UwazumiMix 是一款由 Pony 团队开发的日系美女大模型,旨在通过先进的人工智能技术,生成高质量的日系风格人物图像,具有精致的细节和独特的艺术风格。
关于v2.1版本
我添加了稍微不同的香料,煮至透明,将上清液制成清汤式汤。
嗯,味道怎么样?
这是另一个基于 ver2.0 的模型,基于 pony,也与 sdxl 混合。
我认为基本的pony提示很好,但其中一些有点慢。
我还没怎么用过,但感觉畸形率和手部损伤率略有增加。
然而,它已经变得更加真实了。
我混合了它的假设,它会以日本外观出现,但这次它也会以深色皮肤出现。
我希望面孔更加多样化……
这是一个怪癖,所以这取决于你。
与往常一样,材料未知。
请大家单独欣赏。
PS.这真是一匹锯齿马
它是像鳗鱼酱一样将各种模型拼接在一起创造出来的。
我什至不知道来源是什么了。
请欣赏它供个人使用。
我认为最好使用基于pony的质量提示。
还需要照片级关键字。
我认为在小马(pony)风格的检查点(checkpoint)的反向提示词(negative prompt)中,“大猩猩(gorilla)” 是必不可少的。
作品参照
maximum quality, score_9, score_8_up, score_7_up,
selective focus, photorealism, beautiful details, clarity,
fresh human, livery pin-up, polaroid style, Nikon lens, five-fingers,
BREAK
The rawness of ordinal Japanese girl's life scene, black eyes,
colorful flowers, moist,
Negative prompt: score_6, score_5, score_4, low quality, wrong limbs, simple background, five-fingers, cartoonish, unreal, reverse hand, pov,
ChiBorg, makeup, thick eyebrows, blush, gorilla, dry skin
Steps: 20, CFG scale: 7, Sampler: Euler a, Seed: 2489980785, VAE: sdxl_vae.safetensors, Size: 832x1216, Model: SdxlBestP5_mix, Version: f1.1.5-v1.10.1RC-latest-864-g943a2f48, Model hash: ac6ad9ec92, Cutoff strong: False, Cutoff weight: 1.8, Schedule type: Automatic, Cutoff enabled: True, Cutoff padding: _</w>, Cutoff targets: white, mswmsa_enabled: False, raunet_enabled: False, ADetailer model: mediapipe_face_full, ADetailer version: 24.11.1, ADetailer mask blur: 4, ADetailer confidence: 0.3, Cutoff interpolation: lerp, mswmsa_simple_enabled: False, raunet_simple_enabled: False, ADetailer dilate erode: 4, Cutoff disable_for_neg: True, ADetailer inpaint padding: 32, ADetailer denoising strength: 0.2, ADetailer inpaint only masked: True, Clip skip: 2
maximum quality, score_9, score_8_up, score_7_up,
selective focus, photorealism, beautiful details, clarity,
fresh human, livery pin-up, polaroid style, Nikon lens, five-fingers,
BREAK
The rawness of ordinal Japanese girl's life scene, black eyes,
bra, pantie, posing, necklace, moist, armpit, turnback, clock, poster
Negative prompt: score_6, score_5, score_4, low quality, wrong limbs, simple background, five-fingers, cartoonish, unreal, reverse hand, pov,
ChiBorg, makeup, thick eyebrows, blush, gorilla, dry skin
Steps: 20, CFG scale: 7, Sampler: Euler a, Seed: 2489980784, VAE: sdxl_vae.safetensors, Size: 832x1216, Model: SdxlBestP5_mix, Version: f1.1.5-v1.10.1RC-latest-864-g943a2f48, Model hash: ac6ad9ec92, Cutoff strong: False, Cutoff weight: 1.8, Hires upscale: 1.5, Schedule type: Automatic, Cutoff enabled: True, Cutoff padding: _</w>, Cutoff targets: white, Hires upscaler: R-ESRGAN 4x+, mswmsa_enabled: False, raunet_enabled: False, ADetailer model: mediapipe_face_full, ADetailer version: 24.11.1, Denoising strength: 0.4, ADetailer mask blur: 4, ADetailer confidence: 0.3, Cutoff interpolation: lerp, mswmsa_simple_enabled: False, raunet_simple_enabled: False, ADetailer dilate erode: 4, Cutoff disable_for_neg: True, ADetailer inpaint padding: 32, ADetailer denoising strength: 0.2, ADetailer inpaint only masked: True, Clip skip: 2
maximum quality, score_9, score_8_up, score_7_up,
selective focus, photorealism, beautiful details, clarity,
fresh human, livery pin-up, polaroid style, Nikon lens, five-fingers, (cg, realistic),
BREAK
The rawness of ordinal Japanese girl's life scene, black eyes,
long face, big breasts, (real skin),
BREAK
film grain, side lighting, waist shot, medium-long shot,
BREAK
<lora:add_brightness_XL:0.5>
<lora:add_contrast_XL:-0.4>
<lora:uzzlanggirl-ponyxl:0.4> uzzlanggirl, korean
detailxl
Negative prompt: score_6, score_5, score_4, low quality, wrong limbs, simple background, cartoonish, unreal, reverse hand, pov,
ChiBorg, gorilla, dry skin, back lighting,
Steps: 20, CFG scale: 7, Sampler: Euler a, Seed: 2850464425, TI: [object Object], Size: 896x1152, Model: uwazumimix_v21Consomme, Version: f2.0.1v1.10.1-previous-597-gecd4d28e, Module 1: sdxlVAE_sdxlVAE, Model hash: 95cecb6732, Schedule type: Exponential, ADetailer model: mediapipe_face_mesh, ADetailer version: 24.11.1, ADetailer mask blur: 4, ADetailer confidence: 0.3, ADetailer dilate erode: 4, ADetailer inpaint padding: 32, ADetailer denoising strength: 0.4, ADetailer inpaint only masked: True, Clip skip: 2
UwazumiMix 是一款由 Pony 团队开发的日系美女大模型,旨在通过先进的人工智能技术,生成高质量的日系风格人物图像,具有精致的细节和独特的艺术风格。
它是像鳗鱼酱一样将各种模型拼接在一起创造出来的。
我什至不知道来源是什么了。
请欣赏它供个人使用。
我认为最好使用基于pony的质量提示。
还需要照片级关键字。
我认为在小马(pony)风格的检查点(checkpoint)的反向提示词(negative prompt)中,“大猩猩(gorilla)” 是必不可少的。
关于2.0版本
再次将我随意放入的材料煮沸,我将汤中的液体舀出。我觉得结果很好吃。
对于因使用该模型而引起的任何问题,我们不承担任何责任。
请欣赏它供个人使用。
::::::::::::::::::::::::::::::::::::::::::::::::::::: ::::::::::::::::::::::::::::::::::::::::::::::
皮肤有点黑很难看到的是。
我担心如果增加现实提示的数量,我的下巴会裂开。
如果使用Adtailer,去噪强度应该在0.2左右。
这也很像西方人的脸,下巴裂开。
感谢您提供的精彩材料。
作品参照
maximum quality, score_9, score_8_up, score_7_up,
selective focus, photorealism, beautiful details, clarity,
fresh human, livery pin-up, polaroid style, Nikon lens, five-fingers,
BREAK
The rawness of ordinal Japanese girl's life scene, black eyes,
colorful flowers, moist,
Negative prompt: score_6, score_5, score_4, low quality, wrong limbs, simple background, five-fingers, cartoonish, unreal, reverse hand, pov,
ChiBorg, makeup, thick eyebrows, blush, gorilla, dry skin
Steps: 20, CFG scale: 7, Sampler: Euler a, Seed: 2489980782, Size: 832x1216, Model: UwazumiMix2.0A, Version: f2.0.1v1.10.1-previous-634-g37301b22, Module 1: sdxl_vae, Model hash: c9c00fff41, Schedule type: Automatic, ADetailer model: mediapipe_face_full, ADetailer version: 24.9.0, ADetailer mask blur: 4, ADetailer confidence: 0.3, ADetailer dilate erode: 4, ADetailer inpaint padding: 32, ADetailer denoising strength: 0.2, ADetailer inpaint only masked: True, Clip skip: 2
