chartsAndDataViz
Two layer alignment by correlation
Generates two canvas layers of the same seeded scene, pulls the second one apart by a known displacement and adds noise plus a brightness jump, then finds that displacement on its own with normalized cross correlation at two scales: a coarse sweep over the downsampled layers and a full resolution refinement around the winner. It paints both correlation maps as heat grids with the peak marked against the true offset, overlays the layers tinting their difference, and counts how many comparisons it saved against brute force.
See it live and copy its promptWhat you can adjust
- Real horizontal offset
- Real vertical offset
- Added noise
- Brightness difference
- Search radius
- Coarse scale reduction
- Accent color
- Canvas height
- Initial overlay
How this catalog works
Every component ships with its reproduction prompt: the text you paste into an AI builder to get exactly this piece. The values above are editable, and the prompt you copy carries the ones you chose, so what gets generated is what you were looking at.
