<h1 id="prompt-generator-anime">Prompt Generator Anime</h1><p>Expands a few Danbooru-style seed tags into a full image generation prompt.</p><h2 id="the-model-(click)"><a target="_blank" rel="ugc" href="https://huggingface.co/TRYZER01/Qwen3.5-4B-Danbooru-Prompt-Generator">The Model (click)</a></h2><h3 id="how-it-works">How it works</h3><p>Send a few seed tags as input. The model expands them into a complete prompt.</p><p><strong>Supports semantic negation with -tag (for example -nude).</strong></p><p>Examples:</p><pre><code>1girl, blue_hair</code></pre><pre><code>1girl, long_hair, school_uniform, -nude, rating:general</code></pre><pre><code>2girls, yuri, rating:explicit</code></pre><p></p><h3 id="tag-format-rules">Tag format rules</h3><ul><li><p>Separate tags with comma + space: 1girl, long_hair, blue_eyes</p></li><li><p>Multi-word tags use underscores: looking_at_viewer</p></li><li><p>Rating tags must be written exactly as:</p></li></ul><pre><code>rating:general</code></pre><pre><code>rating:explicit</code></pre><ul><li><p>Prefer real Danbooru tags for best results</p></li></ul><p></p><p><strong><u>Note: The model was trained mostly on NSFW data and has a strong bias toward explicit content. For SFW results always add rating:general and negative tags like -nude.</u></strong></p><p></p><h3 id="comfyui-installation">ComfyUI Installation</h3><ol><li><p>Copy the comfyui-prompt-generator folder into:</p></li></ol><pre><code>ComfyUI/custom_nodes/</code></pre><ol><li><p>Install requirements (use ComfyUI's Python):</p></li></ol><pre><code>python -m pip install -r custom_nodes/comfyui-prompt-generator/requirements.txt</code></pre><ol><li><p>Restart ComfyUI.</p></li></ol><h3 id="node-usage">Node usage</h3><ul><li><p>Put seed tags into the tags field</p></li><li><p>Connect formatted_prompt to your CLIP Text Encode</p></li><li><p>Adjust prefix / suffix if needed for your image model</p></li></ul><p>The node automatically converts:</p><p>rating:general → safe</p><p>rating:explicit → nsfw, explicit,</p><p>Make sure the server is running and accessible, then set in the node:</p><ul><li><p>host: address of your API</p></li><li><p>model: the model name/key that your server uses</p></li></ul><p></p><h3 id="how-to-run-the-model">How to run the model</h3><p>You can run the GGUF model in any backend that supports the following endpoints:</p><ul><li><p>/v1/chat/completions</p></li><li><p>/v1/models</p></li><li><p>/v1/models/load</p></li></ul><p></p>

Files with SHA256: 1b708a04b57a6ee8970a2b834b53c54b444f9e8c901cb251d14b076849da5226