Stable Diffusion - SDXL 模型测试 (DreamShaper 和 GuoFeng v4) 与全身图像参数配置

欢迎关注我的CSDN：https://spike.blog.csdn.net/
本文地址：https://spike.blog.csdn.net/article/details/132085757

图像来源于 GuoFeng v4 XL 模型，艺术风格是赛博朋克、漫画、奇幻。

全身图像是指拍摄对象的整个身体都在画面中的照片，可以展示人物的姿态、服装、气质等特点，也可以表达一种情绪或故事。全身图像的拍摄需要注意构图、光线、角度、姿势等方面。

全身图像的提示词：

full body shot,(head-to-toe shot:1.2),1girl,an asian beatiful woman in a dress and jacket standing,(office lady high heels),
pantyhose,black leggings,outdoors,
people in the center,soft light,natural and comfortable pose,face to camera,
<lora:neg4all_xl_v6:1>,<lora:pantyhose_widget_xl_v10:1.2>,负向提示词: nsfw,breast,(ng_deepnegative_v1_75t:1.2),badhandv4,

SDXL 的提示词一般不需要高质量、低质量等提示词，尽量简洁。

随机种子：3830551278、3193321894、2206410753

常用的描述全身的词汇：

full-body shot
head-to-toe shot
complete body photograph
comprehensive full-body image
overall body shot

注意配置，保持图像稳定：

1. 图像尺寸选择 768x1152，其他尺寸会导致人物比例失衡；
2. 开启高分辨率修复，放大算法4x-UltraSharp，重复幅度 0.3；开启 面部修复；
3. 开启 AfterDetailer，修复脸部和手部细节。

配置如下：

结合不同的风格，绘制全身图像，例如 赛博朋克 (Neonpunk)、幻想艺术 (Fantasy art)、影片风格 (Cinematic)、高增强 (Enhance) 等。

测试模型 DreamShaper_XL1.0_alpha2，即：

测试模型 GuoFeng4_XL_Real-Beta，即：

下载 GuoFeng 模型的真实版，来源于 吐司Art 模型网站：

wget https://download.tusiassets.com/users%2FCHECKPOINT%2F610052112453282992%2FbOP6tNK2FNrxC34OXF-CZ.safetensors?Expires=1691131569&OSSAccessKeyId=LTAI5tA9wgpRFMZ23PwaJcL4&Signature=gVHvtdRPkOrKiEceHLiFd8i%2Fbp8%3D -O GuoFeng4_XL_Real-Beta.safetensors

其他测试模型 ArienmixxlAsian_v10_SDXL1.0、SDVN6-RealXL，效果均不稳定。

测试参数示例：

ethereal fantasy concept art of full body shot,(head-to-toe shot:1.2),1girl,an asian beautiful woman in a dress and jacket standing,(high heels),
pantyhose and leggings,outdoors,
people in the center,soft light,natural and comfortable pose,face to shot,
<lora:neg4all_xl_v6:1>,<lora:pantyhose_widget_xl_v10:1.2>, . magnificent, celestial, ethereal, painterly, epic, majestic, magical, fantasy art, cover art, dreamy
Negative prompt: nsfw,breast,(ng_deepnegative_v1_75t:1.2),badhandv4,, photographic, realistic, realism, 35mm film, dslr, cropped, frame, text, deformed, glitch, noise, noisy, off-center, deformed, cross-eyed, closed eyes, bad anatomy, ugly, disfigured, sloppy, duplicate, mutated, black and white
Steps: 35, Sampler: DPM++ 2M SDE Karras, CFG scale: 7, Seed: 2206410753, Face restoration: GFPGAN, Size: 768x1152, Model hash: 57fdfb1fbe, Model: GuoFeng4_XL_Real-Beta, Clip skip: 2, ADetailer model: face_yolov8n.pt, ADetailer confidence: 0.3, ADetailer dilate/erode: 4, ADetailer mask blur: 4, ADetailer denoising strength: 0.4, ADetailer inpaint only masked: True, ADetailer inpaint padding: 32, ADetailer version: 23.7.6, Lora hashes: "neg4all_xl_v6: 9a735be26f5e, pantyhose_widget_xl_v10: 90b94c2a1974", Version: v1.5.1

推荐 SDXL 版本的相关 LoRA，即：

• 负向提示词 LoRA：neg4all_xl_v6
• 袜子 LoRA：pantyhose_widget_xl_v10