<h1 id="vivid-real">Vivid Real</h1><hr /><span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">中文 / English<h2>简介</h2>这是一款专注于 二次元角色真实/半真实质感 的高质量 LoRA，生成的作品在人体的合理性与构图的选择上比不少私有模型更高*。它通过精选数据集训练，对皮肤与头发的表现力进行了高度优化，能够在 真实感、半真实感、塑料质感、光泽质感 之间灵活切换/叠加。*注：通过反推提示词，对比部分收费画师的作品，本模型在多数情况下生成准确性更高。已知问题：准确性是相对而言的，在使用真实质感时，对手指的绘画准确性仍然会比原来降低些许。<h3>主要特性</h3><ul><li>皮肤细节：支持 realistic skin、half realistic skin、plastic skin 等多种表现，质感自然或光滑可控。</li><li>头发表现：支持 shiny hair、viscous hair，让发丝呈现光泽或浓稠感，当然，你也可以把它们加入负面提示词。</li><li>高度解耦：关键词互不干扰，可独立触发效果，方便组合。</li><li>适配性强：适用于各种二次元角色风格，既能保持原画风，又能增强真实表现力。</li></ul><h2>生成指引</h2><h3>提示词</h3><pre><code>main theme: realistic, half realistic, &lt;empty&gt;,
hair: viscous hair, shiny hair,
skin: half realistic skin, shiny skin, plastic skin, realistic skin,
body type: young, &lt;-- for more age down</code></pre>它会显著影响以下角色的生成，在生成以下角色时，应适度调整该模型权重：<pre><code>castorice \(honkai: star rail\), castorice \(star rail live 2025\) \(honkai: star rail\), formidable \(azur lane\), ganyu \(genshin impact\), taihou \(azur lane\), taihou \(fun on the baize\) \(azur lane\), eyjafjalla \(arknights\), phoebe \(wuthering waves\), wakaba mutsumi, kafuu chino, shinosawa hiro, kasugano sora, yukinoshita yukino, </code></pre><h3>建议权重</h3>一般情况下可以使用 1.0 权重进行作品生成，当画风过重、人体异常时，适当降低模型权重。<h2>模型训练详情</h2><ul><li>硬件：RTX 4090 GPU</li><li>损失曲线：损失曲线总体呈下降趋势，相邻迭代轮数的模型生成的样本图例未出现大幅波动。</li><li>图像数量：490 张</li><li>迭代轮数：64 轮</li><li>迭代步数：≈ 10,000 步</li><li>批量大小：6 张</li><li>优化器：AdamW 8-bit</li><li>底模：Creation NAIXL V1</li></ul><hr /><h2 id="introduction">Introduction</h2>This is a high-quality LoRA focused on realistic / semi-realistic textures for anime-style characters. Compared to many private models, its results show better anatomical accuracy and compositional choices*. It was trained on a carefully curated dataset, with highly optimized rendering of skin and hair, allowing flexible switching/overlaying between realism, semi-realism, plastic-like textures, and glossy effects.*Note: By backtracking prompt words and comparing with some paid artists’ works, this model generally achieves higher accuracy in most cases.Known Issues: Accuracy is relative—when using realistic textures, the rendering of fingers may still show slight inaccuracies compared to the original.<h3 id="key-features">Key Features</h3><ul><li>Skin details: Supports realistic skin, half-realistic skin, plastic skin, etc. Texture can be natural or smooth, fully controllable.</li><li>Hair rendering: Supports shiny hair and viscous hair, making strands look glossy or thick. Of course, you can also add them to negative prompts.</li><li>High decoupling: Keywords do not interfere with one another. Each effect can be triggered independently, making combinations easy.</li><li>High adaptability: Works with various anime character styles—able to preserve the original art style while enhancing realism.</li></ul><h2 id="generation-guide">Generation Guide</h2><h3 id="prompts">Prompts</h3><pre><code>main theme: realistic, half realistic, &lt;empty&gt;,
hair: viscous hair, shiny hair,
skin: half realistic skin, shiny skin, plastic skin, realistic skin,
body type: young, &lt;-- for more age down</code></pre>This LoRA significantly affects the generation of the following characters. When generating them, it’s recommended to moderately adjust the model weight:<pre><code>castorice \(honkai: star rail\), castorice \(star rail live 2025\) \(honkai: star rail\), formidable \(azur lane\), ganyu \(genshin impact\), taihou \(azur lane\), taihou \(fun on the baize\) \(azur lane\), eyjafjalla \(arknights\), phoebe \(wuthering waves\), wakaba mutsumi, kafuu chino, shinosawa hiro, kasugano sora, yukinoshita yukino, </code></pre><h3 id="suggested-weights">Suggested Weights</h3>In general, you can use 1.0 weight for generation. If the art style becomes too strong or the anatomy looks abnormal, try lowering the model weight.<h2 id="training-details">Training Details</h2><ul><li>Hardware: Trained on RTX 4090 GPU</li><li>Loss curve: Overall downward trend; no major fluctuations observed between iterations in sample outputs</li><li>Number of images: 490</li><li>Epochs: 64</li><li>Steps: ≈ 10,000</li><li>Batch size: 6</li><li>Optimizer: AdamW 8-bit</li><li>Base Model: Creation NAIXL V1</li></ul>

