Sign In

Ray Krea2 x Mid Aesthetic

Early Access

Updated: Oct 19, 2026

base modelnijikrea2ray3780

Download

1 variant available

int8 SafeTensor

Ray Krea2 x Mid Aesthetic v1 turbo INT8ConvRot.safetensors

8-bit integer, smaller file • 11.94 GB

Verified:

Type
Checkpoint Merge
Stats

95

18

Reviews
Published

Oct 19, 2026

Base Model
Hash
AutoV2554B9D48B9
图像节点 (10).png

Recommended settings: er_sde + simple, 8-12 steps, 1 cfg

推荐设置:er_sde + simple,8-12 步,1 cfg

This checkpoint is a reconstruction derived from a fusion of several LoRAs created by @OnlyLemon. These LoRAs were trained on the creator’s own original works, incorporating a wealth of exclusive visual language and the refined textural qualities of closed-source models. The resulting imagery clearly reflects the narrative tension characteristic of Niji, the sophisticated lighting and shading of Midjourney, and the somber, aesthetic DNA of "Higgdfield soul."

Building upon this foundation, I have refined the style, optimized weight merging, and fine-tuned the model’s layers. This has broadened its versatility across various subjects—a reflection of my personal preference for more comprehensive, all-purpose checkpoints.

The core appeal of this model lies in its highly distinctive aesthetic—one that is instantly recognizable. It carries an inherent, moody, and romantic Gothic melancholy, excelling at crafting atmospheric scenes through bold color contrasts and cinematic compositions, all while maintaining a consistently high level of polish across a wide range of themes.

In short, rather than being a model specialized for a single style, it operates within a cohesive, high-end aesthetic framework. It is well-suited for diverse creative needs—encompassing portraits, landscapes, sci-fi, nature, and mechanical subjects—while preserving a clear, signature artistic imprint. It is equally capable of supporting targeted, stylized projects and adapting to a broad variety of visual scenarios.

这款检查点脱胎于创作者 @OnlyLemon 的多款 LoRA 的融合重构 —— 这些 LoRA 均基于她本人的原创作品训练而成,注入了大量独家视觉语言与闭源模型的质感精髓,在画面中能清晰窥见 Niji 的叙事张力、Midjourney 的高级光影质感与 Higgdfield soul 的阴郁美学基因。

我在此基础上完成了风格提炼、权重合并与层级精细调校,进一步补全了模型的题材适配能力,这也契合我的个人喜好(我喜欢功能更加全面的检查点)

这款模型最核心的吸引力,在于其辨识度极高的独特美学体系:自带沉郁浪漫的哥特忧郁气质,擅长以强烈的色彩对比、极具电影感的构图搭建画面情绪,同时在多元题材下始终保持统一的高级完成度。

整体而言,它并非单一风格的专精模型,而是在统一的高级审美框架下,兼顾人像、场景、科幻、自然、机械等多元创作需求,同时保留了鲜明的作者风格印记,既能支撑定向风格化创作,也能适配多样化的出图场景。

Have a great time!祝你们玩的开心!