<h3 id="note:">note:</h3>出于未知的原因，v-pred模型不能被自动识别。如果有使用的需求，需搭配ComfyUI的ModelSamplingDiscrete节点使用，将采样模式设置为v_prediction并且将zsnr设置为ture。如果找到了解决方案我会在下个版本修复。For unknown reasons, the v-pred checkpoints cannot be automatically detected. If you need to use a v-pred checkpoint, you’ll have to pair it with the ModelSamplingDiscrete node in ComfyUI, set the sampling to v_prediction, and set zsnr to true. If I find a solution, I’ll fix it in the next release.<h1 id="or-quick-start">快速上手 | Quick Start</h1><h3 id="or-what-is-this">这是什么？ | What is this?</h3><ul><li>Pony: People's Works （ppw）是一个实验性的微调模型系列，数据集有约85%是收集自CivitAI上用户发表的AI生成图片。<a target="_blank" rel="ugc" href="https://civarchive.com/models/856285?modelVersionId=965209">早期ppw</a>的数据集最初建立在由<a target="_blank" rel="ugc" href="https://civarchive.com/models/257749">pony v6</a>生成图片的基础上，因此本系列模型生成的图片也带有pony diffusion的特征。</li><li>本系列模型使用标准Danbooru标签，主要擅长生成中、近景的风格化人像。它们的主要功能是使基础模型可以在不使用画师串、较少的质量提示词的条件下获得相对稳定的图像质量，为提示词节约token空间。 本模型并非风格模型，在不同的提示词和生成条件下可能会有微妙的画风差异。</li></ul><ul><li>Pony: People's Works (ppw) is a experimental fine-tuned model series, approximately 85% of the dataset comes from AI-generated images published by users on CivitAI. Since the <a target="_blank" rel="ugc" href="https://civarchive.com/models/856285?modelVersionId=965209">earlier ppw</a> dataset was built on images generated by <a target="_blank" rel="ugc" href="https://civarchive.com/models/257749">Pony V6</a>, the outputs of this series also carry some characteristics of Pony Diffusion.</li><li>This series uses standard Danbooru tags and is mainly optimized for generating stylized portraits at medium and close range. The primary effect of this model series is to allow the basemodel to achieve relatively stable image quality, without artist keywords or long quality tags, freeing up token space for prompts. These models are not style LoRAs. There may be subtle stylistic variations depending on different prompts and generating conditions.</li></ul><h3 id="or-version-info.">版本信息 | Version Info.</h3><ul><li>本页面发布的是ppw的基础模型版本。这些模型由<a target="_blank" rel="ugc" href="https://civarchive.com/models/833294/noobai-xl-nai-xl">NoobAI-XL</a>训练得来，首要功能是用于提取<a target="_blank" rel="ugc" href="https://civarchive.com/models/1400090/">LoCon版本模型</a>。除此之外还会被用作下一代模型某些特定概念训练数据的合成。</li><li>您也可以把这个模型当做生成模型独立使用，可以不加载LoRA独立生成图像，占用更少计算资源。相比LoCon版本，基础模型版本理论上对训练内容有更强的泛化性，但是与其他LoRA、LyCORIS等模型搭配的效果可能较差。没有针对眼睛、手部等细节做优化。</li><li>可以用于硬盘储存空间较大但是显卡性能有限的情形，或者需要更稳定的生成效果的用户。</li><li>请注意本模型生成内容不可用于商用。详情请阅读<a target="_blank" rel="ugc" href="https://civarchive.com/models/833294/noobai-xl-nai-xl">NoobAI-XL</a>原始许可证。</li></ul><ul><li>This page features the checkpoint versions of ppw. These models were trained from <a target="_blank" rel="ugc" href="https://civarchive.com/models/833294/noobai-xl-nai-xl">NoobAI-XL</a> and primarily served as the source for extracting the <a target="_blank" rel="ugc" href="https://civarchive.com/models/1400090/">LoCon versions</a>. They are also used to synthesize training data for certain specific concepts in next generation of the ppw models.</li><li>You can generate images independently without loading any LoRA. It can produce images independently while consuming fewer computational resources. Compared to the LoCon version, the checkpoints theoretically offers stronger generalization on the training content, but is less flexible in combination with other LoRA or LyCORIS models. It has not been specifically optimized for details such as eyes or hands.</li><li>It is suitable for situations where disk storage is sufficient but GPU performance is limited, or for users who require more stable generation results.