FulGerNet / rePOP

# reSOT
Static recompilation of the original Xbox builds of
**Prince of Persia: The Sands of Time** into native Windows executables
and one native Android app for both games (device validation in progress).
*The Xbox retail game, recompiled, at 16:9 (Hor+), from one run
(2026-10-17) through the first areas (`0101_Porte`, `0102_Rempart_Couloir`):
a cutscene on the ramparts, the Vizier's guard, the courtyard, the wall-run
and ladder tutorials, two fights and the pause menu. Captures of the game
window; inputs from the scripted pad (`RECOMP_PAD_SCRIPT` / `RECOMP_PAD_LIVE`),
not a person. In the two cutscene frames the F5 panel, which sat on the
letterbox bar, was painted over in black.*
*The Xbox demo, recompiled, at 16:9 (Hor+), from one scripted 185-second
run (2026-10-14): main menu, the intro, the first level's opening, the
soldiers' scene, the ramparts under fire, a ledge prompt, the transition into
the next section. Frame dumps from the game window; inputs from a pad script,
not a person.*
| Target | Source executable | State |
| --- | --- | --- |
| `xbox-demo` | E3 2003 demo, `default.xbe` built 2003-07-18 | Runs movies, menus, pause and the scripted level sections at about 58-60 FPS with VSync; played by a person to the third level without a freeze. Compact F5 and F6 options added. User confirms corrected audio and 120+ FPS (2026-10-18); intermittent visual defects remain under validation; no console comparison |
| `xbox-retail` | NTSC retail, title `0x5553001D`, `default.xbe` built 2003-10-20 | Bring-up: first playable level reached by the user and a scripted run. The invalid-refresh timestep that caused extreme slow motion is corrected: gameplay measured 9.003 real seconds / 9.008 simulation seconds with the fix. User confirms corrected audio, 120+ FPS and the save-crash fix (2026-10-18); full-level accuracy and broader audio/visual validation remain incomplete |
The existing post-processing quality selector now budgets auxiliary effect
buffers and projected shadows as well as recognized blur/bloom chains, on
both hosts and targets. The user confirms that it works (2026-10-20).
The effect study records the remaining water,
motion-blur and other effect-identification gaps. Windows native renderers
also synchronize SDL window borders when F11 leaves fullscreen; that separate
correction still awaits confirmation. See VALIDATION.md.
The log viewers have a source-selection repair: Windows follows the logger's
actual file path, and Android reads the `logs/` subfolder. Output from these
repairs still awaits user confirmation.
**Both targets remain under gameplay validation.** The user reports
angle-dependent object flickering in the demo's first level and a crash at
the garden lever, and a retail crash at the second ladder of its first level
(2026-10-16). The user reports the save crash resolved (2026-10-18);
full save/load validation remains incomplete.
Build and runtime observations are recorded in VALIDATION.md.
The goal is both games fully pl