Generating Media

In media mode, Musebot renders images, video, music, and audio from your prompts using ComfyUI. Describe what you want — the bot handles the rest and posts the result in the channel.

Making a Request

Send a prompt in a channel where the bot is allowed to respond (see Getting Started). What the bot generates depends on what your admin has set up — it may render images, videos, music, or audio.

Buttons

Every generation comes with buttons. Which buttons you see depends on the type of media generated and whether the generation is stateful (can be re-rendered with tweaks):

Button	What it does
🔄️ Retry	Re-renders with the same prompt for a different result.
➖ Decrease Guidance Scale	Loosens how strictly the render follows your prompt.
➕ Increase Guidance Scale	Makes the render follow your prompt more strictly.
{ } Show Source	Reveals the render's source JSON. You can paste it back as a prompt to customize the render.
❔ Help	Explains the buttons on that message.
📃 Expand Prompt	An LLM enriches your prompt first, then the render uses the expanded version.
🎲 Randomize	An LLM invents a fresh, unrelated prompt and renders that instead.

Depending on the generation type you may see fewer buttons:

Image and video generations (stateful): Retry, Guidance Scale ±, Show Source, Help, Expand Prompt, Randomize.
Music and audio generations (stateful): Retry, Guidance Scale ±, Show Source, Help — no Expand Prompt or Randomize.
Stateless image/video generations: Randomize and Help.
Stateless audio generations: Help only.

Show Source and Custom Renders

Click { } Show Source to get a .json file with the full render request — the prompt, seed, dimensions, sampler, and more. Paste that JSON back into the channel as a prompt and the bot will re-render it with whatever you changed: edit the prompt text, tweak the dimensions, change the sampler.

Tip: set the "seed" in the JSON to -1 and the bot will pick a random seed for the render.

Editing Images (img2img and img2vid)

Attach an image to your message (or reference a recent message with one) and you'll get buttons for each img2img/img2vid workflow your admin configured — for example, upscaling or style transfer. One render is produced per attachment. Each button is labeled with its workflow's name, exactly as the admin defined it.

Guidance Scale

Guidance scale controls how strictly the render follows your prompt: lower values give the model more freedom; higher values stick closer to your text. The ➕/➖ buttons step the value by an interval your admin configured, within fixed bounds (0–30). If a button would step past a bound, it's unavailable.