Lustrous Highlights Style Ver1.1 is out!
This update focuses on structural stability for complex poses:
V1.1 handles complex NSFW poses significantly better — multi-character scenes hold up much more reliably, with far fewer broken or merged figures;
The trade-off: the pure aesthetic ceiling is slightly reduced — if you mainly generate SFW content, V1.0 still delivers the better look.
Which version should you use?
Mostly SFW / chasing the aesthetic ceiling → V1.0 is more than enough
Mostly NSFW / complex poses / multi-character scenes → V1.1 is the better pick
For V1.1, just set strength to 1.0. Full usage notes below.
My Anima LoRA: trained on a 171-image dataset for 3000 steps.
Faithfully reproduces the Japanese anime/illustration style — clean, smooth linework with even, natural variation; a blend of thick paint and gradients (between cel-shading and thick paint); soft light-to-shadow transitions with strong highlights; a moist, glossy finish; and a strong atmosphere that takes both warm and cool light.
It's a style LoRA: the dataset is mostly futa images, but it won't produce futa content unless you explicitly prompt for it. The trigger word is "Futanari" (just my quirky taste) — it doesn't affect image quality. Safe for everyone who's not into futa.
Recommended strength:
V1.0: 0.7–1.0. I personally use 0.7 — at this level, structural issues are unlikely.
V1.1: 1.0.
Prompt tip: adding "eyeshadow" to your prompt brings out this LoRA's flavor best; without it, the style may come through less noticeably.
Usage: stacking it with other style LoRAs (Anima Masterpiece, Anima Highres, etc.) may dilute this LoRA's style. For the full flavor, use it standalone.
Pairs great with my random prompt gacha workflow, Anima Random Gacha:
Anima Random Gacha/随机提示词抽卡 - v1.0 | Anima Workflows | Civitai
If you like it, a like or favorite means a lot — it keeps me training. Thank you!
Lustrous Highlights Style Ver1.1 发布！
本次更新的重点是复杂动作的结构稳定性：
V1.1 在复杂 NSFW 动作上的表现明显更好——多 P 场景结构更稳，肢体更不容易崩；
代价是纯粹的美学上限略有收敛，如果主要出 SFW 内容，V1.0 的观感依然更好。
版本怎么选？
以 SFW 为主 / 追求美学上限 → V1.0 已足够
常玩 NSFW / 复杂动作 / 多 P 场景 → V1.1 更合适
V1.1 强度直接设为 1.0 即可，其他使用说明见下方。
我的 Anima LoRA：171 张图数据集、3000 步训练。
风格：日式动漫/插画风——线条干净流畅、粗细均匀、变化自然；厚涂与渐变混合（介于赛璐璐与厚涂之间）；光影过渡柔和、高光强烈；湿润光泽感；氛围感强，暖光冷光都吃得开。
定位：这是风格 LoRA。数据集虽以 futa 图为主，但不会主动产生 futa 内容，除非提示词中明确写出。触发词是 Futanari（个人恶趣味），不影响出图质量，不喜欢 futa 的放心用。
强度建议：
V1.0：0.7~1.0，个人常用 0.7，这个强度下最不容易出结构问题；
V1.1：直接 1.0。
提示词小技巧：提示词中加入 eyeshadow（眼影）能最大程度表现出本 LoRA 的风味，不加的话效果可能没那么明显。
使用说明：与其他风格 LoRA（如 Anima Masterpiece、Anima Highres）叠加使用可能稀释本 LoRA 的风格，想体验完整风味建议单独使用。
搭配随机提示词抽卡工作流 Anima Random Gacha 效果更佳：
Anima Random Gacha/随机提示词抽卡 - v1.0 | Anima Workflows | Civitai
喜欢的话欢迎点赞收藏——你们免费的点赞就是我继续炼 LoRA 的动力，谢谢！