A feature on ground below the sea level casts no shadow - #54
Conversation
The only water test in the original's shadow pass turns out to be on map features. At 0x4592D5 the test is skipped for any unit whose type index is non-zero, and type index zero is the Feature Unit that draws the map's features: for those, and only those, ground below the sea level byte casts nothing. A real building is never tested, and a finished unit's copied shadow is not cut at the water line either -- the cut at 0x4BA1B0 needs a height plane a finished bitmap does not carry. So a wreck on the sea bed or a rock in the shallows has no shadow, and everything else is as it is on land. RWE drew every feature's shadow, lifting a floating one to the surface. It now skips a feature whose ground is below the sea level. The notes' reading of that test as a building-on-water rule is corrected, and the upstream request to cast shadows on the water surface is recorded in S:100 as considered and declined: it is not what the original does. Issue: #9 Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PJCbwC9MGZnM6erMDKptun
|
Merging. It builds and passes on MinGW Debug and Release merged onto current One thing for a later look, not this PR. A floating building's shadow is still lifted to the water surface by 🤖 Generated with Claude Code |
|
Settled, in favour of your reading: the shadow goes at the ground under the building, wherever that ground is. The projected pass has no water test in it, and the one water test in the shadow pass is the Feature Unit's, which withholds a shadow rather than moving one. So the lift is gone. Committed locally on 🤖 Generated with Claude Code |
Left over from the shadows-on-water work in #54. The unit shadow loop lifted a floating building's ground height to the sea level before placing its shadow, so a floating structure's shadow sat on the water under it. Nothing in the original does that. The projected pass a building takes has no water test in it at all: the one water test in the whole shadow pass is on the Feature Unit (section 3), which gives a map feature below the sea level no shadow rather than moving its shadow. So the shadow belongs at the ground under the building, wherever that ground is. A floating unit is a different case and is unchanged: it casts no shadow at all, the copied pass skipping canhover and floater alike (0x45957F). Recorded in the section 100 subsection #54 added. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018moxVoyHDBoHqkw8uqS3HX
2 participants
For #9, the third item (shadows on water, upstream #25), as answered: match the original.
The decode. The gate list in §3 read the water test at
0x4592D5-0x4592F1as "a building below the water line casts none unless[unit+0xA6]is set".unit+0xA6is the unit's type index (§18), and the test is skipped when it is non-zero, so the rule is on type index zero only: the Feature Unit that draws map features. A feature whose ground (0x485070) is below the sea level byte casts nothing. A real building is never tested. And the water cut on a mobile unit's copied shadow (0x4BA1B0) lives only in the block taken when the bitmap carries a height plane, which a finished unit's does not (#40), so a finished unit driving through the shallows keeps its whole silhouette.What changed.
featureCastsShadow(groundHeight, seaLevel)inGameScene_util, used by the model-feature shadow loop: ground below the sea level, no shadow. RWE used to draw every feature's shadow and lift a floating one to the surface.compatibility.md, since nothing departs from the original.Verified.
noshadow.test.cpppins the strict-below rule;rwe_test: all 594 test cases pass (71755 assertions), Linux gcc-14 Debug.The other two items of #9 (3DO texture distortion, GAF off-by-one) are untouched here.
🤖 Generated with Claude Code
https://claude.ai/code/session_01PJCbwC9MGZnM6erMDKptun