【CHECKPOINT】ArienMixXL Asian portrait东方女孩美感v40Pruned[E80A4FB006]

资源详情：

Directional tweaks based on ArienRealWordV1 to change the stereotypes of Asian people at the bottom. It can be used as a refiner in comfyUI to acquire Asian characteristics.

Suggestions for use:

Positive: from + perspective,1girl,solo,realistic

Reverse: cgi, illustration, cartoon, deformed, distorted, disfigured, poorly drawn, bad anatomy, wrong anatomy,ugly, deformed, blurry,Noisy,log,text

VAE: No external VAE is required

Recommended size: 1024*1024, 896x1180, 896x1152, 1024x1280, 1280x768

Recommended sampler: DPM++ 3M SDE Karras

基于 ArienRealWordV1 的方向性调整，以改变底层亚洲人的刻板印象。它可以用作 comfyUI 中的精炼器，以获得亚洲特色。

使用建议：

正面：从+视角，1个女孩，独奏，现实

反面：cgi，插图，卡通，变形，扭曲，毁容，画得很差，解剖学不好，解剖学错误，丑陋，变形，模糊，嘈杂，日志，文本

VAE：无需外部VAE

推荐尺寸：1024*1024、896x1180、896x1152、1024x1280、1280x768

推荐采样器：DPM++ 3M SDE Karras

建议步数：50-150

提示词：arien photography，chinese art photography，nfsw

Hash parameter SHA256 E80A4FB006FA48A2238DFEE368E291DCF09A240727FEA40DE61B3E754F5EC615 CRC32 19BD398D BLAKE3 EB3A6AAB3FD192EE90CD2162E8CF698BDCE7EDED699FD4FADE7601956F1ABF97 AUTOV1 5FA1EFDD AUTOV2 E80A4FB006

作品展示：

作品参数：

1. arien photography,chinese art photography,1girl,solo,hair ornament,black hair,jewelry,earrings,holding,red flower,looking at viewer,long hair,spider lily,red lips,black eyes,makeup,portrait,eyelashes,lips,lipstick,hair flower,blue eyes,tassel,upper body,parted lips,
   
2. Negative prompt: cgi, illustration, cartoon, deformed, distorted, disfigured, poorly drawn, bad anatomy, wrong anatomy,ugly, deformed, blurry,Noisy,log,text,watermark
   
3. Steps: 50, Size: 768x1280, Seed: 2492757457, Model: ArienMixXL亚洲人像V4, Version: v1.7.0, Sampler: DPM++ 3M SDE Karras, CFG scale: 7, Clip skip: 2, Model hash: e80a4fb006, ADetailer model: face_yolov8n.pt, ADetailer version: 24.1.0, 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

复制代码

作品参数：

1. 1girl, solo, black hair, ponytail, smile, realistic, lips, blue eyes, long hair, tank top, closed mouth, upper body, shirt, looking at viewer, blurry, bare shoulders, collarbone, black eyes, traditional media, white shirt, watermark, head tilt, portrait, nose, blurry background, looking to the side, sleeveless
   
2. Negative prompt: cgi, illustration, cartoon, deformed, distorted, disfigured, poorly drawn, bad anatomy, wrong anatomy,ugly, deformed, blurry,Noisy,log,text,watermark
   
3. Steps: 50, Size: 768x1280, Seed: 3857740008, Model: ArienMixXL亚洲人像V4, Version: v1.7.0, Sampler: DPM++ 3M SDE Karras, CFG scale: 7, Clip skip: 2, Model hash: e80a4fb006, ADetailer model: face_yolov8n.pt, ADetailer version: 24.1.0, 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

复制代码

作品参数：

1. arien photography, chinese art photography, 1girl, solo, jewelry, earrings, black hair, hair ornament, blurry background, dress, blurry, upper body, bare shoulders, long hair, profile, off shoulder, green dress, closed mouth, realistic, red lips, chinese clothes, from side, lips, half-closed eyes, from behind, makeup, flower, eyelashes, off-shoulder dress, aqua dress, watermark, hair flower
   
2. Negative prompt: cgi, illustration, cartoon, deformed, distorted, disfigured, poorly drawn, bad anatomy, wrong anatomy,ugly, deformed, blurry,Noisy,log,text,(worst quality, low quality),(Poorly Made Bad 3D, Lousy Bad Realistic:1.1),(blurry),body painting,
   
