Why draw shots on one image
When an image model draws four shots in one picture, it keeps lighting, costume and faces consistent between them more easily than across four separate generations. The grid is then cut into cells and each cell starts a clip. It also makes review faster: a sequence reads at a glance.
Chained grids
In a chained grid, the cells are boundaries between shots rather than the shots themselves. Cell 2 is where shot 1 ends and shot 2 begins; the video model gets cell 1 as the first frame and cell 2 as the last. Consecutive clips then join on the same frame, which makes cuts smooth and keeps continuity details in place.
2x2, not 3x3
| Grid | Result |
|---|---|
| 3x3 (nine cells) | Cells drifted off the cut lines, so frames were cropped wrongly and needed fixing |
| 2x2 (four cells) | Clean cells; every four shots became one grid |
With 972 clips rendered for BROODFALL, a clean cut line on every grid mattered more than fitting more shots per image. Larger cells also give the video model a higher-resolution starting frame.
When to use grids
- Action and fast sequences where shots must flow into each other.
- Long productions where reviewing one frame per shot is too slow.
- Not for isolated shots, or when one shot needs a very specific composition; give those their own frame.
A 2x2 grid costs one image (13 credits in Arc) for four shots, about 3 credits a shot, against 136 credits for a 6-second 720p clip on Seedance 2.0 Fast.
Where Katechon Arc fits
Arc generates grid storyboards, including chained grids, and sends each cell to video as a first frame.
Grids are a planning aid, not a substitute for references; a project without approved character sheets will still drift across grids.