Image-model pass (mocked: no image model is configured on this machine)

1. Control image (segmentation colours, from the data)

2. Line art (walls + props, for a lineart/canny ControlNet)

3. What would come back: a painted image, same size, aligned to the squares

4. In the game: the painted image is only a backdrop; grid, tokens, fog and cover stay drawn from the data on top.

The request it would send

Local / tactical maps · 2D

One 5-ft square = one cell. The generator writes data; this page only draws it. all prototypes · same map in 3D

Generator

seed

Style

Rules overlays

Click a square to select it (an attacker or a mover). Cover: half three-quarters no line of sight.

Stats

Placement rules used

Data (what would be stored)