Version: Illustration:<ul><li>LoRA Strength: 1.20</li><li>Model: flux1-dev-fp8</li><li>Steps: 20</li><li>CFG Scale: 3.5</li><li>Sampler: DPM++ 2M SDE Heun Karras</li><li>KSampler: dpmpp_2m_sde_gpu</li><li>Schedule: karras</li><li>Guidance: 3.5</li><li>VAE: Automatic</li><li>Clip Skip: 2</li></ul>V1:This Lora doesn’t need any trigger words or keywords like illustration, digital art, or anime. It creates subjects ranging from semi-realistic to hyper-realistic.V2: This LoRA was trained on Civitai using a dataset of 21 images, over 7 epochs, with around 25 repeats.

Files with SHA256: bbab57098719e805e997fa5372587973daa013e24f8a417849c5baf1b1d5f1c6