Batch runs

Clicking one opens a checklist first: the shots it can actually do are listed, you tick the ones you want, then pick a model, an aspect ratio and the rest. Each of the three has its own bar and its own settings — storyboard and video list only shots whose description is filled in (the video one also lets you set a one-off duration per shot), while dubbing lists only shots with dialogue and swaps the settings for voices. If you open it from the toolbar after ticking rows, the ticked shots are already checked for you.

That sheet also says what this batch will cost: the AI compose entry gives a range (estimated shot by shot and added up), because it settles on the model's real input and output tokens and carries reasoning output; Join fields costs nothing, so no figure is shown for it.

Shots you have already done are unticked by default, marked Generated at the end of the row. To redo one, tick it by hand. This is so you can work in batches: do three shots, see whether the direction is right, then fill in the rest.

The batch lands on the canvas edges and all, and is fenced automatically into one group. From there:

• Run group on the group toolbar queues up every node in it that is waiting to generate and runs them through (see 2.3).
• While it runs, each node lights up a Running badge to the right of its title, and fills its content in when it finishes.
• Unhappy with one shot? Change that one and rerun it on its own — no need to redo the group.