http://bugs.winehq.org/show_bug.cgi?id=9774
Summary: Max Payne 1.05 - T-junction-like graphic bugs Product: Wine Version: 0.9.45. Platform: PC-x86-64 OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: wine-directx-d3d AssignedTo: wine-bugs@winehq.org ReportedBy: liquid.acid@gmx.net
Hi there,
I recently installed Max Payne on Wine and I'm now in the progress of playing the entire game on my Linux machine. It works fine but I encountered some graphical bugs.
The bug is somehow like the old t-junction errors (I know them from my half-life mapping experiences), causing single flashing pixels of some odd color at points where game geometry is put together. Most of the time at polygon edges. This errors appear nearly everywhere in the game, sometimes you hardly notice them, but sometimes they are very intense.
I know for sure that when I played the game on my Windows machine there we no t-junction errors at all.
I would like to supply some screenshots to illustrate the problem, but the integrated screenshot functionality doesn't seem to work. It simply outputs a completly black TGA file. I'm going to file another bugreport for this one.
Any idea how to get a screenshot with another method?
http://bugs.winehq.org/show_bug.cgi?id=9774
Vitaliy Margolen vitaliy@kievinfo.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Severity|normal |trivial
--- Comment #1 from Vitaliy Margolen vitaliy@kievinfo.com 2007-09-25 20:42:15 --- Use virtual desktop and take a screen-shot if the window with external tools.
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #2 from Vitaliy Margolen vitaliy@kievinfo.com 2007-09-25 20:43:02 --- Also helps to know your exact configuration (video card, driver version, any Wine settings that were changed)
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #3 from Tobias Jakobi liquid.acid@gmx.net 2007-09-28 15:43:31 --- OK, here we go. I set up a virtual desktop and made the screenshot:
http://img118.imageshack.us/my.php?image=testxh3.png
Notice the three bluish dots at positions where polygon edges meet each other. Positions vary when changing the viewing angle or moving the camera.
Video card is a nvidia Geforce FX 5900 with 128MB RAM, attached to a IBM P260 21 inch CRT. Driver version is x11-drivers/nvidia-drivers-100.14.11
I haven't changed any wine settings related to graphics.
Also going to recheck this with wine 0.9.46
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #4 from Tobias Jakobi liquid.acid@gmx.net 2007-10-01 18:49:24 --- Created an attachment (id=8350) --> (http://bugs.winehq.org/attachment.cgi?id=8350) screenshot showing t-junction errors
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #5 from Tobias Jakobi liquid.acid@gmx.net 2007-10-01 18:52:34 --- Hello again,
I've created some other screenshots and uploaded one directly into bugzilla (in case imageshack deletes the first one).
I updated wine to 0.9.46 today but this doesn't fix the problem.
I have asked a friend of mine to try Max Payne via wine on his machine. He also has a gentoo box but with different gfx hardware (his machine is a laptop with a mobility nvidia chipset). I think he'll give me feedback tomorrow about how it worked out.
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #6 from Tobias Jakobi liquid.acid@gmx.net 2007-10-01 18:54:51 --- Oh yes, I completly forgot:
I tested the whole thing with WINEDEBUG=warn+all,fixme+all but didn't find anything useful or anything that I think could be related to the error. I really would like to help to find and fix this problem. Is there anything else I can do to debug this stuff?
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #7 from Tobias Jakobi liquid.acid@gmx.net 2007-10-02 13:35:15 --- My friend cloned by Max Payne installation on his notebook.
The machine is powered by a Pentium 4 2.53GHz, 768MB RAM and GeForce4 440 Go with 64MB dedicated RAM. As already stated his box is also a Gentoo one.
Driver version on his system is: x11-drivers/nvidia-drivers-1.0.9639 Kernel version is: 2.6.20-gentoo-r8
The gfx driver is the last driver in gentoo portage that works with this mobile card.
He experiences exactly the same problems. So this doesn't seem to be a problem of the graphics drivers.
