Text nodes

Text nodes hold words: ideas and prompts you write yourself, or text the AI generates or reads out of an image or audio clip. They often feed image and video generation downstream.

What they're for

Text plays two main roles on the canvas. It can be a prompt or script connected to downstream image or video nodes, where it takes part in their generation. Or it can be the result of a text generation task itself (the AI continuing or rewriting text).

Create and generate

• Choose “Text” in the “Add node” menu to create a text generation node, then type in it or let the AI continue or rewrite the text;
• Connect a text node's output to an image or video node, and its text is combined with that node's own prompt for generation;
• What you write in a text node is never machine-translated. It stays exactly as you wrote it.

Reading images and speech to text

Open the “Text” group from the “+” beside a neighboring node to create nodes whose result is text:

• “+” beside an image node → “Text”: creates a text generation node that can read the image, for example to describe the picture or write a prompt for it;
• “+” beside an audio node → “Text”: speech to text, which pulls the words out of an audio clip.

Both are paid generations, and you see the credit estimate before you create them.