Guide
English
Search the docs⌘K
2.2

Text nodes

AI video creation runs on words: a line of narration, some copy, a prompt meant to be fed to another node. DOTV's text node is the card that holds those words, and it's the only kind of node on the canvas you can write yourself — everything else has to be generated. This section covers what it looks like, the two ways to write one, and how to wire it into an image or video node as an input.

What a text node looks like

One with contentThe body renders as Markdown: headings, lists and quotes all work. Long content scrolls inside the card.
1 / 2
On the left one that already has content, on the right one just created and still empty.
  • The card is a little narrower than an image node and its height is fixed; longer content scrolls inside the card.
  • The body renders as Markdown: headings, lists, quotes and code blocks all work.
  • Select a text node that has content and a Download text button floats up above the canvas.

With it selected, the scroll wheel scrolls the content inside the card and doesn't drag the whole canvas along; to zoom the canvas it's still ⌘ plus the wheel.

Two ways to write

Have it generated. Select the empty node, the input bar floats up at the bottom, and you spell out what you want ("write 15 seconds of product narration, keep it conversational") and submit. This route takes references: wire another text node in and its content is material for this generation.

Write it yourself. Hit Write it yourself on the empty node and the card turns into an editor on the spot, cursor already in it.

Writing it yourselfHit Write it yourself and the card turns into an editor on the spot; select some text and a formatting bar floats up (heading / bold / list / divider).
The "Write it yourself" line on an empty node, which turns into an editor in place.

Select some text and a formatting bar floats up, left to right: Clear formatting, Heading 1 / 2 / 3, Body, Bold, Italic, Bullet list, Numbered list, Divider. Click an empty part of the canvas when you're done; the content is saved with the canvas.

Feeding it to other nodes

The most common use of a text node isn't being read itself, it's feeding the next one:

  • Wire it to an image node → this text takes part in that generation's prompt.
  • Wire it to a script node → the model works backwards from your screenplay to a shot sheet (see Chapter 6).
  • Type @ in Agent's input box to reference it directly.

How to wire things up, and how a whole group runs once wired, is in Chapter 9.

Changing the size

A text node that has content can be dragged by its bottom-right corner to make it wider or taller, and the size is remembered afterwards. Drafts, nodes still generating and failed nodes don't have this handle; text nodes inside a group do. Grid slices don't get one either, since their size comes from the grid of the whole image and dragging one alone would scramble the group. Script nodes always have it (see 6.1).

The script node is its other face

The script node uses the same card, but what comes out of it isn't a passage of text, it's a shot sheet: the body of the card is a shot table with a progress line on top. With it selected, the toolbar lays out a whole group of shot images, a whole group of clips, a whole group of voice tracks in one hit, and you can also tick a few shots and create nodes for just those. All of that is in Chapter 6.

Try it in DOTVReading only goes so far: open DOTV and make an AI video with the feature this section covers.