maximum quality, score_9, score_8_up, score_7_up,
selective focus, photorealism, beautiful details, clarity,
fresh human, livery pin-up, polaroid style, Nikon lens, five-fingers,
BREAK
The rawness of ordinal Japanese girl's life scene, black eyes,
bra, pantie, posing, necklace, moist, armpit, turnback, clock, poster
Negative prompt: score_6, score_5, score_4, low quality, wrong limbs, simple background, five-fingers, cartoonish, unreal, reverse hand, pov,
ChiBorg, makeup, thick eyebrows, blush, gorilla, dry skin
Steps: 20, CFG scale: 7, Sampler: Euler a, Seed: 2489980780, Size: 832x1216, Model: UwazumiMix2.0A, Version: f2.0.1v1.10.1-previous-634-g37301b22, Module 1: sdxl_vae, Model hash: c9c00fff41, Schedule type: Automatic, ADetailer model: mediapipe_face_full, ADetailer version: 24.9.0, ADetailer mask blur: 4, ADetailer confidence: 0.3, ADetailer dilate erode: 4, ADetailer inpaint padding: 32, ADetailer denoising strength: 0.2, ADetailer inpaint only masked: True, Clip skip: 2
maximum quality, score_9, score_8_up, score_7_up,
photorealism, beautiful details, clarity, indoors,
BREAK
Japanese woman, cute face, pores, goosebumps,
brown eyes,
(round face:1.2), bob cut,
pout,
curled fingers, paw pose, leaning forward,
curvy, groin, white one-piece swimsuit,
[shadow::0.5],cinematic lighting,
Negative prompt: score_6, score_5, score_4, low quality, wrong limbs, simple background, five-fingers, cartoonish, unreal, reverse hand, pov,
blush, gorilla, muscular, smooth skin, female child,
Steps: 50, CFG scale: 7, Sampler: DPM++ 2M SDE Heun, Seed: 2573174241, Size: 832x1216, Model: uwazumimix_v20_1089578, Version: f2.0.1v1.10.1-previous-634-g37301b22, Emphasis: No norm, Module 1: sdxl.vae, Model hash: 805592afb8, Hires steps: 25, Hires upscale: 2, Original Size: 832x1216, Schedule type: Karras, Hires Module 1: Use same choices, Hires upscaler: 4x_foolhardy_Remacri, Hires CFG Scale: 7, Denoising strength: 0.35, Hires schedule type: Karras, Clip skip: 2
使用介绍
关于v1.2版本
我感觉手失败的概率降低了一些。 在保持面部印象的同时进行调整是很困难的。 这次,我把熬好的汤里的上清液取出来。 请仅将其用于个人用途。
我认为最好使用基于pony的质量提示。还需要photoreal关键字。 我认为pony大模型的负面提示需要加入 gorilla
如果提示太长,事情往往会失控。如果将score_6, score_5, score_4, low quality, bad quality, 差质量增加1.2,它可能会平静下来。 我觉得在皮肤纹理中添加一个”fresh”这样的关键词会让它看起来更真实。 问题是脸部往往会变得模糊。 我觉得如果合适的话,它会产生一个非常好的图像。
作品参照
score_9, score_8_up, score_7_up, masterpiece,
hyper photorealism, fair pale skin,
The raw photo of Japanese woman's vacation scene,
at beach, hand on chest, bikini,
Negative prompt: score_6, score_5, score_4, low quality, bad quality, gorilla,
blurry simple background,