Style: Vivid Real

ai_artist_group_1.safetensors

v1.0

THIS IS A <a href="https://civitai.com/models/1944150" alt="Reprint url">REPRINT</a>!Go to original author to check their other models or to support them. <hr /> <h1 id="vivid-real">Vivid Real</h1><hr /><span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">中文 / English<h2>简介</h2>这是一款专注于 二次元角色真实/半真实质感 的高质量 LoRA，生成的作品在人体的合理性与构图的选择上比不少私有模型更高*。它通过精选数据集训练，对皮肤与头发的表现力进行了高度优化，能够在 真实感、半真实感、塑料质感、光泽质感 之间灵活切换/叠加。*注：通过反推提示词，对比部分收费画师的作品，本模型在多数情况下生成准确性更高。已知问题：准确性是相对而言的，在使用真实质感时，对手指的绘画准确性仍然会比原来降低些许。<h3>主要特性</h3><ul><li>皮肤细节：支持 realistic skin、half realistic skin、plastic skin 等多种表现，质感自然或光滑可控。</li><li>头发表现：支持 shiny hair、viscous hair，让发丝呈现光泽或浓稠感，当然，你也可以把它们加入负面提示词。</li><li>高度解耦：关键词互不干扰，可独立触发效果，方便组合。</li><li>适配性强：适用于各种二次元角色风格，既能保持原画风，又能增强真实表现力。</li></ul><h2>生成指引</h2><h3>提示词</h3><pre><code>main theme: realistic, half realistic, &lt;empty&gt;,
hair: viscous hair, shiny hair,
skin: half realistic skin, shiny skin, plastic skin, realistic skin,
body type: young, &lt;-- for more age down</code></pre>它会显著影响以下角色的生成，在生成以下角色时，应适度调整该模型权重：<pre><code>castorice \(honkai: star rail\), castorice \(star rail live 2025\) \(honkai: star rail\), formidable \(azur lane\), ganyu \(genshin impact\), taihou \(azur lane\), taihou \(fun on the baize\) \(azur lane\), eyjafjalla \(arknights\), phoebe \(wuthering waves\), wakaba mutsumi, kafuu chino, shinosawa hiro, kasugano sora, yukinoshita yukino, </code></pre><h3>建议权重</h3>一般情况下可以使用 1.0 权重进行作品生成，当画风过重、人体异常时，适当降低模型权重。<h2>模型训练详情</h2><ul><li>硬件：RTX 4090 GPU</li><li>损失曲线：损失曲线总体呈下降趋势，相邻迭代轮数的模型生成的样本图例未出现大幅波动。</li><li>图像数量：490 张</li><li>迭代轮数：64 轮</li><li>迭代步数：≈ 10,000 步</li><li>批量大小：6 张</li><li>优化器：AdamW 8-bit</li><li>底模：Creation NAIXL V1</li></ul><hr /><h2 id="introduction">Introduction</h2>This is a high-quality LoRA focused on realistic / semi-realistic textures for anime-style characters. Compared to many private models, its results show better anatomical accuracy and compositional choices*. It was trained on a carefully curated dataset, with highly optimized rendering of skin and hair, allowing flexible switching/overlaying between realism, semi-realism, plastic-like textures, and glossy effects.*Note: By backtracking prompt words and comparing with some paid artists’ works, this model generally achieves higher accuracy in most cases.Known Issues: Accuracy is relative—when using realistic textures, the rendering of fingers may still show slight inaccuracies compared to the original.<h3 id="key-features">Key Features</h3><ul><li>Skin details: Supports realistic skin, half-realistic skin, plastic skin, etc. Texture can be natural or smooth, fully controllable.</li><li>Hair rendering: Supports shiny hair and viscous hair, making strands look glossy or thick. Of course, you can also add them to negative prompts.</li><li>High decoupling: Keywords do not interfere with one another. Each effect can be triggered independently, making combinations easy.</li><li>High adaptability: Works with various anime character styles—able to preserve the original art style while enhancing realism.</li></ul><h2 id="generation-guide">Generation Guide</h2><h3 id="prompts">Prompts</h3><pre><code>main theme: realistic, half realistic, &lt;empty&gt;,
hair: viscous hair, shiny hair,
skin: half realistic skin, shiny skin, plastic skin, realistic skin,
body type: young, &lt;-- for more age down</code></pre>This LoRA significantly affects the generation of the following characters. When generating them, it’s recommended to moderately adjust the model weight:<pre><code>castorice \(honkai: star rail\), castorice \(star rail live 2025\) \(honkai: star rail\), formidable \(azur lane\), ganyu \(genshin impact\), taihou \(azur lane\), taihou \(fun on the baize\) \(azur lane\), eyjafjalla \(arknights\), phoebe \(wuthering waves\), wakaba mutsumi, kafuu chino, shinosawa hiro, kasugano sora, yukinoshita yukino, </code></pre><h3 id="suggested-weights">Suggested Weights</h3>In general, you can use 1.0 weight for generation. If the art style becomes too strong or the anatomy looks abnormal, try lowering the model weight.<h2 id="training-details">Training Details</h2><ul><li>Hardware: Trained on RTX 4090 GPU</li><li>Loss curve: Overall downward trend; no major fluctuations observed between iterations in sample outputs</li><li>Number of images: 490</li><li>Epochs: 64</li><li>Steps: ≈ 10,000</li><li>Batch size: 6</li><li>Optimizer: AdamW 8-bit</li><li>Base Model: Creation NAIXL V1</li></ul>

TAGS:
realistic, half realistic, viscous hair, shiny hair, half realistic skin, shiny skin, plastic skin, realistic skin

Files with SHA256: 03a84f2f5f659413ca3107855e8f50d9333481005bf8f981379ac3733e30449d