santa hat
deerdeer nosedeer glow
Sign In

SCMix

973
9.6k
70
Updated: Oct 5, 2024
styleanime
Verified:
PickleTensor
Type
Checkpoint Merge
Stats
5,476
Reviews
Published
Mar 15, 2023
Base Model
SD 1.5
Hash
AutoV2
FFA9C19068
default creator card background decoration
Getsc's Avatar
Getsc

Best practice: Resolution=768*960, hires=2x, clip=2, sampler DMP++2M karras, 40 steps, for more parameters please refer to my diagram.

SCMIX -

Mixed from pastel and abyssorangemix3AOM3_aom3a3

I really love detailed and chaotic images, as well as intense lighting and shadows. Therefore, I extracted the weight layers related to lighting and shadows from aom3 and mixed them with pastel through layer weighting. After a lot of filtering and iteration, I chose this version that is most suitable for me!

This is my favorite version, containing everything I love perfectly!

SCMIX_NSM -

Mixed from pastel and abyssorangemix3AOM3_aom3a3, as well as subtle nightSkyYOZORAStyle and meinamix_meinaV8,

I hope to achieve stronger contrast and complex backgrounds with this version, which is more suitable for long-range and full-body shots, of course, depending on your preferences~

You will notice that my sample images contain a lot of out border and tarot-style elements, thanks to two lora, one is the trained gsr, which makes the character's decorations more complex and gorgeous, and additional related items appear in the scene.

The second is tam, which adjusts the fusion from tachi-e, and I deleted the parts that would greatly change the composition and destroy the lighting.

As well as the fusion of the two, you can download it at the following link.

SCMix_grc_tam | Stable Diffusion LORA | Civitai

• 最佳实践:分辨率=768*960,hires=2倍,clip=2,采样器DMP ++2M karras,40step,更多参数可以参考我的示意图

SCMIX—

—混合自pastel and abyssorangemix3AOM3_aom3a3

我超级喜欢细节丰富,杂乱的画面,以及浓度强烈的光影,故我将aom3中的涉及光影的权重层提取出于pastel进行分层权重混合,在通过大量的筛选迭代中选出了这个最合适的版本!

这是我最喜欢的版本,包含完美的我喜爱的一切!

SCMIX_NSM—

—混合自pastel and abyssorangemix3AOM3_aom3a3,以及细微的nightSkyYOZORAStyle和meinamix_meinaV8,

希望从中获得更强的对比度和复杂的背景,这个版本更适合远景,全身,当然看你喜好~

你会注意到,我的示例图片中包含了大量的out border,塔罗牌等风格的元素,这得益于两个lora,一个是训练的gsr,它更会让人物的装饰更加复杂和华丽,并在场景中出现额外的相关物品

第二个是tam,他调整融合自tachi-e,我删除了其中会大幅度改变构图,并破坏掉光影的部分

以及两者的融合版,你可以在以下链接中去下载

SCMix_grc_tam | Stable Diffusion LORA | Civitai