/SpaceCadetPinball

Emscripten port of 3D Pinball for Windows – Space Cadet decompilation

Primary LanguageC++MIT LicenseMIT

SpaceCadetPinball

Summary: Reverse engineering of 3D Pinball for Windows – Space Cadet, a game bundled with Windows.

How to play: Place compiled executable into a folder containing original game resources (not included).
Supports data files from Windows and Full Tilt versions of the game.

Known source ports:

Platform Author URL
PS Vita Axiom https://github.com/suicvne/SpaceCadetPinball_Vita

Platforms covered by this project: desktop Windows and Linux.





Source:

  • pinball.exe from Windows XP (SHA-1 2A5B525E0F631BB6107639E2A69DF15986FB0D05) and its public PDB
  • CADET.EXE 32bit version from Full Tilt! Pinball (SHA-1 3F7B5699074B83FD713657CD94671F2156DBEDC4)

Tools used: Ghidra, Ida, Visual Studio

What was done:

  • All structures were populated, globals and locals named.
  • All subs were decompiled, C pseudo code was converted to compilable C++. Loose (namespace?) subs were assigned to classes.

Compiling:
Project uses C++11 and depends on SDL2 libs.
On Windows:
Download and unpack devel packages for SDL2 and SDL2_mixer.
Set paths to them in CMakeLists.txt, see suggested placement in /Libs.
Compile with Visual Studio; tested with 2019.

On Linux:
Install devel packages for SDL2 and SDL2_mixer.
Compile with CMake; tested with GCC 10, Clang 11.

Plans:

  • Decompile original game
  • Resizable window, scaled graphics
  • Loader for high-res sprites from CADET.DAT
  • Misc features of Full Tilt: 3 music tracs, multiball, centered textboxes, etc.
  • Cross-platform port
    • Using SDL2, SDL2_mixer, ImGui
    • Maybe: Android port
  • Maybe x2: support for other two tables
    • Table specific BL (control interactions and missions) is hardcoded, othere parts might be also patched

On 64-bit bug that killed the game:
I did not find it, decompiled game worked in x64 mode on the first try.
It was either lost in decompilation or introduced in x64 port/not present in x86 build.
Based on public description of the bug (no ball collision), I guess that the bug was in TEdgeManager::TestGridBox