This is my first attempt at making a lora. That being said, i am very surprised with the results. It came out better than i expected. it can even take your photorealistic prompts you may have generated already and turn it into anime/semi-realistic variants. Its recommended to use the trigger words: high_budget_anime_movie_styleOther words that may help for a different style is (used all together): key_visual, ultra_detailed_background, 4k, 8k, cinematic_lighting, painterly_environment, soft_cel_shading,suggested sampler: euler, er_sde, euler_cfg_pp and euler_ancestral_cfg_pp (adds a lot more color)I'm pretty sure this lora could be used with flux klein base As well. The images i generated were used with klein distilled model.This LoRA fine-tunes Flux Klein 9B toward a polished anime aesthetic, emphasizing consistent anatomy, crisp outlines, and strong facial expression control. It improves stylistic cohesion while maintaining prompt accuracy. The recommended strength is 1.00.I appreciate anyone trying out my first lora! feel free to give me feedback :)

Mrpopo's Flux.2 Klein 9B anime

Further improved dataset while also using an experimental training settings. Very noticeable improvements overall

Files with SHA256: 8e15bcaa9347e3a84e6deb0c35696443c585786a013fc6632b2bf66942aefebe