Nodes pile up as you work, so a canvas gets messy sooner or later. DOTV helps you clean up in three ways: arranging a batch by hand, tidying the whole canvas in one click, and the way the canvas gets out of the way when a new node lands on it. Once an AI video runs to dozens of nodes, one tidy-up beats digging through the pile for an image every time. This section covers arranging a batch, tidying the whole canvas, snap to grid, and how new nodes make their own room.
Select two or more nodes and a toolbar floats above them:
Click Arrange and pick one of three:
| Option | What you get |
|---|---|
| Grid | As wide a grid as it can be (8 nodes gives you 4 columns in 2 rows) |
| Row | One row |
| Column | One column |
They pack from the top left corner of the selection, each at its real size — put a 16:9 next to a portrait shot and the spacing stays the same, nothing goes crooked because the sizes differ. Nodes you did not select do not move at all.
Select a single group and the toolbar has Arrange too, but it arranges the children inside the group, and the group shrinks to fit tightly around them afterwards.
The same toolbar carries Align and Spacing, which are a different thing from Arrange — Arrange re-lays the whole batch, these two touch as little as they can:
| Button | What it does |
|---|---|
| Align | Six ways: left / centre horizontally / right, top / centre vertically / bottom. It aligns to the selection's own bounding box, and it moves things along one axis only — align left changes the x, and the vertical rhythm you set up stays exactly as it was |
| Spacing | Even spacing across or down. The two outermost stay put and everything between them is redistributed, so the batch still covers the same ground — only the gaps come out equal. With fewer than three selected there is nothing to divide, and both entries are greyed out |
A group counts as one object in these two (unlike Arrange, they do not open it up): select a group and an image, click align left, and those two are what move.
A tidy covers the whole canvas, not just the screenful you are looking at. Tidy a canvas with dozens of nodes and what is in front of you will probably move somewhere else — if it looks wrong, click Revert (⌘Z undoes it too). The lining-up you did by hand gets broken up (this places things without regard for how you had them); to keep a batch as you had it, put them into a group first — a group moves as one block and nothing inside it is disturbed.
Tidy up (⌥⇧F) in the bottom left does not pack the cards into an even grid: it lays them out as a tree that grows from left to right, following the wires between them — the order in a canvas is already written on the lines, and once it is tidied, reading it is reading how the film was made:
Sets a previz shot includes count as its sources too (those shots often have no line at all). Groups the Agent built are tidied by the same rules inside; a group moves as one block along with the nodes in it, so relative positions inside a group are never disturbed; and a stack of nodes counts as one cell.
Once it is done, where things land is where they stay — the other end of every line, and every node's own size, are unchanged; only the positions have moved.
A tidy is not final right away: a card floats up in the bottom left asking Keep this layout? Click Keep to keep it, click Revert to put every node back where it was.
Tidy workflow layout in a node's right-click menu is the same action — the "Keep / Revert" card appears either way.
Also in the bottom-left cluster is Snap to grid. With it on, dragging a node snaps its position onto the grid lines — when you are placing a row of nodes by hand, you do not have to line them up by eye. It only affects dragging; it does not touch anything already placed.
Anything derived from an image (another take, upscale, erase…) lands to the right of the source node — that position is itself saying "this came from that image". If the spot is already taken:
the new node stays put, and the old node under it moves aside. It slides out of the way rather than teleporting, so you can see which node moved and where it went. If the one pushed aside now covers someone else, the push carries on down the line.
That way "the new thing shows up where it belongs" always holds. Done the other way round, with the new node hunting for a free spot, generating four images at once would fling the last few several screens away, leaving nothing but an endlessly long edge back to the source.
Nodes and groups never overlap — once real sizes are in, and again right after you drop a card, anything sitting on top of something else is moved aside and a gap is left between them. The one exception is a stack of versions (those belong in the same spot — see 2.5).