FluxionbySynthemo

Guide

Node mode

Node mode is a canvas of steps you wire together. Use it when one panel isn't enough: several models side by side, one image feeding many, or a whole pipeline you want to rerun.

The canvas

  • Add a node from Add Node in the header. Click a node to open its settings in the panel on the right; it's the same panel as Simple mode.
  • Drag from a node's output dot (right) to another node's input dot (left) to connect them. Only connections that make sense will join.
  • The gallery sits on the left. Drag either side panel's inner edge to resize it; double-click the edge to jump to half the screen.
  • Your canvas saves itself as you work and follows you to other devices. Flow saves and loads named copies.

The nodes

  • Text Prompt holds a prompt you can feed to several nodes. AI Prompt writes or improves one with AI.
  • Image Diffusion is every image model; connect images into it as references. Image nodes hold a picture to pass along.
  • Mask and Inpainting change only a painted area of an image.
  • Video Diffusion takes a prompt and a start (and end) frame. Video nodes hold a clip.
  • Story is the Story editor. Its storyboard can put shots, cast and locations on the canvas as linked nodes, so a shot's frame or a character's portrait can feed any other node.
  • Edit cuts clips together; wire videos into it.

Start from a recipe

A recipe is a small pipeline, already wired, with example prompts so you can run it straight away and change it from there. Pick one from Recipes in the header (a canvas with nothing wired yet offers them too). It's added beside whatever is on your canvas, never over it, and a card lists its steps until you close it.

  • Prompt → image → video: Write once, make an image, then animate it.
  • One prompt, two models: The same prompt on a fast model and a detailed one, side by side.
  • Three clips, one cut: Three short clips wired into an Edit, ready to trim and export.
  • Story to an edit: A premise into Story, and an Edit waiting for its clips.

Build one yourself

  1. Add a Text Prompt, an Image Diffusion and a Video Diffusion node.
  2. Connect the prompt into the image node, and the image node's output into the video node's start frame.
  3. Generate the image, then the video: the clip starts from the image you just made.
A thin bar along the bottom of a node shows it rendering. You can keep working, switch panels or reload: the result still lands on its node.

More

Everything a node does is explained on its own page: Images, Video, Story and Edit.