Project1998 · reports
Everything the mobs-page build surfaced beyond the page itself: era cross-reference counts, game-data nits, and sprite-renderer palette bugs from the side-by-side look check.
Side-by-side art comparison: the full look-check report — 356 suspicion-scored cards, our rendered sprite next to the NexusAtlas image.
| Mobs total | 715 |
|---|---|
| Matched on NexusAtlas | 479 — 350 on both the live site and the 2005 archive, 4 live-only, 122 archive-only (the live site deleted those era mobs, which is why the flag uses the union) |
Flagged no atlas | 236 — likely RTK/private-server additions |
| Mythic kits / boss-only | 72 / 9 |
| Rare-spawn pills | 4 |
| Ambush spawn provenance | 31 mobs |
| No rendered sprite | 53 — all look ids ≥ 617, beyond the 5.33 client table; none of them atlas-matched, so zero comparable mobs lack a sprite |
Catalogued during the docs mobs-page build; none of these were changed anywhere — they are for
the project owner's separate game-repo / render passes. Docs-side context: the union atlas cross-reference and
per-mob findings live in project1998-docs/tools/data/atlas_monsters{,_2005}.csv and the
side-by-side look report mob-look-check.html (this directory — 356 suspicion-scored cards).
722,rabbit_invasion_mob,Rabbot Invasion 4 — typo for "Rabbit Invasion 4" (its five siblings are all "Rabbit Invasion N").736,abyssal_rab_guardian,Rabibit Guardian — typo for "Rabbit Guardian".testxyz (MobId 730) — leftover test mob with a valid MobLook (0), so the server LOADS it and the docs page shows it. Either delete the row or make it unloadable like test is filtered docs-side. (test, MobId 0, line 2, also loads server-side for the same reason.)abyssal monkey1, abyssal_dog3, abyssal_oxbossf, abyssal_sheepguardian etc. (MobIds 738–791; exact list greps as Description islower/contains _). They render raw on the live page. Real names or a naming pass needed if this content ships.entries is empty (version 2, never populated by re/monster-matcher). Dead file; delete or populate.In-game looks are probably fine; the committed docs sprite sheet is what mis-renders. From the
scored look report (top ~20 cards):
palette_index % 33 over the DLPalette blocks.Colour bytes 11 (pale/albino family: ghost_mantis, albino_buck, ice_panther, pale_scorpion,
stout_pig, boa) and 29 (white family: white_chicken, white_rabbit, white_wolf, shadow_mantis)
render dark/green instead of pale/white. The 5.33 client's colour-byte → palette mapping for
values ≥33 needs real reverse-engineering.
diamond_horse and golden_horse (both look 17, colours 11/30) collapse to the same palette 3.
Needs (Look, Colour) keying or per-mob rows.
show a small dark shape where the atlas shows the big white yeti-tiger. Verify look 130's frame
index in the 5.33 MONSTER.DNA (possible id shift vs 4.95).
the 5.33 client's monster table — no sprite possible from this client; none are atlas-matched.
NexusAtlas pages deleted since 2005 usually still host their monster GIFs at the predictable URL
https://www.nexusatlas.com/photo/monster60/<lowercase-alphanumeric-name>[1].gif
(100/124 era-mob images recovered that way; Mod_Security blocks bare curl — send browser headers).
Spawns.csv also still carries the spy_hwan rows (mob 729) that the server filters at
load — cleanable in the same pass.