Helen Parr (Elastigirl) character LoRA trained on Krea2.

Trigger word: helen_parr

Trained on the Krea2 RAW model (fp8) using Fizgig with a 31-image dataset. Works with the Krea2 Turbo model (kromaKrea2_opd) for inference.

Recommended settings:

• Supersuit scenes: strength 0.8-1.0, include "elastigirl, red supersuit, black mask, no cape" in prompt

• - Civilian/casual: strength 0.6-0.7

• - Photorealistic style: strength 0.5-0.6, add "photorealistic, real woman, cinematic photography"

• - 3D animated style: strength 0.7-1.0, add "3d render, pixar style"

Workflow: UNETLoader > LoraLoaderModelOnly > CLIPLoader (qwen3vl_4b_bf16, type: krea2) > VAELoader (qwen_image_vae) > CLIPTextEncode > EmptySD3LatentImage (896x1152) > KSampler (euler, simple, 8 steps, cfg 1.0) > VAEDecode > SaveImage