Makima – Chainsaw Man Character LoRA
A character LoRA inspired by Makima from Chainsaw Man, trained for realistic and cinematic cosplay-style generations.
The dataset includes multiple adult models with varied facial features, body types and visible skin tones. It covers Makima’s signature red hair, braided hairstyles, golden eyes, formal office outfits, black dresses, themed costumes, portraits, mirror selfies, indoor scenes, beach settings and dramatic studio lighting.
Trigger word: makima
Recommended prompt:
makima, adult woman, red braided hair, blunt bangs, golden ringed eyes, white button-up shirt, black necktie, black trousers, confident expression, cinematic lighting, highly detailed
Recommended LoRA strength: 0.7–1.0
Lower strengths provide more creative freedom. Higher strengths reinforce the hairstyle, costume and overall Makima aesthetic.
Because the training set contains several different models, this LoRA is designed primarily to reproduce Makima’s character concept and visual identity, rather than the exact face of one specific person.
Base model: KREA 2
Best suited for: realistic portraits, cosplay photography, fashion images, cinematic scenes and full-body character compositions.
Description
Usage Notes
Trigger word:
makimaRecommended strength:
0.7–1.0Start at
0.8and adjust depending on the desired character accuracy.For the classic appearance, describe red braided hair, blunt bangs, golden ringed eyes, a white shirt and black necktie.
Include the desired pose, outfit, camera angle, body type, visible skin tone, lighting and background for better control.
Lower strength allows more variation in outfits and facial features.
Higher strength reinforces Makima’s hairstyle, colors and overall character design.
The dataset contains multiple adult models with varied appearances, so facial identity may change between generations.
Designed primarily for realistic cosplay, cinematic portraits, fashion photography and full-body compositions.
Complex hands, small facial details and the ring pattern in the eyes may require additional prompting or refinement.
Compatible with the KREA 2 base model.
Comments (2)
love this
this lora work so well