Text nodes

Four nodes work with text instead of pictures. Each has a purple text output on its right. Wire that output into the prompt input on an Image Gen or Video Gen node, and the wired text replaces what is typed in that node's own prompt box.

The Prompt node is a text box. Add a variable in the Inspector and it gets its own input. Type @ and the variable's name in the text, and whatever text is wired into that input is put there when the graph runs. The name must match exactly, capital letters included. An @ name in the text that is not a variable stops the run, so check the spelling. Seedance's own names, such as @Image1, are left as they are.

The Prompt Joiner puts several pieces of text together, in the order of its inputs, with a space or whatever you choose between them. Add an input from the Inspector when you need more.

The Prompt Enhancer rewrites the prompt wired into it. Its instruction is shown in the Inspector, and you can change it. Press Enhance to run it.

The Describe node looks at an image or a video wired into it and writes a prompt that describes it. For a video it looks at three frames. Press Describe to run it.

1. Add a Prompt node and type your prompt.
2. Wire its output into the prompt input on an Image Gen or Video Gen node.
3. Run the generation. It uses the wired text.

Prompt and Prompt Joiner are free. Prompt Enhancer and Describe cost a small number of credits each run, like the Assist button, and nothing on your own Anthropic key.

If you change the text going into a Prompt Enhancer or Describe after it has run, its answer is out of date. Anything wired from it waits until you run it again.