github moesuito/gt4-staticrecomp v1.3.0-rc14
GT4Recomp 1.3.0-rc14 (test build)

pre-release5 hours ago

GT4Recomp 1.3.0-rc14 (test build)

A test release candidate of 1.3.0: camera zoom in races, the throttle and brake over the whole trigger travel by default,
Spec II's speed and timer text kept in shape on wide screens, a first-start note for creating your save, a guard for graphics
drivers that close the game while CoreGS 2.0 prepares its shaders, and any Spec II 1.10 image installing. The race itself is unchanged.

New

  • Camera zoom in races (TASK-160, both games, on by default). Hold R3 and press up or down on the D-pad to zoom the race
    camera in or out, one step per frame, from half to twice the game's view; L3 + R3 resets it. The zoom scales the projection the
    game itself builds for its race cameras, with the game's own divide, so at the normal view nothing changes; the race clock stays
    the same while zooming. In races the game does nothing else with these buttons. fov_adjust=0 in game.ini (or
    GT4_FOV_ADJUST=0) turns it off. Idea from the community PCSX2 patch, used as a reference only.
  • Full trigger range (TASK-161, on by default). The game turns a trigger's pressure into its own coarse scale: 46 throttle
    levels with 100% at a pressure of 45 out of 255, and 85 brake levels with 100% at 84 (our controls already stretch the trigger
    travel over those, TASK-147). The throttle and brake now follow the whole travel in about 254 levels, still reaching 100% at
    the end. Controls > "Full trigger range" off (or trigger_full_range=0 in game.ini, GT4_TRIGGER_FULL_RANGE=0) is the
    game's own scale. Idea from the community "Full Trigger Sensitivity" PCSX2 patch, used as a reference only.
  • First-start note on Play (TASK-164, issue #160). On a new card Gran Turismo 4 first asks whether to create its game data,
    and only Cross answers; players who pressed another key thought the game had frozen. When the selected game has no save yet
    (none in the launcher's library, none on the card the game left behind), Play now shows a short note before the game starts:
    confirm with Cross, named with the keys of your own controls (W or Enter on the keyboard and A on a controller by default), and
    when a question starts on Cancel or No, move to Yes or OK first. Play starts the game; Cancel starts nothing. The note is gone
    once the save exists.
  • Any Spec II 1.10 image installs (TASK-173). Spec II images people make (translations, car packs) keep Spec II 1.10's
    game code and change only its data. The setup now checks that code exactly: the game program (which holds the whole Spec II
    mod) and the 30 IOP modules must be byte-identical to 1.10's; then the image installs whatever its data. An image with other
    code, such as an earlier Spec II release, is refused as before: GT4Recomp is compiled ahead of time for 1.10's code, so other
    code cannot run.
  • Spec II PT-BR + CPSP image (TASK-167). The Brazilian Portuguese Spec II image with some cars replaced by Gran Turismo PSP
    cars is recognised by name. Its game program is the same as Spec II 1.10's, and every file but the game data volume is
    identical to the PT-BR image, so it ran before as an unnamed "variant"; a recorded lap on it matches the PT-BR image frame for
    frame.

Fixes

  • A graphics driver that closed the game while CoreGS 2.0 prepared its shaders did it again at every start (TASK-170). A
    player's log (RX 6700 XT) showed the launcher's shader step stopping about halfway and the game then closing at boot, every
    time, while GSdx ran fine on the same PC. The likely cause is one shader compile that ends the process inside the driver.
    CoreGS 2.0 now notes each compile before it starts: when a start finds that a compile never ended, it prepares the shaders one
    at a time from then on, and if a single compile was running, that one shader is left out and its draws use CoreGS 2.0's
    general shader instead (same picture, a little slower there). A driver update starts clean. The exported log bundle now
    includes logs/shaders.log, the shader step's own log.
  • 2D screens over the race stretched on screens wider than 16:9 (TASK-095, re-does Luan Antony's pull request #177). At
    21:9 and 32:9 the race view fills the screen, but the pre-race screens, the Time Trial lobby and the pause menu were stretched
    with it. Their canvas is now the centred 16:9 box, and the pause menu's text is narrowed with it, so they keep their shape. At 16:9, and wherever the picture is pillarboxed, nothing changes; the race monitor stays as the game draws it.
    Checked by replaying drives on both games: at 16:9 the race clock, the simulation state and every frame are identical; at
    21:9 the race is identical and only those screens differ, narrowed by exactly the ratio. The results and the pit menu use the
    same calls but were not reached in a replay.
  • The 360 camera also stopped a left-stick or wheel throttle (TASK-159b). While the camera is on, only the right stick's
    up and down no longer accelerate and brake; a throttle or brake set on the left stick or on a wheel's pedals works again.
  • Spec II on screens wider than 16:9 stretched the speedometer, the odometer and the lap timer (TASK-166). Gran Turismo 4
    draws that text with a font whose width it sets for a 16:9 screen; our ultrawide support corrects that width (and the rear-view
    mirror's) on v2.00, but the correction had not been carried over to Spec II. The same two instructions are now corrected in
    Spec II. At 16:9 nothing changes: every compared frame is identical to rc13.
  • The trigger meters on the Controls page changed size while a trigger moved. The number beside each bar now keeps the
    width of its widest text, so the bar stays the same size.