He tested this with a in-game resolution of 1024x768 and 800x600, both with 32 bit color-depth. Setting bitdepth to 16 bit doesn't change anything.
http://bugs.winehq.org/show_bug.cgi?id=9774
Tobias Jakobi liquid.acid@gmx.net changed:
What |Removed |Added ---------------------------------------------------------------------------- Keywords| |regression
--- Comment #8 from Tobias Jakobi liquid.acid@gmx.net 2007-10-04 14:25:43 --- Hi there,
I went down to wine version 0.9.20 to check if the bug could be a regression (the screenshots in the appDB don't show any obvious t-junction errors). And yes, it seems like one. The problem doesn't appear with 0.9.20.
Still the also mentioned screenshot bug (and missing savegame pictures) remains.
I'm now bisecting between wine version 0.9.20 and 0.9.45 to find the commit that breaks correct rendering. I report back as soon as git bisect finishes.
http://bugs.winehq.org/show_bug.cgi?id=9774
Tobias Jakobi liquid.acid@gmx.net changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |stefandoesinger@gmx.at
--- Comment #9 from Tobias Jakobi liquid.acid@gmx.net 2007-10-04 20:50:53 --- git bisect output: [c9b8a79e0bd61204cfe7b88362617da9a485f0e0] wined3d: Implement vertical refresh sync.
Search in the git tree: author: Stefan Dösinger stefandoesinger@gmx.at Sun, 3 Jun 2007 11:20:27 +0000 (13:20 +0200)
committer: Alexandre Julliard julliard@winehq.org Thu, 14 Jun 2007 11:21:54 +0000 (13:21 +0200)
commit: c9b8a79e0bd61204cfe7b88362617da9a485f0e0 tree: 0307f1b5768eb9576a36ef85fa7606000fc4b803 parent: 43fda1f4f0d4b88003b2ac1ff092aa9ba8783b95
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #10 from Tobias Jakobi liquid.acid@gmx.net 2007-10-05 15:53:07 --- I'm very sorry. I went through the vertical sync commit myself because I was very confused that this rendering error could have anything to do with vsync artifacts. And I didn't find any obvious errors.
So I re-bisected, this time between 0.9.38 and 0.9.40 (the rendering errors occur in 0.9.40 but NOT in 0.9.38).
Obviously I made some mistakes with my last bisect (it was quite late when I did it). The correct bisect output is:
078523f73e7b708dab06e888c24a1595bb5a63d8 is first bad commit commit 078523f73e7b708dab06e888c24a1595bb5a63d8 Author: Stefan Dösinger stefandoesinger@gmx.at Date: Thu Jun 14 11:18:03 2007 +0200
wined3d: Present does not clear the depth stencil.
:040000 040000 81558b3157da6386c88704dc87c4ff492779eeff 7ef9b91372f0382feeaac98f8a17b2f4df59ce6a M dlls
Now this looks more like it. Both depth and stencil mentioned in the commit.
I hope Stefan didn't spend too much time hunting for ghosts :-)
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #11 from Tobias Jakobi liquid.acid@gmx.net 2007-10-20 13:46:39 --- Errors still exist with wine-0.9.47
http://bugs.winehq.org/show_bug.cgi?id=9774
Jan Buecken jb.faq@gmx.de changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |jb.faq@gmx.de
--- Comment #12 from Jan Buecken jb.faq@gmx.de 2007-10-24 14:12:36 --- confirming the problem
tested with wine-0.9.46
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #13 from Stefan Dösinger stefandoesinger@gmx.at 2007-10-24 14:39:04 --- The screenshot looks a bit dark. Can the bug be seen in a brighter area?
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #14 from Tobias Jakobi liquid.acid@gmx.net 2007-10-24 17:38:27 --- I have a savegames with an area where the 'effect' is much more noticeable, meaning a lot pixels with 'wrong' color are rendered at polygon edges. I can provide a screenshot tomorrow, but I must say it's not going to be any brighter. The area is in the subway station on the rails, naturally a quite dark setting.
The whole game is very dark, so I'm going to have a hard time finding an area with lots of light and the same effect (the rail tracks seem to be constructed in such a way that triggers the 'effect').