Steps: 20, CFG scale: 7, Sampler: Euler a, Seed: 2355372291, VAE: sdxl_vae.safetensors, Size: 832x1216, Model: test04, Version: f1.0.2-v1.10.1RC-latest-695-gf9f9cd12, Model hash: 4be29f095c, Cutoff strong: False, Cutoff weight: 1.8, Schedule type: Automatic, Cutoff enabled: True, Cutoff padding: _, Cutoff targets: white, ADetailer model: mediapipe_face_full, ADetailer version: 24.9.0, ADetailer mask blur: 4, ADetailer confidence: 0.3, Cutoff interpolation: lerp, ADetailer dilate erode: 4, Cutoff disable_for_neg: True, ADetailer inpaint padding: 32, ADetailer denoising strength: 0.2, ADetailer inpaint only masked: True, Clip skip: 2
score_9, score_8_up, score_7_up, JPN, (HDR,:0.5)
hyper photorealism, fair pale skin, sharp focus, lively Glitter skin, realistic background, cosplay crowd,
The raw photo of 2 young Japanese girl's night vacation scene, petit, very shorty, low brightness,
night party, so fun, happy, having drink, drinking, cheek to cheek, (Yuri shot,:1.5)
Helloween party, black witch cosplay, (midnight, dark, :1.4)
gaze away,
Negative prompt: score_6, score_5, score_4, low quality, bad quality, gorilla, solo focus,,
blurry simple background, (cartoonish:1.2), big eyes, teeth
Steps: 20, CFG scale: 7, Sampler: Euler a, Seed: 4151667818, VAE: sdxl_vae.safetensors, Size: 832x1216, Model: test04, Version: f1.0.2-v1.10.1RC-latest-695-gf9f9cd12, Emphasis: No norm, Model hash: 4be29f095c, Cutoff strong: False, Cutoff weight: 1.8, Schedule type: Automatic, Cutoff enabled: True, Cutoff padding: _, Cutoff targets: white, ADetailer model: mediapipe_face_full, ADetailer version: 24.9.0, ADetailer mask blur: 4, ADetailer confidence: 0.3, Cutoff interpolation: lerp, ADetailer dilate erode: 4, Cutoff disable_for_neg: True, ADetailer inpaint padding: 32, ADetailer denoising strength: 0.2, ADetailer inpaint only masked: True, Clip skip: 2
score_9, score_8_up, score_7_up, 8k, JPN,
hyper photorealism, fair pale skin, dramatic lighting, bloom, (backlit,:0.5)
The raw photo of a young Japanese girl,
detailed waring wedding dress costume,
at room, selfie shot, clothes, looking up, leaning, happy, awaiting kiss face, shy, (kiss shaped lips,:1.1) kiss on viewer, (closed eyes,:1.1) solo focus, so lovely, (from above:0.9), tilt head, open arms
Negative prompt: score_6, score_5, score_4, low quality, bad quality, gorilla, loli, (phone:1.3), teeth,
simple background, (cartoonish:1.2), bad-finger, bad hand, big eyes, thick lips, makeup
Steps: 25, CFG scale: 7, Sampler: Euler a, Seed: 3084719719, VAE: sdxl_vae.safetensors, Size: 832x1216, Model: test07, Version: f1.0.2-v1.10.1RC-latest-695-gf9f9cd12, Emphasis: No norm, Model hash: 8c92db667b, Cutoff strong: False, Cutoff weight: 1.8, Schedule type: Automatic, Cutoff enabled: True, Cutoff padding: _</w>, Cutoff targets: white, ADetailer model: mediapipe_face_full, ADetailer version: 24.9.0, ADetailer mask blur: 4, ADetailer confidence: 0.3, Cutoff interpolation: lerp, ADetailer dilate erode: 4, Cutoff disable_for_neg: True, ADetailer inpaint padding: 32, ADetailer denoising strength: 0.2, ADetailer inpaint only masked: True, Clip skip: 2