The same game

Against rc13, on this build (both games built again with a fresh profile, from the hooked units of the rc14 test build):

  • Physics: the race clock at every VBlank and the whole simulation state, byte for byte, on three recorded Spec II laps
    (George V Paris, El Capitan, Tokyo R246) and on the v2.00 night race with 5 AI cars. The recordings press the throttle and
    brake fully or not at all, where the full trigger range and the game's own scale give the same values.
  • Frames: every compared frame identical: 24, 35 and 35 on the three Spec II laps at 1x, 12 of 12 on a lap at 6x, 10 of 10
    on the v2.00 night race at 3x and 14 of 14 on the v2.00 Arcade path.
  • Checks while running, against the exact interpreter: 236,448 VU1 samples on Spec II and 355,388 on v2.00, and the MTVU check
    over 8,032 frames, all with 0 differences.
  • Wide screens (TASK-095): on a Spec II drive with the pause menu, 16:9 identical to a build without TASK-095; at 21:9 and
    32:9 the race identical and the overlays narrowed by the ratio (an image check: 10 of 10 frames at 21:9, 9 of 10 at 32:9, the
    tenth a frame of the pause menu's fade). At 32:9 this build draws every frame identical to the rc14 test build.
  • Launcher: 665 tests, focus checks 294/294 (en-US) and 301/301 (pt-BR), navigation tests 217/217; locale keys
    identical in both languages (695). Native tests 301/301 on each game, Python tests 790.

Known issues

  • Photo mode on slower graphics cards (CoreGS 2.0): each picture still waits for the graphics card twice, so frame times can
    be uneven (TASK-162).
  • The first license (v2.00, Test Course) on mid-range laptops (CoreGS 2.0): the scene has about 26,000 primitives per frame,
    well above a race, and can drop below 60 FPS on a laptop processor; planned for rc15 (TASK-171).
  • An earlier Spec II release (not 1.10) is refused at setup: its mod code differs from 1.10's (TASK-172). Any image with
    Spec II 1.10's code is supported (TASK-173).

Thanks

  • Luan Antony, whose study of the first save creation (pull request #167) the first-start note comes from, and whose pull
    request #177 the wide-screen 2D canvas was re-done from.
  • Duchi/Vyerq ("FoV Adjustment") and Silent & Aero_ ("Full Trigger Sensitivity"), whose PCSX2 patches gave the ideas for
    the camera zoom and the full trigger range.
  • The player on Discord who sent the log that led to the shader guard.

Install or update

  • From rc8 to rc13 or another 1.3.0 test build: the launcher offers rc14 on the Beta channel (Settings > General).
  • Installer: run GT4Recomp-Setup-1.3.0-rc14.exe. It keeps existing settings, saves and replays.
  • Portable: unpack GT4Recomp-1.3.0-rc14-portable.zip anywhere. Settings, saves and data stay inside the folder.
  • After the update, Play may compile CoreGS 2.0's shaders once, as after every update.
  • The full trigger range is on after the update. Turn it off on the Controls page for the game's own scale.
  • Disc images: you need your own: Gran Turismo 4 (USA, v2.00), and your own Spec II 1.10 image for the mod. GT4Recomp ships no
    copyrighted game files.

Requirements

Windows 10 or 11, 64-bit, a processor with SSE4.1. CoreGS 2.0 needs a graphics card with Vulkan 1.2.

Credits and legal

Our code is MIT. GSdx is the PCSX2 GS (GPL-3.0-or-later), downloaded by the launcher and run in its own process; the game
executable contains no GPL code. FSR1 sharpening is written from AMD's published MIT design. Community PCSX2 patches (camera zoom,
trigger sensitivity) were used as references only; no patch code is copied.

Don't miss a new gt4-staticrecomp release

NewReleases is sending notifications on new releases.