<h1 id="moss-image-character-and-style-catalog-prompt-library-and-local-browser">MOSS-Image Character &amp; Style Catalog — Prompt Library and Local Browser</h1><h2 id="moss-image-character-and-style-catalog-v0.1.0">MOSS-Image Character &amp; Style Catalog v0.1.0</h2>Browse 33,937 character images and 221 style examples — 34,158 images in total, with generated previews, reusable MOSS-Image prompts, original tags and saved workflows.Full download: approximately 2.05 GB / 1.91 GiB.This is a local catalog application and prompt library. Model weights are separate downloads.The interface follows the familiar Krea2 Character &amp; Style Catalog layout, with English and Japanese available.<h2 id="features">Features</h2><ul><li>Search and filter characters and styles.</li><li>Browse by copyright, appearance tags and style category.</li><li>Sort by name, copyright count, or generation order — newest or oldest first.</li><li>Enlarge images and copy reusable prompts or original tags.</li><li>Save favorites, rate entries, and hide unwanted images.</li><li>Add your own images and edit their names, prompts, tags and workflows.</li><li>Optionally compare images with your own local AnimaDex references.</li><li>Review MOSS style images even without reference images.</li><li>Export confirmed review decisions as JSON and import them into the catalog.</li></ul>Comparison judgments are made by the user. Importing confirmed decisions is a separate operation, with a preview and image-hash checks before applying changes.<h2 id="what-is-included">What is included?</h2>The collection is based on an Anima-oriented character list, expanded with additional candidates, outfit variants and tests.<ul><li>33,937 character images, including 247 additional character images beyond the retained source-list entries.</li><li>221 styles, covering visual media, painter-inspired descriptions, traditional art and animation styles.</li><li>Reusable prompts, original tags and per-image workflow JSON files.</li><li>Two portable ComfyUI sample workflows:<ul><li><code>MOSS_Image_Character_Sample.json</code></li><li><code>MOSS_Image_Style_Sample.json</code></li></ul></li></ul>Outfits, forms and test images count as separate entries. These totals are image counts, not counts of unique characters or artists.<h2 id="about-the-results">About the results</h2>This release has not been selected by pass/fail judgments.Incorrect character identities, mismatched prompts and weak style results remain included. Use the previews to judge which results and prompts are useful for your own work.Resemblance does not prove accurate reproduction or confirm that a character or artist appeared in the model’s training data.The catalog is a practical reference for exploring MOSS-Image’s behavior. Changing the checkpoint, quantization, LoRA, steps, CFG, prompt or seed can change the result.Corrections are welcome. Please include the entry name or ID and catalog version when reporting a problem.<h2 id="how-the-previews-were-generated">How the previews were generated</h2>The previews were generated with MOSS-Image-SFT in ComfyUI.SettingCharactersStylesDiffusion modelSFT INT8 ConvRotSFT BF16Acceleration LoRA8-step LoRA, strength 1.0NoneSteps825CFG1.02.5Text encoderQwen3.5-9B INT8 ConvRotSameVAEFlux2 initially; Small Decoder laterSmall DecoderSampler / scheduler<code>er_sde</code> / <code>simple</code>SameSeed<code>8961184</code>SameResolution896 × 1184SameCatalog image formatWebP quality 70SameThe accelerated character workflow made a large collection practical to generate. It should not be treated as a maximum-quality benchmark of unaccelerated SFT or BF16.Style prompts describe visual characteristics in English and use a shared flower-workshop scene to make differences easier to inspect.Generation hardware: RTX 5060 Ti 16 GB, Ryzen 9 3900X, 64 GB RAM, Windows 11 Pro. This is the rendering environment, not the minimum requirement for browsing.Full settings, model download links and workflow instructions are documented in the <a target="_blank" rel="ugc" href="https://github.com/ukr8b3g-cmyk/MOSS-Image-Character-Catalog#readme">README</a>.<h2 id="download-and-links">Download and links</h2><ul><li><a target="_blank" rel="ugc" href="https://github.com/ukr8b3g-cmyk/MOSS-Image-Character-Catalog/releases/download/v0.1.0/MOSS-Image-Character-Catalog-v0.1.0.zip">Download the complete v0.1.0 ZIP</a> — approximately 2.05 GB / 1.91 GiB</li><li><a target="_blank" rel="ugc" href="https://ukr8b3g-cmyk.github.io/MOSS-Image-Character-Catalog/">Live demo</a> — 120 samples: 96 characters + 24 styles</li><li><a target="_blank" rel="ugc" href="https://github.com/ukr8b3g-cmyk/MOSS-Image-Character-Catalog">GitHub</a></li><li><a target="_blank" rel="ugc" href="https://github.com/ukr8b3g-cmyk/MOSS-Image-Character-Catalog/releases/tag/v0.1.0">Release notes and checksums</a></li><li><a target="_blank" rel="ugc" href="https://github.com/ukr8b3g-cmyk/MOSS-Image-Character-Catalog/blob/main/README.ja.md">日本語README</a></li></ul>The live demo displays only a small subset of the collection. It supports browsing, searching, enlargement and prompt copying. Download the full ZIP for image registration, editing and review features.GitHub’s Code → Download ZIP and automatically generated Source code archives contain the demo and documentation. Choose the named catalog ZIP above for the complete application and image collection.<h2 id="getting-started">Getting started</h2><ol><li>Download and extract the complete ZIP.</li><li>Keep the extracted folder together.</li><li>Open the START_HERE file for your operating system.</li></ol>Python 3.10+ and a web browser are required. A GPU is not required to browse the catalog. Initial dependency setup may require internet access.Windows, macOS and Linux launchers are included. Windows operation has been checked; physical macOS/Linux desktop testing has not been performed for this MOSS package.The completed ZIP passed CRC and SHA256 verification for all 136,691 files, and all 34,158 image hashes matched the catalog metadata. These checks verify file integrity, not visual accuracy.Model weights, reference image libraries, personal settings and review verdicts are not bundled. Image generation is performed separately in ComfyUI.

Files with SHA256: 9b0bd22dc4930eb0842059548daf98831a1894ff960ecc086181547843c125a0