id Software to release Doom 3 source code
  • IKE
  • Butterfly
    Butterfly
    IKE
    Posts: 73 from 2009/11/7
    From: Southern CA
    Great work so far! Very smooth on PowerMac G5 2.0 DP (Radeon 9650/256MB). Worked thru the first two objectives despite the missing shaders. Seemed to avg 30-40 FPS when running.

    Wasn’t able to compile, aborts at ~35% complaining ‘undefined reference to main’ (during d3xp.so linking)

    Noticed ‘timedemo demo1’ doesn’t run on the console and causes program to shut down.

    Keep up the good work, I’ll try it out some more once I load it on my MacMini 1.5 and PowerBook 1.67!
    IKE

    MacMini G4 1.5Ghz/PowerBook G4 1.67Ghz/PowerMac G5 2.0Ghz DP 7,2 Radeon 9650/256MB
  • »06.01.21 - 05:57
    Profile Visit Website
  • Just looking around
    Posts: 9 from 2020/12/24
    @IKE

    Thanks. I hope it will be better.

    dx3.so needs the same treatment as base.so: Just add dllglue.c code to it. Other stuff inside dx3 needed to be done I don't know if it works because I didn't work on it.

    I tried to compile with dlfcntl.h include but so far only works for main executable (as other versions compilers do) but it fails with "main" needed for the ".so". So the dll library I left there is needed.

    And no, timedemo fails here too.

    I didn't use the cmake way to compile BTW: Barebones single makefile for me.

    [ Edited by Cowcat 06.01.2021 - 08:34 ]
  • »06.01.21 - 08:20
    Profile
  • Just looking around
    Posts: 9 from 2020/12/24
    And thanks to Waldiamiga for being the betatester. He found that the only way to run the game "almost always" was from an icon with stack.

    Different options added are in my Github: People should go through the code to see what is done. No README for now.
  • »06.01.21 - 08:29
    Profile
  • Just looking around
    Posts: 9 from 2020/12/24
    So far:

    Two bugs (and a half?). One is related to "brightness" of the scenes and the other the failure of representing the models.

    - You can dim the lights with r_skipLightScale.
    - Models rendering can be seen somewhat with r_skipUpdates (heads still shake).

    The other half bug happens in the first scene in the hangar: In my system is basically all red with some spots till entering the labs. In other systems with more vram or other gfx cards is a triangle/texture dance.

    Still shims of black triangles and quads here and there, probably related to models and scene render.

    My workaround for the scenes to be viewed implied leaving out a couple of binding textures and keeping the one that has all colours.

    Did some tests and concluded that texture mapping fails in the driver (glTexGenfv).
    Got a couple of simple glut programs that also fail to work with that function (probably more are involved).

    Not much can be done for now till this is resolved by the morphos team developers.

    Some bugfixes for mirrors and cleanup for draw_arb soon.





    [ Edited by Cowcat 14.01.2021 - 12:19 ]
  • »14.01.21 - 11:52
    Profile
  • IKE
  • Butterfly
    Butterfly
    IKE
    Posts: 73 from 2009/11/7
    From: Southern CA
    Great work! Can't wait to test the bugfixes. Keep up the good work.
    IKE

    MacMini G4 1.5Ghz/PowerBook G4 1.67Ghz/PowerMac G5 2.0Ghz DP 7,2 Radeon 9650/256MB
  • »14.01.21 - 14:32
    Profile Visit Website
  • Order of the Butterfly
    Order of the Butterfly
    tolkien
    Posts: 371 from 2013/5/29
    Damn...It seems I have a bad Doom3 version . It fails to load with this error:
    ERROR: LoadTGA( textures/sfx/monitor_glass2.tga ): Only type 2 (RGB), 3 (gray), and10 (RGB) TGA images supported

    Will try with another version :/


    EDIT: I have used the latest patch version of DOOM3 and now it runs. It goes at 20 fps +- with a lot of gfx glitches but it is a really good start!

    [ Editado por tolkien 16.01.2021 - 01:37 ]
    PowerMac G5 - PowerBook G4 - MacMini with MorphOS. Amiga 1200/060 AmigaOS 3.x
  • »15.01.21 - 22:50
    Profile
  • Just looking around
    Posts: 9 from 2020/12/24
    Did someone realize that in my latest "Project UAC" video the first scene kinda works?

    I got a theory that lights, textures and entities that go beyond some boundaries are the cause of the render errors all the way.

    In the testing video two functions that check a certain situation exceeding > 1024, I make it return from those instead of continuing the drawing that normally the engine does.

    And that happens all over the place: The debug option "r_checkBounds" spits a whole bunch of problems in the original game but it does nothing about it. (tr_light.cpp).

    The bypassed option test I did is where "r_showUpdates" is called in tr_lightrun.cpp (and more): In the first scene (the hangar) the problems are more apparent.

    The missing model bodies, the shaking model heads and general gfx glitches here and there have probably a root cause in those referenced boundaries.

    I checked a whole bunch of source code ports for different systems and no apparent problems are in those versions.

    Morphos gfx drivers allergic to this engine? All is theory.

    And no response from MOS developers about why texture mapping fails to do its thing.
  • »24.01.21 - 10:00
    Profile
  • MorphOS Developer
    cyfm
    Posts: 422 from 2003/4/11
    From: Germany
    Quote:

    Cowcat schrieb:


    I checked a whole bunch of source code ports for different systems and no apparent problems are in those versions.

    Morphos gfx drivers allergic to this engine? All is theory.

    And no response from MOS developers about why texture mapping fails to do its thing.





    I've been discussing this with bigfoot lately and we came to the conclusion that various things related to glTexGenfv most likely don't work as required or the required infrastructure to make it work correctly isn't available in TinyGL. So no easy solution to it as of now.
  • »24.01.21 - 12:37
    Profile Visit Website
  • Just looking around
    Posts: 9 from 2020/12/24
    Thanks for the explanation.

    The Texgen bug is not exclusive of Doom3 but game works ok with the morphos workaround I did running on another system. It is what it is now.

    The other part about boundaries surpassed hopefully will be discussed in dhewm3 Github.

    [ Edited by Cowcat 24.01.2021 - 13:20 ]
  • »24.01.21 - 13:19
    Profile