参数名 | 参数值 |
---|---|
Cache FP16 weight for LoRA | False |
Denoising strength | 0.5 |
Downcast alphas_cumprod | True |
ENSD | 31337 |
FP8 weight | Enable for SDXL |
Hires steps | 16 |
Hires upscale | 1.6 |
Hires upscaler | R-ESRGAN 4x+ Anime6B |
Model | AnythingXL-Beta4 |
Model hash | 8421598e93 |
Old prompt editing timelines | True |
Size | 832×1216 |
VAE | sdxl_vae.safetensors |
VAE hash | 63aeecb90f |
Version | v1.8.0 |
cfgScale | 5 |
hashes | {“model”:”8421598e93″} |
negativePrompt | lowres, bad, error, fewer, extra, missing, worst quality, jpeg artifacts, bad quality, watermark, signature, extra digits, artistic error, username, abstract, artist logo, artist name, signature, bad fingers, bad hand |
prompt | masterpiece, best quality, 1girl, solo, animal ears, bow, teeth, jacket, tail, open mouth, brown hair, orange background, bowtie, orange nails, simple background, cat ears, orange eyes, blue bow, animal ear fluff, cat tail, looking at viewer, upper body, shirt, school uniform, hood, striped bow, striped, white shirt, black jacket, blue bowtie, fingernails, long sleeves, cat girl, bangs, fangs, collared shirt, striped bowtie, short hair, tongue, hoodie, sharp teeth, facial mark, claw pose |
resources | [{“hash”:”8421598e93″,”name”:”AnythingXL-Beta4″,”type”:”model”}] |
sampler | Euler a |
seed | 3665112698 |
steps | 28 |
在下载模型之前,请详细查看模型介绍。本次引入的AnythingXL β4版是基于SDXL模型高度开发的,模型质量得到更高的提升。模型合并是死胡同,关于艺术风格多样性和一些提示词的准确性,完全是一团糟。
这是专门给中国和日本的AI玩家的建议:除非你需要生成较大尺寸的图像,否则不建议使用高清修复。如果你觉得模型有所偏向,首先检查提示词。别用SD1.5的习惯来使用SDXL,两者本质不同。
模型融合的顺序无关模型质量。按照FairAIPublicLicense1.0-SD,公开披露配方,并明确合并此模型的所有模型必须依照此许可披露融合配方。
我们推荐的提示词模板是结构化的,这有助于引导生成结果走向希望的方向。而且,如果你想生成高质量的图片,你可以使用质量词和负面提示词。
AnythingXL采用了Fair AI Public License 1.0-SD许可,和Stable Diffusion模型兼容。 总体讲,更改Anything XL后,你必须分享你的更改和原许可证,任何分发都必须在本许可证或其他类似规则的许可证下进行。
下载地址1立即下载
评论0