Prototype Comic/Semi-realism Hybrid Model
This prototype is designed to push the boundaries of the my own learning of model merging with a comic inspired aesthetic while blending elements of semi-realism. The goal is to create a dynamic, ever-evolving generation output that balances the detailed anime sketch-like nature of flux with and expressive features typical of comic art. While I was experimenting with a unique LORA that is still in development ("coming soon™" 😜), I’ve paired it with some hand-crafted comic book-style LORAs and a fashion LORA I created to explore new creative possibilities. Did I achieve something truly unique? I’ll leave that for you to decide — feel free to let me know your thoughts! 😄
Model Merge Breakdown:
Custom Style LORAs: A combination of my personal style LORAs, tailored to create distinctive, artistic outputs. Each one adds its own flavor to the model's overall aesthetic.
Comijc by Reijlita: This LORA adds a comic book-inspired touch, imbuing the artwork with bold outlines, dramatic shading, and the dynamic storytelling feel characteristic of graphic novels.
Hassaku XL (Illustrious) by Ikena: A powerful checkpoint by Ikena that injects a touch of high-quality semi-realism, resulting in stunning, refined details and a more polished, mature finish to the images.
A huge thank you to all the amazing creators who made this checkpoint possible — without their incredible work, this wouldn't have come together! 🙏
Description
FAQ
Comments (11)
EP2 is such a great model. Been prompting on ComfyUI with an All-In-One TXT2IMG and the results after an ADetailer pass are wonderful. Makes wonderful faces and great anatomy. Next, will try to make some multi-char images.
Thank you so much :D
please do i have only one it once, would love to see if face verity hasn't been comprised i really worked hard on that when i make EP1
Lol i don't want to use your model anymore, everything i generate is so well crafted and so perfect, the colors, to the composition, to the details, etc... i want to upscale and post each and everyone of them, now! It is that good, i hate it 😛
good because I am working on a possible episode 3 :D :P
P.S. Thank you :D
@purplelady Nooooooooo!
I didn't even try the episode 2 yet, please don't release episode 3 yet, i don't want no spoilers 😅
where the femboys at?!
If you are not using this model, you don't know what you are missing. Opens up whole new possibilities, by getting away from the dominant "anime face" and getting closer to a graphic novel feel. Use this model from start to finish, or use it as a finisher, you will see incredible results. You could start with something that gives wild anime style (e.g. like sometyome or yoneyama mai) and then use Plasmaviolet to highres fix to pull the character back to a western comic look, while preserving anime design in the background (if you do this, resize your first generation down by 66% with an image editing program so you get two shots to highres fix using Plasmaviolet). It also gives great results just on its own. Use this model so it can be self-sustaining and stay on the online generator!!!
Thank you so much :D
A little more than a decade and Episode 3 still hasn't been released. You gotta stop delaying it, man.
:D its only been a few decades, perfection takes time... also episode 3 might actually be zimage based if i can figure out training for it