Krea2 Style GPT Image-2 LoRA
A Krea2 LoRA trained on 450 images generated using GPT Image 2.
Designed to bring recurring visual and rendering characteristics of GPT Image 2 into your base model, while keeping your prompt's subject under your control.
Recommended Strength
0.8–1.0
Generation Settings
Samplers used: euler, euler_a, er_sde
Scheduler: Simple
Training
Images: 450
Network Rank: 64
Training Steps: 5,500