If you like my model or AI Comic Art in general please consider joining our Discord Community AI Comic Corner
Here are some Useful Articles:
Prompting Guide - Faces Guide - Mod Catalogue
I highly recommend using my embeddings with this model!
ComicMix:
This model focuses on a western comic book art style such as Marvel Comics and DC Comics with influences from Jim Lee and Todd Mcfarlane. It has a Bias towards Women and sexy attire. Prompting for Men can sometimes be a bit tricky.
Here are my go to settings for generating Images with v3.25:
Sampling: euler a
Steps: 15 - 25
Dimensions: 512x640
CFG: 7 and below
Addetailer for faces
Upscaler: 4x-UltraSharp
Prompt: Comic artwork of
Negative: Anime, Manga, 3D, Render, Painting, Photo, RAW, Cartoon, Animated, muted colors, (monochrome:1.1), (greyscale:1.1), (black and white:1.1)
Embeddings: Beyondnegative, Antiblur, verybadimage, Mr_Monster_Style
Detail Slider Lora and Colorize Slider Lora
Reviews and feedback are always welcome!
Description
FAQ
Comments (19)
This excellent comic checkpoint merge is highly flexible and reliable. It delivers a consistent style and works with a wide variety of prompts. It is now my go-to checkpoint for DC and Marvel style images. There is not much to complain about with it. It produces clear images with strong lines, good color, and sharp contrast, which can be prompted to be as bright or dark as you wish.
I did discover one minor quirk. If you are trying to create images of superheros, I would avoid using the actual prompt "superhero". It tends to key on the word "super" and produces the superman/supergirl logo on many costumes. Instead use "male hero" or "female hero". Great work. Five stars.
your feedback is very much appreciated. I am gonna take a look into the "Superman Problem" and see if I can fix it with the next version. Thank you for the kind words, that means a lot to me.
why do you recommend euler a over other samplers?
I always use it. Its fast and gives good outputs. No real reason otherwise. It's just so people know my settings if they want to reproduce my results ^.^
I'm pretty new to this so sorry for what might be a stupid question, but how do I stop it from adding multiple people in the picture? I'm getting a lot of two headed characters for some reason. Otherwise I love this style thank you :)
Hi ^.^ put "solo" or "alone" in your prompts. But to be honest I never had that problem. If you could upload some of your images for me so I can take a look ;-)
This is the sort of image I mean, it doesn't always happen but it's fairly common. Just wondering if there's something obvious I'm missing, thanks!
@ladysabrena182 Link gives me a 404. Without seeing your output and generation settings it is hard to point at anything ^.^ could be 1000 things.
I had that problem a lot when I was trying ComfyUI. Didn't figure it out.
@dudemansir for 1.5 models it is better to generate in 512x512 and upscale. Generating in 1024x1024 causes many problems like multiple heads/people, deformations, bad proportions, etc. That's what Ladysabrenas problem seem to have been. Stable Diffusion and XL are pretty different in many cases so what works for XL doesn't automatically work for 1.5
Well done! Thank you 👍
Thanks for the commpliment! Means a lot to me ^.^
The best impant model. Perfectly corrects feet and fingers. Corrects anatomy correctly. My new favorite.
Very clean yet artistic style. i love it! ⭐⭐⭐⭐⭐
Thanks for the compliment ^.^ means a lot to me
Is there an inpaint model?
No. But as far as my experience goes Inpainting works fine with this model.
An incredibly cool model! I'm in love with this amazing graphic comic style. According to my observations, this is the best thing I have seen on CivitAI in such an art style! Also, excellent responsiveness, gorgeous variability, lack of focus only on female characters, a very expressive picture...
Divine work! Bravo!
Thanks a lot for the compliment. Means a lot to me ^.^
Dude how many images did u upload for the review? You are crazy! <3