Sign In

Comic Book Style LoRA v1

40

Download

1 variant available

SafeTensor

217.87 MB

Verified:

Type

LoRA

Stats

29

2

Reviews

Published

Jan 28, 2026

Base Model

Illustrious

Hash

AutoV2
3AA44FB47E

Comic Book Style LoRA v1 is a style-focused LoRA trained to capture the visual language of modern comic books, with a strong emphasis on line strength, controlled shading, and disciplined coloring.

Rather than chasing photorealism or over-detailing, this LoRA prioritizes:

  • Bold, readable linework

  • Clear silhouettes that survive scaling and cropping

  • Intentional shading that defines form without muddying color

  • Clean, saturated colors inspired by print-ready comic art

The training set was built from multiple full comic issues, not isolated panels, allowing the model to learn:

  • consistent character proportions across scenes

  • action framing and foreground/background hierarchy

  • quiet moments as well as dynamic splash-page compositions

This makes Comic Book Style LoRA v1 suitable for:

  • hero poses and action scenes

  • splash-page style illustrations

  • character sheets and promo art

  • calm, intimate scenes that still retain comic clarity

The goal of this LoRA is consistency and usability, not gimmicks. It layers cleanly over a wide range of base models and works especially well with controlled prompts where line clarity and color separation matter.

This is a foundation release (v1)—designed to be reliable, flexible, and easy to build on for future refinements.

Alt 1 was the first runner up while creating the loras for the same foundation.

Alt 2 was done is a 1990s retro futuristic style, which I always found appealing.

Alt 3 was done in a 70s, 80s, 90s fashion where it was more of a dotted colored printing than your standard digital printing today.

Alt 4 was done in a 1930s, 40s, 50s fashion, where it was more single color backgrounds, flat colors and strong lines.

Alt 5 was done in the style of the artist Breno over at MilfToon.


Please Note that I'm keeping them named and set activating words based on the style i chose for the version. I'm trying to be semi-transparent in that manner.