get at least something you could call an actual image. But still.. I did a full copy including identical seed number cfg steps etc of your holding hands in front of face image. If this woud've been a quiz, nihei's style would be at least in the last two or three answers left

My problem is Not that I can't generate an identical copy or the perfect image. I know I am by far not the best prompter or most creative mind. But having a nice collection of like 30 grey patterns pics somehow letting me doubt my abillities a bit more ^^

@freckledvixon good to know, thanks

@freckledvixon btw did you include the namens from the Manga?

@freckledvixon okay, thanks

@Pandaofd00m I can see from your other images that your abilities are fine. Illustrious can be tricky if you haven't used it before. I'm not sure what tool you are using to make images. Experiment with it until you have a workflow that can generate a clear image before you add any LoRA. There's nothing special about the prompt for this one as long as you have 'tsutomu nihei' in there somewhere. EDIT: try the illustrious checkpoint I used in my images, I also got muddy images from the base model.

@bertrand_krenshaw Thanks, even if this shouldn't be fishing for compliments ^^

I have done those images so far with the onsite generator but will try in 2 or 3 days when I am back home with comfy (have to do some travel stuff for work atm so only via mobile)

@bertrand_krenshaw okay so this really seems to be a Checkpoint related Thing. I am still not fully satisfied with the outcome but now I am back on track and have to do some adjustments tto my likings. Thank you very much for pointing me in the right direction

@Pandaofd00m it looks as though it could potentially be the use of the wrong clip, illustrious, noob, and pony, typically work best and sometimes only work with clip skip 2 or -2 if using comfyui

@CHESHIRE_OS As I am running illustious in comfy on the workflow I did build for pony models (with just some minor tweaks) there never occured a Clip skip related issue. Crazy thing is that my problem only happened while using the onsite generator.

@Pandaofd00m i think that was brought up on the twitch stream today that the base illustrious on site is v1 and doesnt do great image generation while merges finetunes and later version should give better outputs

@CHESHIRE_OS ah, good to know. So if this was adressed already and will be resolved eventually I'll give it a try from time to time.

Thanks a lot for still looking into this