mirror of
https://github.com/bilawalsidhu/gods-eye-view.git
synced 2026-11-04 10:24:02 +00:00
fix(street-level): Panoramax IC8 round 2 — shared point budget, selected line, cone cache, DP segments
From the round-2 verification (approve with follow-ups), all follow-ups: - The point budget is shared across the tiles in view once loading settles (water-filling): a sparse tile keeps what it has, a dense city centre takes the rest instead of drawing thinner than its suburbs. The selected sequence is always drawn, and selecting one a cap had left out draws it. - The cones' picture-list cache keeps only finished lists: a re-centre while the cones loaded inherited the aborted load's tiles and drew none in them. - Douglas–Peucker measures to the segment, so dead ends and U-turns keep their far end. - Switching Panoramax off lets its decoded tiles and lists go. - Picture downloads are counted while they stream (a host sending no length cannot fill memory before the cap). - Comments on the via caption and the simplification tolerance match the code. Co-Authored-By: Claude Opus 5.5 <[email protected]>
This commit is contained in:
1 parent
1a546bec72
commit
3ae367d0dd
11 files changed
+343
-33
No files matched your search
+10
-1
@@ -3256,7 +3256,16 @@ IGN, MapComplete…; IGN's redirect to its OVH object storage). Its chip
|
||||
vector tiles (`/api/map/{z}/{x}/{y}.mvt`, an hour's HTTP cache): sequence
|
||||
lines from z14 below 1.8 km above ground, z13 to 5 km, z12 to 12 km, z11
|
||||
to 30 km and z10 to 60 km, as Mapillary's (a dense z10 tile over Paris is
|
||||
~2.1 MB and 8,600 sequences; each tile draws its 1,600 longest), and above
|
||||
~2.1 MB and 8,600 sequences, 750,000 points). `decode.js` simplifies each
|
||||
line (Douglas–Peucker to the segment, 4 units of a 4,096 tile: about a
|
||||
pixel where it is drawn, 6–10% of a GPS track's points), and the tiles in
|
||||
view draw their longest lines within 90,000 points (`maxPoints`,
|
||||
`keepOrder`), shared out once loading settles so a sparse tile keeps all
|
||||
it has and a dense one (a city centre) takes the rest; the selected
|
||||
sequence is always drawn. Every point becomes ~8 ground-line vertices:
|
||||
unsimplified, the lines hung the tab. Decoded tiles are kept while the
|
||||
provider is on (16, ten minutes, `tileCache.js`, shared with KartaView),
|
||||
and above
|
||||
60 km the `grid` layer's 0.1°
|
||||
density cells, binned to at most 96×96 points a tile (a z0 tile holds
|
||||
~70,000 cells) and not counted as sequences: the panel says to zoom in. Both
|
||||
|
||||
Reference in new issue
Block a user