Atomsphere Style LoRA can enhance the atmosphere, details, and composition of an image.It is a model with a relatively flat style.Atomsphere Style LoRA 可以改善画面的氛围、细节与构图。模型风格较平。 It trains on anima-base 1.0 and works properly with most Anima-series models and is especially suitable for improving illustration generation. Specific version for miaomiao harem v1.3 are listed above as well.以 anima-base 1.0 模型为基础训练，能够在绝大多数 Anima 系列模型上正常使用，非常适合用于改善插画生成。对于 Miaomiao harem v1.3 有特别适配版。<ul><li>Recommended strength: 1.0 ( 0.5-1.1，depend on your base model)</li><li>Trigger word (optional): atomsphere_style</li><li>推荐强度：1.0（0.5-1.1，根据你的基座模型调节）</li><li>触发词 (可选)：atomsphere_style</li></ul>If you find it useful, please consider giving it a like. Thank you ❤Recommend to prepend trigger words to your prompt. Note: "atomsphere" not atmosphere" to avoid natural language confliction.As anima-branch models become more heavily fine-tuned, the effect may become weaker on certain models. You can let me know if this happens, and I am currently adapting for these cases to achieve the best results across different models.但是随着 anima 分支模型的全量微调程度增加，某些模型上可能效果会变弱，可以通知我，我正在针对此类情况适配，以在各个模型上达到最佳效果。<h2 id="style-description">Style Description</h2> For Anima base models, it can add detail, enrich colors, and improve composition. 针对 Anima 基础模型，它可以增加细节、丰富颜色与改善构图。 It focuses more on a modern anime-style aesthetic, with visuals that tend to look less explicit. 它更注重当代二次元美型，画面表现看起来不那么 explicit。Preserves fine details with a high level of refinement, without introducing overly stylized sketch-like strokes or textures, which also helping the result follow the prompt. 保留细节的高精细度，不会出现过于风格化的草稿笔触与质感，同时也有助于结果遵循指令。 Enhances the sense of air and depth, while reducing the oily texture commonly seen in anime AI outputs. This oily look often comes from rendering all materials as high-contrast, glossy surfaces, a tendency commonly seen in early CG and mordern semi-realistic game art. 加深空气感与层次感，减少anime AI 的油皮质感。AI 的油感大多源于将所有材质都刻画为高对比度光滑材质，在早年 CG 与写实风游戏较为常见。 To reduce this oily appearance, some models tend to move toward the opposite: a flatter cel-shaded style. However, the recent aesthetic trend in anime illustration is a combination of both. While emphasizing graphic composition and color composition, material contrast is mostly defined through binary value contrast, with only subtle volume rendering preserved. 为了减少油感，一些模型又会走另一边：风格平面塞璐璐风格。近年来的 anime 插画审美趋势是两者的结合，在注重平面构成与色彩构成的同时，材质对比大多由二分的对比度确定，保留微弱的体积塑造。<h2 id="changelog">Changelog </h2>The creation of this LoRA, from dataset preparation and training to effect evaluation, was completed through multiple rounds of manual iteration, aiming to achieve the best balance between prompt adherence, detail quality, and stylization.From v1.1 to v1.2, the process went through several aesthetic regressions, and three intermediate versions were discarded.If you find it useful, please consider giving it a free like to show your support. Thank you.制作此 lora 从数据集训练效果评估均人工多次迭代完成，以达到prompt 遵循度、精细度与风格化的最佳平衡。从 v1.1 到 v1.2 中途经历了好几次美学效果倒退的过程，废弃了三个版本。如果您觉得对您有用，可以考虑点个免费的赞表示支持，谢谢～ v1.2 - Improve the naturalness of disordered arrangements of organic shapes.v1.1 - Compared with v1.0, the effect is stronger and the style is flatter.

Atomsphere Style - Anima Lora

改进构成： - 一直以来，画圆形的花时，大部分模型容易等距等大小排列；画海浪等有机形背景时，容易画成几何圆贴图，这些排列都是不太自然的。此版本对有机形的无序分布稍做了改进。 - 人物气质上，男生会比上个版本显更硬朗。上个版本的氛围对比度稍低，气氛柔和，你可以选择喜欢的版本用 最新的 atomsphere_lora 基于 anima_base 训练，而在目前的 miaomiao herum v1.3 分支中，此 lora 的效果几乎不再生效，因此从 v1.2 开始，将开启多个分支。未来的更多模型全量微调后，应该会有许多分支。此 lora style 已经几乎定下了，不会再有大的风格更新，转为维护流行模型的适配。

Files with SHA256: 73623ebb866b9b8a977ef5f19c74fb929f85282600b5f4725faf9fd48193aaac