<h1 id="ctlf">🐱 CTLF</h1>作者发现市面上几乎所有的图像模型都无法生成猫咪不露出脚的“catloaf“姿态（GPT、Gemini等多模态模型除外），于是决定自己训练了这个模型，目前这个版本只适合生成猫咪，不支持和人合照。<h3>模型权重</h3><ul><li>基于juggernaut reborn训练，但适配各种风格的 checkpoint，推荐权重：0.9</li><li>如果生成的猫咪有脚课适当提高权重</li></ul><h3>触发词指南</h3>全局触发词：1cat, CTLF推荐结构：1cat, CTLF, 毛色，毛长，镜头方向，环境，风格，质量毛色：<ul><li>solid blue （纯蓝）</li><li>orange tabby （橘虎斑）</li><li>brown tabby （棕虎斑/狸花）</li><li>blue point （蓝点）</li><li>cream and white （奶油拼白）</li><li>blue and white （蓝白）</li><li>orange and white （橘白）</li><li>silver shaded （银渐层）</li><li>calico （三花）</li><li>silver tabby and white （银虎斑拼白/起司）</li></ul>毛长：<ul><li>shorthair</li><li>longhair</li></ul>数据集比较有限所以毛色还原不是很稳定，使用juggernaut_reborn效果会稍好<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/0efebd46-57a1-4c46-a1ee-b5041f218ec3/width=525/0efebd46-57a1-4c46-a1ee-b5041f218ec3.jpeg" /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/5e0b4efd-0e1b-452f-a3f9-ffa502ecff09/width=525/5e0b4efd-0e1b-452f-a3f9-ffa502ecff09.jpeg" /><h3>出图尺寸</h3>本模型基于 41 张网络素材及作者自己的猫咪“布墩”的实拍训练，数据集尺寸为：<ul><li><code>512×512</code></li></ul>只推荐直出512×512，避免崩图。<h3>高清放大</h3>推荐重绘幅度0.5，超过0.7会出现双头猫推荐：<ul><li>放大算法：<code>R-ESRGAN 4x+</code></li><li>重绘幅度 ≤ 0.6</li><li>建议启用 ControlNet Tile</li></ul><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/f1bb0182-23a6-42e4-84e7-08f8130fecff/width=525/f1bb0182-23a6-42e4-84e7-08f8130fecff.jpeg" /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/61c33171-331f-411d-a361-d888704ea4ff/width=525/61c33171-331f-411d-a361-d888704ea4ff.jpeg" /><h3>使用声明</h3>CUC AI STUDIO 出品欢迎大家使用这款 LoRA 自由创作！我们很期待看到你的返图作品~本 LoRA 训练所使用的数据集由以下内容混合构成：<ul><li>来自社交媒体平台的公开可获取猫咪图像</li><li>作者本人拍摄的猫咪图像</li></ul>训练数据仅用于学习猫咪“catloaf”姿势的一般性视觉特征与形态特征，并不旨在复现或指向任何特定个体动物或具体来源。本 LoRA 可用于个人及商业用途。使用者需自行确保其生成内容的使用方式符合相关法律法规、平台政策及第三方权利要求。除作者本人创作的内容外，作者不对训练图像主张任何所有权，亦不对生成内容中可能涉及的第三方权利作出任何保证。<hr /><h1 id="ctlf">🐱 CTLF</h1>I discovered that almost all image models currently on the market are unable to generate the “catloaf” pose in which a cat’s paws are completely hidden (with the exception of multimodal models such as GPT and Gemini). Therefore, I decided to train this model myself. This version is only suitable for generating cats and does not support images with humans.<h3 id="weights">Weights</h3><ul><li>Trained based on juggernaut reborn, but compatible with various style checkpoints</li><li>Recommended weight: 0.9</li><li>If the generated cat still shows paws, you can slightly increase the weight</li></ul><h3 id="trigger-guide">Trigger Guide</h3>Global Trigger:1cat, CTLFRecommended structure:1cat, CTLF, coat color, fur length, camera angle, background/environment, style, qualityCoat colors:<ul><li>solid blue</li><li>orange tabby</li><li>brown tabby</li><li>blue point</li><li>cream and white</li><li>blue and white</li><li>orange and white</li><li>silver shaded</li><li>calico</li><li>silver tabby and white</li></ul>Fur length:<ul><li>shorthair</li><li>longhair</li></ul>Due to the limited dataset, coat color reproduction is not always stable. Using juggernaut_reborn generally produces better results.<h3 id="output-resolution">Output Resolution</h3>This model was trained on 41 online images plus real photos of my cat, “Mop”.Dataset resolution:<ul><li><code>512 × 512</code></li></ul>Only direct output at 512 × 512 is recommended to avoid image collapse.<h3 id="hires.fix">Hires.fix</h3><ul><li>Recommended denoising strength: 0.5</li><li>Values above 0.7 may result in double-headed cats</li></ul>Recommended settings:<ul><li>Upscaler: <code>R-ESRGAN 4x+</code></li><li>Denoising strength ≤ 0.6</li><li>ControlNet Tile is recommended</li></ul><h3 id="disclaimer">Disclaimer</h3>Produced by CUC AI STUDIOFeel free to use this LoRA for your creative works — we’d love to see what you come up with! 💛This LoRA was trained on a mixed dataset consisting of:<ul><li>Publicly available cat images from social media platforms</li><li>Images of the author's own cat</li></ul>The training data was used solely to learn general visual features and characteristics of the “catloaf” posture, and does not aim to replicate or reference any specific individual animal or source.This LoRA is released for both personal and commercial use. Users are responsible for ensuring that their use of the generated outputs complies with applicable laws, platform policies, and third-party rights.The author makes no claims of ownership over the training images beyond those personally created, and does not guarantee the absence of third-party rights in generated content.

Files with SHA256: 5f31b7293d5366a8964cfcc790fbbff78e796ff9d12cb34df25e52c2580edf0d