<h1 id="2222">训练集收集自群友水无22，因此本模型命名为水无22。我也不知道这是具体什么画风</h1><h1 id="anima-lora">一个基于 <strong>Anima</strong> 训练的柔光系画风 LoRA，主要强化：</h1><ul><li><p>柔和自然的光影氛围</p></li><li><p>高曝光、低对比的梦幻观感</p></li><li><p>通透细腻的皮肤质感</p></li><li><p>淡彩色调与空气感表现</p></li><li><p>发丝透光与环境光融合</p></li><li><p>适合人物特写、海边、窗边、室内柔光等场景</p></li></ul><h1>整体风格偏向：</h1><p>柔光、梦幻、空气感、奶油肌肤、浅色系、氛围感插画</p><hr /><h1>触发词</h1><pre><code>shuiyue
</code></pre><p>推荐放在 Prompt 前部使用。</p><hr /><h1>推荐参数</h1><h3 id="base-model">Base Model</h3><pre><code>Model: anima-base-v1.0
Model hash: bd43b7cffe
Module 1: qwen_3_06b_base
Module 2: qwen_image_vae
</code></pre><h3 id="generation">Generation</h3><pre><code>Steps: 36
Sampler: Euler a
Schedule type: Beta
CFG scale: 5
Shift: 3
Seed: 114514
Size: 1024x1536
</code></pre><h3 id="hires-fix">Hires Fix</h3><pre><code>Denoising strength: 0.6
Hires upscale: 1.5
Hires upscaler: 4xHFA2k_ludvae_realplksr_dysample

Hires steps: 32
Hires CFG Scale: 5
Hires Shift: 3
</code></pre><h3 id="other">Other</h3><pre><code>RNG: CPU
Emphasis: Original
Beta schedule alpha: 0.6
Beta schedule beta: 0.6
</code></pre><hr /><h2>推荐权重</h2><h3>平衡模式</h3><pre><code>&lt;lora:shuiwu:0.8&gt;
</code></pre><p>兼顾风格与 Prompt 服从性。</p><h3>推荐模式</h3><pre><code>&lt;lora:shuiwu:1&gt;
</code></pre><p>获得最完整的画风表现。</p><h3>强风格模式</h3><pre><code>&lt;lora:shuiwu:1.2&gt;
</code></pre><p>风格影响更强，但可能降低部分 Prompt 自由度。</p><hr /><h2 id="prompt">示例 Prompt</h2><pre><code>masterpiece, best quality, shuiyue, soft light, 1girl
</code></pre><hr /><h2>推荐关键词</h2><pre><code>soft light,
dreamy atmosphere,
pastel colors,
window light,
backlighting,
glowing hair,
blurry background,
warm sunlight,
high exposure,
cinematic lighting,
delicate skin,
airy atmosphere
</code></pre><hr /><hr /><h2>训练信息</h2><pre><code>Resolution: 1216x1664
Rank: 32
Alpha: 32
Learning Rate: 2e-5
Mixed Precision: bf16
</code></pre><hr /><h2 id="metadata">Metadata</h2><pre><code>Lora hashes: "shuiwu-000045: b2965c7974b6"
Version: neo-2.24
</code></pre><p>感谢下载与使用，希望你喜欢这套柔光空气感画风。</p><h1 id="shuiwu-style-lora">Shuiwu Style LoRA</h1><p>A soft-light anime illustration LoRA trained for <strong>Anima</strong>, focusing on dreamy lighting, delicate skin rendering, atmospheric color grading, and high-quality character illustrations.</p><h2 id="style-features">Style Features</h2><ul><li><p>Soft and diffused lighting</p></li><li><p>High-exposure aesthetic</p></li><li><p>Smooth skin shading</p></li><li><p>Pastel color palette</p></li><li><p>Dreamy atmosphere</p></li><li><p>Delicate hair highlights</p></li><li><p>Gentle contrast</p></li><li><p>Strong ambient lighting</p></li><li><p>Cinematic mood</p></li></ul><p>The overall style emphasizes:</p><p>Soft Light · Airy Atmosphere · Dreamy Colors · Delicate Skin · High-Key Lighting · Anime Illustration</p><hr /><h2 id="trigger-word">Trigger Word</h2><pre><code>shuiyue
</code></pre><p>Recommended to be placed near the beginning of the prompt.</p><hr /><h2 id="recommended-settings">Recommended Settings</h2><h3 id="base-model">Base Model</h3><pre><code>Model: anima-base-v1.0
Model hash: bd43b7cffe
Module 1: qwen_3_06b_base
Module 2: qwen_image_vae
</code></pre><h3 id="generation">Generation</h3><pre><code>Steps: 36
Sampler: Euler a
Schedule type: Beta
CFG scale: 5
Shift: 3
Seed: 114514
Size: 1024x1536
</code></pre><h3 id="hires-fix">Hires Fix</h3><pre><code>Denoising strength: 0.6
Hires upscale: 1.5
Hires upscaler: 4xHFA2k_ludvae_realplksr_dysample

Hires steps: 32
Hires CFG Scale: 5
Hires Shift: 3
</code></pre><h3 id="other-settings">Other Settings</h3><pre><code>RNG: CPU
Emphasis: Original
Beta schedule alpha: 0.6
Beta schedule beta: 0.6
</code></pre><hr /><h2 id="recommended-lora-weight">Recommended LoRA Weight</h2><h3 id="balanced">Balanced</h3><pre><code>&lt;lora:shuiwu:0.8&gt;
</code></pre><p>Good balance between style strength and prompt flexibility.</p><h3 id="recommended">Recommended</h3><pre><code>&lt;lora:shuiwu:1&gt;
</code></pre><p>Best overall visual quality and style representation.</p><h3 id="strong-style">Strong Style</h3><pre><code>&lt;lora:shuiwu:1.2&gt;
</code></pre><p>Maximum style impact with reduced prompt flexibility.</p><hr /><h2 id="example-prompt">Example Prompt</h2><pre><code>masterpiece, best quality, shuiyue, soft light, 1girl
</code></pre><hr /><h2 id="suggested-keywords">Suggested Keywords</h2><pre><code>soft light,
dreamy atmosphere,
pastel colors,
window light,
backlighting,
glowing hair,
warm sunlight,
blurry background,
cinematic lighting,
high exposure,
delicate skin,
airy atmosphere
</code></pre><hr /><h2 id="training-information">Training Information</h2><pre><code>Resolution: 1216x1664
Rank: 32
Alpha: 32
Learning Rate: 2e-5
Mixed Precision: bf16
</code></pre><hr /><h2 id="metadata">Metadata</h2><pre><code>Lora hashes: "shuiwu-000045: b2965c7974b6"
Version: neo-2.24
</code></pre><hr /><h3 id="credits">Credits</h3><p>Model description and documentation generated with assistance from <strong>ChatGPT 5.5</strong>.</p><p></p>

Files with SHA256: e6aa038596c16e775f072b6d9dbab966f5f82cae7bf5a9eaeea94905367e7b85