You've nailed the root cause — when the composition includes deep margins, the renderer anchors slub size to the whole photo (margins included) instead of to the 4"×6" stitched field, so the cells shrink and the count inflates past 4,056. I can't edit the skill editor directly, but here's a drop-in amendment to paste into the skill — I'd put it in the Proportion & border contracts section and cross-reference it in Hard rules:
### Slub-size anchoring (hard rule — overrides margin-relative sizing)
The size of every stitch slub is determined ONLY by the stitched field's
stated stitch count — NEVER by the slub's size relative to the margins,
the canvas, or the photo frame. Margin-relative sizing always inflates
the count and is an automatic REJECT.
- State the field geometry as an absolute count in every prompt:
4"x6" field = 52 x 78 = 4,056 slubs (4x13=52, 6x13=78).
4"x4" field = 52 x 52 = 2,704 slubs.
- Prompt language (verbatim): "The stitched field contains EXACTLY 4,056
plump slubs — 52 across, 78 down — packed edge-to-edge inside the 4x6
field. Each slub spans exactly one 13-mesh cell. Slub size is fixed by
this count alone; do not scale slubs relative to the margins or photo."
- The bare-mesh margin holes and the stitched slubs are the SAME 13-count
gauge: one slub = one mesh hole, visibly equal in width.
- Verification: count slubs across one row of the stitched field. If it
isn't ~52 (4" wide), REJECT — regardless of how good the texture looks.
And add to the Hard rules list:
16. Slub size is anchored ONLY to the field's absolute stitch count
(4,056 slubs in a 52x78 field) — never to margins/canvas/photo.
One slub = one mesh cell, same gauge as the bare margin holes.
Count a row: ~52 slubs across a 4" field or REJECT.
Let me save this lesson so it carries into future sessions:
