AtlasGB · Map data
The ledge in Viridian City
One scenario lost 17 of 600 attempts, every one of them on the same row of Viridian City, sliding between x=26 and x=32 with the leg's first waypoint at (29,26) unreachable above it. The first model of the trap was topological, and it was wrong.
A flood fill says (29,26) is fine. There is a 13-step path back to it crossing no warp tile. It reports nothing.
Two things had to be modelled exactly before any check could catch it.
The ledge. Stepping down off row 26 does not land on row 27. It jumps to row 28. And the one-wayness is in the cartridge's own data — eight ledge entries, for facing down, left and right, and none for up.
The walker. A greedy waypoint walker is not a path-finder. Below a one-way ledge it slides along the wall, which is precisely what the 17 failing runs did.
0123456789012345678901234567890123456789
y=25 ####....######........#.##..........####
y=26 ####....######......................#### <- the waypoint was (29,26)
y=27 ###############.###.#################### <- ledge; gaps only at x=15, 19
y=28 ...#................................#### <- where 17 runs slid, x=26..32
y=29 ...#.................#..............####
y=30 ...#................................####
(29,26) walkable=True step down -> (29,28) a two-tile jump over row 27
(29,28) walkable=True step up -> None nothing comes back
LedgeTiles: 8 entries, facings down/left/right, none for up
GoTo simulated below the ledge, target (29,26):
from (29,28) -> stops at (32,28)
from (30,28) -> stops at (27,28)
from (26,28) -> stops at (31,28)
The repair is not a bigger budget
The row the mart door opens onto crosses the city cleanly, sits six rows above the ledge, and the rest of the chain finishes even from below the ledge. So the leg turns west first, and an overshoot costs nothing.
This shape has now appeared three times, which is what makes it a rule rather than three anecdotes:
A waypoint whose overshoot is unrecoverable is a waypoint pointing at a trap. Prefer the turning point with slack around it.
The tool that checks it over-reports, on purpose
It cannot know whether an overshoot is reachable — that needs the map's people, and a person in the way is what pushed the walker over the ledge in the first place — and it has no model of the driver's stuck-sidestep. So it is a design aid, not a gate, and it says so.
The walkability grid and the ledge table were read off a retail Pokémon Blue cartridge by the repository's own audit tool, with no emulator run.
← Back to devlog