The Lora greatly improves lighting and character aesthetics ,use example pictures of my own workflow to train.
KREA2 version
Recommended Settings:
Sampling Method:Euler
Schedule Type: simple
CFG Scale:1
Recommendation weight is between 1.5-2,or 0.5-1
If you only want lighting effects, set the weight to 0.5-1. If you want style, set it to 1.5-2.
You can try downloading both the attached epoch_ 12 and 16, I prefer 16,trained on krea2 raw fp8。
masterpiece, best quality, intricate details, soft painterly anime illustration, subtle faint sketch pencil outlines, delicate hand‑drawn linework, gentle rim backlighting, warm soft back‑glow, strong rim light from behind, beautiful lens flare, bokeh, shallow depth of field, soft atmospheric haze, muted gentle colour grading, delicate skin rendering, soft brush texture, subtle canvas grain, quiet sentimental mood, cinematic soft diffused lighting,
It's recommended to add a prompt prefix, this way it works better
sdxl version
Recommended Settings:
Sampling Method: DPMPP_2M
Schedule Type: karras
CFG Scale: 5-6
Recommendation weight is between 0.8 -0.95
Anima version
Recommended Settings:
Sampling Method: er_sde
Schedule Type: simple/beta
CFG Scale: 5-6
Prompt:
masterpiece, best quality, ultra-detailed, absurdres, 8k, (@yoneyama mai:0.7), (@wlop:0.5), (@rella:0.6), @kongya, @z3zz4 soft lineart, painterly anime style, soft brush strokes, cinematic composition, depth of field, soft focus, film grain, warm color grading, soft painterly illustration, dreamy ethereal atmosphere, soft golden hour lighting, hazy light rays, lens flare, backlighting, rim light
Description
Overfitted version, stronger lighting and character face optimization, but still has poor body performance
FAQ
Comments (1)
anima_lora v1.1比v1 稳定性差点,但美感好一些








