Theron original Akutuba completion capture — 2026-08-21
This source lock records a research-only original-engine transition against the authenticated US CUE. It does not authorize synthetic gameplay state.
The Mednafen hook enters the already-resident, byte-verified DMS-SG.001
dispatcher at its original $DE21 entry. It maps the exact System Card RAM
page $6D, rejects unless DMS-SG.001 and the $DE38 CMP #$06 bytes are
present, and replaces only A at $DE38 with bounded ordinal 0. Original game
code owns the subsequent CD command and RAM mutation.
Evidence:
- authentic savestate MD5:
f17f377df210b4a3ae904a13fb85a7f0 - initial 8 KiB BaseRAM MD5:
add6f5e61b849263f2a546164c08b496 - initial campaign byte
$267C:00 - original SCSI command: start LBA 4201, four sectors
- final 8 KiB BaseRAM MD5:
1b1fd5f634710c3ce9f3a388a0c30e24 - final campaign byte
$267C:01 - CD trace MD5:
c210f4675bcbbe689ad002d0c013cc7e
This proves that original post-Akutuba execution sets campaign bit 0. It does
not yet prove the later Drator bank load: $20DB remains 0 and $20DA remains
1 in the final snapshot. Generator RNG/spawn publication therefore remains
closed. The next capture must continue through the original restart/stage
selection until $20DB == 1, then enter Drator level 2 without substituting
host or generated dungeon bytes.
Original next-program selection and backup-RAM manager
The authentic savestate’s System Card RAM page $6D was extracted and the
resident DMS-SG.001 body disassembled. At $DE21 it first saves the old
campaign mask from $267C into $DF0E. The bounded ordinal path then updates
$267C, but $DEFB-$DF29 uses the saved old mask to calculate the four-sector
program record, loads it at $4000, and jumps to $4000. This explains why a
state with old mask 00 could set a completion bit without selecting Drator.
A research-only state changes only $267C from 00 to the already proven
Akutuba completion value 01. The original ordinal-1 dispatcher then selects
absolute LBA 4205. Its MODE1/2048 payload is the exact Track 02 byte range at
sector index 971, four sectors, MD5 9aa38417c25bd6d1717081d545fdcfa0
and FNV-1a 337de858. The System Card polling path cannot finish when entered
asynchronously from the research hook, so the capture tool may replay exactly
that hash-checked 8192-byte CD_READ result into the original $4000
destination and resume at the dispatcher’s original $DF29 continuation.
This is synthetic execution used only to derive real data; it does not provide
production gameplay data.
The resumed original overlay made nine further real CD requests: LBA
4229/2, 4233/17, 4257/1, 4994/4, 5002/7, 5014/1, 5015/1, 5017/1 and 4271/1.
The final VDC/VCE frame changed from the Akutuba dungeon to character artwork,
but disassembly of the exact LBA-4205 overlay rejects the earlier visual
interpretation that this was a Drator introduction. The program contains the
original file-cabinet strings CHOOSE A FILE TO DELETE, SURE?, and
THIS GAME WILL NOT BE SAVED, calls the System Card backup-RAM APIs, and loops
on controller state $2228. It is a backup-RAM manager in the post-dungeon
chain. This proves original next-program selection, not entry into Drator’s
dungeon or presentation. $20DA/$20DB remained raw values 01/00 and the
generator context count remained zero.
The capture wrapper now also emits the exact 2048-byte PCE backup-RAM snapshot.
An ordinal-0 run from the unmodified Akutuba state let original DMS-SG.001
execute its own LE051 write. The resulting real backup-RAM image has MD5
ffabc8d19b0915d4d9632a7ae2e90a97; its campaign payload byte at file offset
$20 is 01, and its System Card-managed header/check fields differ from the
input image. This is the authentic Drator-unlock save record and removes the
need to invent or repair a checksum synthetically. The same run issued the
original four-sector LBA-4201 request after saving.
Bounded ordinal sweep
The same authenticated state, System Card page, dispatcher entry and original
LBA 4201–4204 program were then run with bounded ordinals 1 through 6. The
completed snapshots establish the following campaign-byte mapping at $267C:
| Dispatcher ordinal | Final $267C |
Proven result |
|---|---|---|
| 0 | 01 |
sets bit 0 |
| 1 | 02 |
sets bit 1 |
| 2 | 04 |
sets bit 2 |
| 3 | 08 |
sets bit 3 |
| 4 | 10 |
sets bit 4 |
| 5 | 20 |
sets bit 5 |
Ordinal 6 is deliberately not promoted as a campaign-bit result. Its run
reaches the original CMP #$06 special branch but stalls after only two of the
four requested sectors, leaving $267C == 00. That is evidence of a distinct
final-stage path, not evidence that dungeon 7 has no completion state.
The ordinal-1 capture is the first direct original-engine proof that Drator’s completion identity is campaign bit 1. It still does not load Drator’s dungeon bank: the final bank bytes remain Akutuba-owned. Production therefore may use the bit mapping for campaign progression, but may not infer a Drator generator RNG/spawn sequence from this capture.