3. Steps: 50, Size: 768x1280, Seed: 597990517, Model: ArienMixXL亚洲人像V4, Version: v1.7.0, Sampler: DPM++ 3M SDE Karras, CFG scale: 7, Clip skip: 2, Model hash: e80a4fb006, ADetailer model: face_yolov8n.pt, ADetailer version: 24.1.0, 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

复制代码

作品参数：

1. 1girl, black hair, closed eyes, water, blurry, blurry background, solo, partially submerged, swimsuit, red lips, upper body, lips, outdoors, splashing, bikini, wet, breasts, bare shoulders, day, realistic, hair ornament, depth of field, makeup, closed mouth, parted lips, eyelashes, from side, head tilt, striped, hair bun, single hair bun
   
2. Negative prompt: cgi, illustration, cartoon, deformed, distorted, disfigured, poorly drawn, bad anatomy, wrong anatomy,ugly, deformed, blurry,Noisy,log,text,watermark
   
3. Steps: 50, Size: 768x1280, Seed: 1992171442, Model: ArienMixXL亚洲人像V4, Version: v1.7.0, Sampler: DPM++ 3M SDE Karras, CFG scale: 7, Clip skip: 2, Model hash: e80a4fb006, ADetailer model: face_yolov8n.pt, ADetailer version: 24.1.0, 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

复制代码

作品参数：

1. isometric style Chinese martial arts movie stills,cowboy shot,motion blur,A Chinese woman uses Qing Kung to fly on a black background and Wearing Hanfu,martial,Motion blurred hands,water surrounds her body,glowing,(flowing light particles),flowing smoke,fighting stance,best quality and best aesthetic and (photorealistic:1.4) and (Realistic:1.4),Fujifilm XT3,film still,film grain,cinematic,<lyco:film grainV2:0.4>,looking at viewer and facing viewer,Detailed facial features,detailed eyes,detailed nose,detailed lips,(Laser light) and (glowing magic circle:1.4), . vibrant, beautiful, crisp, detailed, ultra detailed, intricate
   
2. Negative prompt: deformed, mutated, ugly, disfigured, blur, blurry, noise, noisy, realistic, photographic, (worst quality,low quality,illustration,3d,2d,painting,cartoons,sketch:1.3),(blurry:1.4),(ugly nails:1.2) and (ugly fingers:1.2) and (fused fingers:1.2) and (missing finger:1.2) and (extra fingers:1.2) and (bad anatomy),(watermark),(bamboo forest:1.4) full body,
   
3. Steps: 50, VAE: sdxl_vae.safetensors, ENSD: 31337, Size: 608x1248, Seed: 3831776447, Model: sdxl10ArienmixxlAsian_v40Pruned, Version: v1.7.0-RC-39-g39ebd568, Sampler: Euler a, VAE hash: 235745af8d, CFG scale: 2, Clip skip: 2, Model hash: e80a4fb006, Hires steps: 15, Hires prompt: "Chinese martial arts movie stills,cowboy shot,motion blur,A Chinese woman uses Qing Kung to fly on a black background and Wearing Hanfu,martial,Motion blurred hands,water surrounds her body,glowing,(flowing light particles),flowing smoke,fighting stance, film grainV2: 0.4>,looking at viewer and facing viewer,Detailed facial features,detailed eyes,detailed nose,detailed lips, Hires upscale: 2, Hires upscaler: R-ESRGAN 4x+, Denoising strength: 0.3, 1.4) and (Realistic: 1.4),Fujifilm XT3,film still,film grain,cinematic, Style Selector Style: Isometric Style, Style Selector Enabled: True, Style Selector Randomize: True, (Laser light) and (glowing magic circle: 1.4),"

复制代码

下载地址：

游客，如果您要查看本帖隐藏内容请回复

这模型训练的很不错啊！骨架和身材比例随便生成都没看到书面问题，用其他大模型是会经常不正常比例或者手脚奇怪多出来。

我想问下，生成后最好refiner一下，refiner模型就用官方的还是？

一方渺小天地 发表于 2024-2-29 00:11
我想问下，生成后最好refiner一下，refiner模型就用官方的还是？

目前是直接用官方的就好，如果是使用webui重绘幅度建议0.35以下，不使用refiner可以尝试高清修复，一些算法放大一倍也可以提升质感

支持下这模型训练的很不错啊！

这是我见过最真实的人像之一，好好学习下

感谢分享