<h2 id="introduction">Introduction</h2>Scene Composer is a worklfow for <a target="_blank" rel="ugc" href="https://github.com/comfyanonymous/ComfyUI">ComfyUI</a> that let you generate random stories in just one run. The system roll parts of the prompt that are used for generation, like:<ul><li>Render: light, artstyle, etc.</li><li>Character: body, hair, eyes, attitude, etc.</li><li>Clothes: Dress, swimsuit, underwear, etc.</li><li>Action: SFW/NSFW, position, gestures, etc.</li><li>Environment: indods, outdoor, background, etc.</li></ul>To maintain consistency, it also keeps certain parts of the prompt to inject across all scenes (like the environment, the main character and their clothes). You can read my <a target="_blank" rel="ugc" href="https://civarchive.com/articles/5760/scene-composer-overview-and-usecases">Overview &amp; Usecases</a> article for more explanations!If you're looking for a simpler workflow, check my <a target="_blank" rel="ugc" href="https://civarchive.com/articles/5366/comfyui-workflow">Main ComfyUI Workflow</a>. I also suggest to have a look at my <a target="_blank" rel="ugc" href="https://civarchive.com/articles/3160/prompt-notebook">Prompt Notebook</a> to better understand how I structure tags.If you have any comment or request, please feel free to share!<h2 id="features">Features</h2>Note: features in gray are in progress and need to migrate to v2.x<ul><li>Random procedural generation of prompts<ul><li>Main character (e.g. body, hair, eyes, fantasy traits, …)</li><li>Attire (e.g. clothes, swimsuit, underwear, uniform, accessories.…)</li><li>Environment (e.g. place, weather,…)</li><li>Action in scene (e.g. starting scene, sexual encounter, ending scene,…)</li><li>Fantasy characters (demon, goblin, furry, slime, etc)</li><li>Famous personas (princess peach, hatsune miku, etc)</li><li>Male partner or tentacles </li></ul></li><li>Control scenes parameters<ul><li>Seed, steps, CFG, image size, etc.</li><li>HighRes-Fix (2nd Pass)</li><li>LoRA stacker</li></ul></li><li>Keep control over scenes<ul><li>Re-generate one or many elements by changing its seed</li><li>Re-generate one or many scene by changing its seed</li><li>Overwrite and compose the final prompt with variables</li></ul></li><li>Scenes consistency<ul><li>Persona automatically switch with their uniform (<code>&lt;if:persona-clothes&gt;</code>)</li><li>Tags update dynamically according to the scene (e.g. <code>wet</code> is added if there's rain)</li><li>Attire state stay across scenes (if character lose clothes, it stay lost)</li><li>If clothes are torned, they stay so</li><li>Bondage ropes stay on character, with clothes</li></ul></li><li>Output<ul><li>Upscaling, pre-processing</li><li>Images scene merged into one</li><li>CivitAI metadatas &amp; workflow embedded</li></ul></li></ul><h2 id="setup">Setup</h2>Simply import the <code>workflow.json</code> file attached to this article in ComfyUI.I personally don't have a very powerful computer. For people in the same situation, check my environment on my <a target="_blank" rel="ugc" href="https://civarchive.com/articles/5366/comfyui-workflow">Main ComfyUI Workflow</a> article, I explain how I rent and setup machines on remote.<h3 id="models">Models</h3>In theory, the workflow can work with any models that use Danbooru-like tags. I personally use Illustrious/NoobAI/Pony-based models and mix with some anime-oriented LoRAs. Have a look at <a target="_blank" rel="ugc" href="https://civarchive.com/user/taches/images">my last images</a> metadatas if you're curious!

Scene Composer: Workflow

<ul><li>Cleaner version with a loop system to avoid bloating the workflow.</li><li>Every scene's prompt are grouped</li><li>All generations happen at the same place</li><li>Images are upscaled and merged to a single image</li><li>Famous personas available as character</li><li>Tentacles available as partner</li></ul>

Files with SHA256: cb37095aa450ac501ded0b914496d7fd09b24d3d851d49bea196f63c046d0eb9