Mika Pikazo Style

An Artstyle LoCon

V3l LoCon is out. Please check it out!

Update Note:
For the V3L version, I have enhanced the tag quality of the training set and extended the training time, resulting in improved prompt obedience. However, I have noticed that the visual effects of this version are slightly inferior to the previous one. I am currently working on resolving this issue, and I have trained 11 versions before releasing V3L. I will release a new version as soon as I find a better solution to the problem.

In conclusion, I still recommend you download the new version since I think it is more important for a model to have a strong prompt obedience.

对于 V3L 版本，我增强了训练集标签质量并延长了训练时间，导致提示服从度有所提高,然而，我注意到这个版本的画面效果略逊于之前的版本,我正在努力解决这个问题，并在发布 V3L 之前进行了 11 个版本的训练,我将在找到更好的解决方案后尽快发布新版本,

总的来说，我仍然建议您下载新版本，因为我认为对于一个模型来说，具备强大的提示服从度更为重要,

Here are some comparisons between v2a and v3l (两个版本之间的对比) :

Please ignore the broken hand... （请无视损毁的手部,,,,）

parameters

1girl, sei shounagon (fate), visor cap, headphones, headphones around neck, shorts, swimsuit, bikini, breasts, yellow eyes, one eye closed, navel, twintails, multicolored hair, jacket, blue hair, thigh strap, chewing gum, red hair, green headwear, black hair, open clothes, bubble blowing, open jacket, looking at viewer, white bikini, short shorts, solo, long hair, black shorts, thighs, bikini top only, bare shoulders, paint splatter, off shoulder, medium breasts, large breasts, skateboard, white jacket, sidelocks, long sleeves, cleavage, blush, colorful, multicolored background, splashing, <lora:mika_locon_v2a:1>

Negative prompt: verybadimagenegative_v1.3, EasyNegative

Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 5.5, Seed: 140136480, Size: 512x1024, Model hash: c870e984cd, Model: anythingv5.5, Clip skip: 2, ENSD: 31337, Hashes: {"vae": "f921fb3f29", "embed:EasyNegative": "66a7279a88", "embed:verybadimagenegative_v1.3": "d70463f870", "lora:mika_locon_v2a": "4f03e1877d", "model": "c870e984cd"}

parameters

kaguya luna, 1girl, :o, bangs, bare shoulders, black dress, blue background, blue eyes, blush, cleavage cutout, close-up, closed mouth, d-pad, d-pad hair ornament, dress, eyebrows visible through hair, finger to mouth, grey hair, hair ornament, hairclip, looking at viewer, colorful, multicolored background, multicolored eyes, multicolored hair, open mouth, paint, paint on body, paint on clothes, paint on face, paint splatter, pink eyes, pink hair, red background, short hair, short twintails, sleeveless, sleeveless dress, smile, twintails, upper body, v-shaped eyebrows, virtual youtuber, x hair ornament, yellow background, clothing cutout, solo, breasts, paint splatter on face, white hair, cleavage, character name, x, portrait, print shirt, print hair, <lora:mika_locon_v2a:1>

Negative prompt: verybadimagenegative_v1.3, EasyNegative

Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 5.5, Seed: 2946328660, Size: 512x1024, Model hash: c870e984cd, Model: anythingv5.5, Clip skip: 2, ENSD: 31337, Hashes: {"vae": "f921fb3f29", "embed:EasyNegative": "66a7279a88", "embed:verybadimagenegative_v1.3": "d70463f870", "lora:mika_locon_v2a": "4f03e1877d", "model": "c870e984cd"}

Recommeneded prompts: multicolored hair,long hair,looking at viewer,smile,teeth,jacket,grin, (Please check sample pictures for more details)
Trigger word: you don't need any trigger word for this LoCon.
Recommeneded Weight: 0.7-1 (You can adjust it by your perference.)
CFG scale: 5.5-7
Check point recommendation: Anything V5
How to use LoCon/LyCORIS? Please read instruction of a1111-sd-webui-locon.