</li><li>Please note that the outputs generated by this model are not permitted for commercial use. For more details, please refer to the original <a target="_blank" rel="ugc" href="https://civarchive.com/models/833294/noobai-xl-nai-xl">NoobAI-XL</a> license.</li></ul><a target="_blank" rel="ugc" href="https://civarchive.com/models/1400090/pony-peoples-works">高规格版LoCon | High-spec LoCon ver.</a> <a target="_blank" rel="ugc" href="https://civarchive.com/models/1774946/pony-peoples-works-sketch">精简版LoCon | Lightweight LoCon ver.</a><a target="_blank" rel="ugc" href="https://civarchive.com/collections/8769046">更多 | more</a><h3 id="or-usage">使用方法 | Usage</h3>positive:<pre><code>masterpiece, best quality, very aesthetic</code></pre>negative:<pre><code>low quality, displeasing</code></pre><h1 id="or-change-log">更新记录 | Change log</h1><h3 id="v8">v8</h3>肌理更新：强化了以下tag的学习：Texture Update: The following tags have been reinforced in training:<pre><code>realistic, photorealistic, flat color,</code></pre><pre><code>shiny skin, matte skin, shiny hair,</code></pre>请注意，在danbooru数据集中有很多个用于描述“照片”和“接近照片的风格”的tag。我在训练集中统一标注这些图片为“photorealistic”。但是使用danbooru训练集训练的SDXL模型大多并不能很好地画出写实图像，因此“photorealistic”只建议在较小权重下用于改变画面的肌理。“realistic”可以在高权重下正常运作。Please note that Danbooru dataset contains multiple tags to describe "photo" or "photo-like styles". I’ve tagged all such images as “photorealistic” in dataset.However, most SDXL models trained on the Danbooru dataset do not render realistic images well. “photorealistic” is only recommended at low weight, where it can help adjust texture rather than create realism images. The “realistic” tag can work properly at higher weight.<h2 id="or-dataset-source-and-license">数据集来源及许可证 | Dataset Source &amp; License</h2><ul><li>数据集中每一张图片都经过作者本人的人工筛选、分类和标注编辑，其中数百张图片经过手工编辑、修正。</li><li>此模型为免费、开源模型，用户可以在私人设备上自行部署该模型。作者并不从模型出售中获取任何报酬。作者并不限制本系列模型用于商业生成服务或者生成图像用于商业用途，但是请注意配合使用的Checkpoint和其他LoRA的许可证限制。</li><li>此数据集中约90%-95%数据为AI生成，但是请注意有约250+张收集自公共媒体、新闻媒体和出版物的图像用于概念补充。未来的版本会逐渐更替相关素材。请有商用意向的用户自行注意相关风险。本数据集没有训练任何独立画师的数据，也没有标注任何画师信息（不排除AI错误标注的情况）。</li><li>另外，本模型不允许用作闭源商用、模型出售，也禁止用于闭源商用模型的融合。对于开源融合模型用于生成服务的情形不做限制，但是建议标注融合模型的出处。</li></ul><ul><li>Every image in the dataset has been manually selected, categorized, and annotated by the author. Additionally, hundreds of the images have been manually edited and corrected.</li><li>This model is free and open-source model, allowing users to deploy it on their personal devices. The author does not receive any compensation from selling the model. The author does not impose restrictions on using this model for commercial image generation services or generating images for commercial purposes. However, please be mindful of the license restrictions of the Checkpoint and other LoRAs used alongside this model.</li><li>Approximately 90%-95% of the dataset consists of AI-generated images. However, around 250+ images have been collected from public media, news outlets, and publications to supplement concepts. Future versions will gradually replace these materials. Users with commercial intentions should be aware of the potential risks.This dataset does not include training data from any individual artist, nor does it contain explicit artist attributions (though AI mistagging cannot be entirely ruled out).</li><li>Additionally, this model is not permitted for use in closed-source commercial applications, model resales, or merged into closed-source commercial models. There are no restrictions on open-source merged models being used for image generation services, but it is recommended to credit the sources of any merged models.</li></ul>

Files with SHA256: 38ea7bb67bdb0cc278cc0142aa57c0b591d56ea11f12c006212d90eefa03d6d7