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #15 from Tobias Jakobi liquid.acid@gmx.net 2007-10-25 12:41:23 --- Created an attachment (id=8766) --> (http://bugs.winehq.org/attachment.cgi?id=8766) screen with obvious gfx errors (1)
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #16 from Tobias Jakobi liquid.acid@gmx.net 2007-10-25 12:41:59 --- Created an attachment (id=8767) --> (http://bugs.winehq.org/attachment.cgi?id=8767) screen with obvious gfx errors (2)
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #17 from Tobias Jakobi liquid.acid@gmx.net 2007-10-25 12:42:27 --- Created an attachment (id=8768) --> (http://bugs.winehq.org/attachment.cgi?id=8768) screen with obvious gfx errors (3)
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #18 from Tobias Jakobi liquid.acid@gmx.net 2007-10-25 12:43:04 --- Created an attachment (id=8769) --> (http://bugs.winehq.org/attachment.cgi?id=8769) screen with obvious gfx errors (4)
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #19 from Tobias Jakobi liquid.acid@gmx.net 2007-10-25 12:43:36 --- Created an attachment (id=8770) --> (http://bugs.winehq.org/attachment.cgi?id=8770) screen with obvious gfx errors (5)
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #20 from Tobias Jakobi liquid.acid@gmx.net 2007-10-25 12:43:59 --- Created an attachment (id=8771) --> (http://bugs.winehq.org/attachment.cgi?id=8771) screen with obvious gfx errors (6)
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #21 from Tobias Jakobi liquid.acid@gmx.net 2007-10-25 12:44:25 --- Created an attachment (id=8772) --> (http://bugs.winehq.org/attachment.cgi?id=8772) screen with obvious gfx errors (7)
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #22 from Stefan Dösinger stefandoesinger@gmx.at 2007-10-25 15:21:27 --- Hmm, this looks like a precision problem, and the clear background color comes through. This could actually be a game bug, where the game doesn't make sure that the vertices are close enough at each other. So the in princible undefined old colors come through.
Now while Windows doesn't give any guarantees regarding the back buffer content after a present call, it is unlikely to be the color that wine currently clears the back buffer with. On most drivers it is the old front buffer content, or just black. In both cases the lack of precision won't be that visible due to the general color theme of the game. So the question is if it is wine's fault that the old color shows up, or if the old color just shouldn't be such an obvious color.
Can anyone download the directx sdk and install it on a windows box? In this case you get a new control panel entry which allows you to switch to the debug runtime. The debug runtime fills the back buffer after a present with a yellow and black color pattern according to the docs. With this a simmilar bug might be visible on Windows.
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #23 from Jan Buecken jb.faq@gmx.de 2007-10-27 05:26:03 --- You are right! I downloaded the directx sdk 9.0c and tested it with the debug-mode in windows. The effect is the same [with another color (pink dots)].
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #24 from Tobias Jakobi liquid.acid@gmx.net 2007-10-28 05:41:10 --- In the commit that I found through bisecting you changed the clear color to pink. Maybe that's the problem? What about reverting the clear color to black (?) again? Also the color that appears in the polygon gaps doesn't seem to be the clear color directly, but a 'processed' color - but it's definitely dependent on the clear color (according to what Jan found out). Maybe a clear color of black always results in a polygap-color that one can't notice?
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #25 from Stefan Dösinger stefandoesinger@gmx.at 2007-10-28 05:53:32 --- (In reply to comment #24) Correct, yes, we should change the clear color back to black, or remove that clear call entirely.
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #26 from Tobias Jakobi liquid.acid@gmx.net 2007-11-14 17:18:19 --- Rechecked this with wine-0.9.49 and the visual errors are still there.
http://bugs.winehq.org/show_bug.cgi?id=9774
--- Comment #27 from Tobias Jakobi liquid.acid@gmx.net 2007-12-02 11:42:32 --- Updated to wine-0.9.50 today and rechecked. Bug is still there. Why isn't this fixed, especially since the source of the bug was found?
http://bugs.winehq.org/show_bug.cgi?id=9774
Tobias Jakobi liquid.acid@gmx.net changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |RESOLVED Resolution| |FIXED
--- Comment #28 from Tobias Jakobi liquid.acid@gmx.net 2007-12-27 06:12:44 --- A big thanks goes to Stefan for fixing this in commit http://source.winehq.org/git/wine.git/?a=commit;h=cd858dfa220b05d0233010ea08...
Closing this bug now...
Thanks again! Tobias
http://bugs.winehq.org/show_bug.cgi?id=9774
Dan Kegel dank@kegel.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #29 from Dan Kegel dank@kegel.com 2008-01-28 05:44:56 --- Closing all RESOLVED FIXED bugs older than four weeks.