<h1 id="moss-bbox-prompter-for-comfyui">MOSS BBOX Prompter for ComfyUI</h1>A visual bounding-box editor for MOSS-Image. Place characters, objects, and text on the canvas, then give each box its own prompt.<h2 id="what's-new-in-v2">What’s New in V2</h2><ul><li>Image Analysis writes a Character or Object prompt into the selected box from a reference image.</li><li>Keep &amp; Edit lets you switch analysis OFF and edit the generated prompt. A successful analysis run with it ON can replace the text in the selected box.</li><li>Target Selection lets you use Auto or choose a specific box from the subgraph controls.</li></ul><h2 id="installation">Installation</h2><h3 id="install-the-custom-nodes-with-git">Install the custom nodes with Git</h3>Open a terminal in your ComfyUI folder and run:<pre><code>cd custom_nodes
git clone --branch v2.0.1 https://github.com/ukr8b3g-cmyk/MossImage-BBOX-Prompter.git</code></pre>If rgthree-comfy is missing, install it from the same <code>custom_nodes</code> folder. It provides the workflow’s Power Lora Loader.<pre><code>git clone https://github.com/rgthree/rgthree-comfy.git</code></pre>Restart ComfyUI and refresh your browser.<h3 id="2.-download-and-open-the-v2-workflow">2. Download and open the V2 workflow</h3>Download the ZIP from this Civitai page and extract it. It contains one workflow (<code>MOSS_BBOX_Canvas_V2.json</code>) and one reference image.Open the JSON in ComfyUI, select your installed model files, and load the included reference image or one of your own. Image analysis starts OFF. Turn it ON when you want to generate a prompt for the selected box.<h2 id="model-files">Model Files</h2><ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/MOSS-Image/blob/main/diffusion_models/moss_image_sft_int8_convrot.safetensors">MOSS-Image SFT — ComfyUI INT8</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/MOSS-Image/blob/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors">Qwen3.5-9B INT8 Text Encoder</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/black-forest-labs/FLUX.2-small-decoder/blob/main/full_encoder_small_decoder.safetensors">FLUX.2 Small Decoder VAE</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/lodestones/MOSS-Image-SFT-8-steps-dmd">MOSS-Image SFT 8-Step Acceleration LoRA</a></li><li><a target="_blank" rel="ugc" href="https://civarchive.com/models/3020298/mossluc">MOSS_LUC — Alternate Anime Model</a></li></ul><h2 id="main-github-repository-custom-nodes-v2.0.1"><a target="_blank" rel="ugc" href="https://github.com/ukr8b3g-cmyk/MossImage-BBOX-Prompter">Main GitHub repository</a> · <a target="_blank" rel="ugc" href="https://github.com/ukr8b3g-cmyk/MossImage-BBOX-Prompter/releases/tag/v2.0.1">Custom nodes v2.0.1</a></h2>

Moss Image  BBOX Prompter

### What’s New in v2.0- Image Analysis: Automatically create a Character or Object box prompt from a reference image.- Keep &amp; Edit: Switch analysis OFF to keep the result and edit it freely.- Simple Targeting: Use Auto or select a specific box, including inside subgraphs.- Updated V2 Workflow: Existing V1 workflows remain compatible.Note: Running analysis ON again overwrites the edited prompt.

Files with SHA256: b300efe45581ddcd463056c490a75bb0c5040c3c424d7af0b15fab0a9811480a