http://bugs.winehq.org/show_bug.cgi?id=14751
Summary: Max Payne 2: PP effects produce black screen Product: Wine Version: CVS/GIT Platform: PC-x86-64 URL: http://www.rockstargames.com/maxpayne2/mp2_downloads.htm l OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: directx-d3d AssignedTo: wine-bugs@winehq.org ReportedBy: liquid.acid@gmx.net CC: stefandoesinger@gmx.at
Hi there,
I already mentioned in http://bugs.winehq.org/show_bug.cgi?id=14724 that MP2 breaks even more in current git master (snapshot from today, latest commit is 428fd07e73c08e01f463982fab0d4b4267695b8f).
Tested this with ORM=fbo and PPE=medium. The screen simply goes black as soon as PP effects are used.
Regression testing between wine-1.1.2 (no such issue there) and 428fd07e73c08e01f463982fab0d4b4267695b8f reveals that this commit introduces the problem: [bc4435e4063c70d0ef2fcc2ab7bb6df3efcfc0e7] wined3d: Use a hashmap to store the ffp shaders.
Adding Stefan Dösinger to CC.
Greets, Tobias
http://bugs.winehq.org/show_bug.cgi?id=14751
Tobias Jakobi liquid.acid@gmx.net changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |jb.faq@gmx.de Keywords| |download, regression
http://bugs.winehq.org/show_bug.cgi?id=14751
--- Comment #1 from Tobias Jakobi liquid.acid@gmx.net 2008-08-04 18:48:22 --- Created an attachment (id=15275) --> (http://bugs.winehq.org/attachment.cgi?id=15275) Screenshot showing the "darkness"
This is with ORM=fbo and PPE=medium.
In fact the scene is not rendered completly black, but Max is a bit visible (when positioned correctly, e.g. under a light source).
http://bugs.winehq.org/show_bug.cgi?id=14751
Tobias Jakobi liquid.acid@gmx.net changed:
What |Removed |Added ---------------------------------------------------------------------------- Summary|Max Payne 2: PP effects |Max Payne 2: PP effects |produce black screen |produce (nearly) black | |screen
http://bugs.winehq.org/show_bug.cgi?id=14751
--- Comment #2 from Tobias Jakobi liquid.acid@gmx.net 2008-08-04 18:58:56 --- Reconfirming this with newest wine git master snapshot.
Latest commit id: 9b572170da5f35a80c6bb07ab8ebcda7a854f27f
Only shows up with ORM=fbo, NOT with ORM=backbuffer.
http://bugs.winehq.org/show_bug.cgi?id=14751
Tobias Jakobi liquid.acid@gmx.net changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |alexd4@inbox.lv
http://bugs.winehq.org/show_bug.cgi?id=14751
Jan Buecken jb.faq@gmx.de changed:
What |Removed |Added ---------------------------------------------------------------------------- CC|jb.faq@gmx.de |
--- Comment #3 from Jan Buecken jb.faq@gmx.de 2008-08-04 19:12:27 --- Cannot confirm this on my ati mobility radeon 2600 hd and propritary driver (8.512)
Hence I delete me from CC, if you need further info or if you see a way I can help than add me again.
http://bugs.winehq.org/show_bug.cgi?id=14751
--- Comment #4 from Tobias Jakobi liquid.acid@gmx.net 2008-08-08 17:26:05 --- Created an attachment (id=15353) --> (http://bugs.winehq.org/attachment.cgi?id=15353) max can also be white
fooling around a bit with the VT switching and other stuff got me this screen
http://bugs.winehq.org/show_bug.cgi?id=14751
--- Comment #5 from Tobias Jakobi liquid.acid@gmx.net 2008-08-20 14:59:32 --- I just found out that the issue is dependent on two additional ingame setting: - pixel shader skins - shadow setting
The original issue was produced using no pixel shader skins and shadows set to low. However a multitude of other issues arise when fiddling around with the other ingame settings. For example simply setting PSS (pixel shader skins) to medium returns correct rendering of the bullettime effect. Blur effect is still broken, but appearance has also changed. Furthermore with medium shadow rendering gets broken as soon as a transparent object comes into view, like the windows in the hospital room. The screen is nearly black then. However switching into BT mode then produces a correct image.
Very weird....
http://bugs.winehq.org/show_bug.cgi?id=14751
Jan Buecken jb.faq@gmx.de changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |jb.faq@gmx.de
--- Comment #6 from Jan Buecken jb.faq@gmx.de 2008-08-20 18:05:15 ---
Furthermore with medium shadow rendering gets broken as soon as a transparent object comes into view, like the windows in the hospital room. The screen is nearly black then. However switching into BT mode then produces a correct image.
Confirming this on ati. Other things mentioned in comment 5 not tested.
http://bugs.winehq.org/show_bug.cgi?id=14751
--- Comment #7 from Alexander Dorofeyev alexd4@inbox.lv 2008-08-21 13:58:36 --- I checked this on geforce 6100 in git yesterday and finished today (with yesterday's git). I do see the problem. Disabling NPOT extension does not seem to be necessary. But ORM=fbo and Use_GLSL=disabled do appear to be mandatory, both (removing either leads to problem disappearing). PPE settings do not appear to have any effect, it occurs even with PPE=off. Occurs with all Character Shadow settings, but medium and high shadows show rendering errors in different scenes than shadows=low.
Also I found out that disabling (commenting out) selection of arbfp_fragment_pipeline in select_fragment_implementation (directx.c) makes the problem go away. This means the problem is somehow related to fixed function replacement shaders (arb backend).
But, I can't reproduce this in bc4435e4063c70d0ef2fcc2ab7bb6df3efcfc0e7. Maybe I made some mistake, but, Tobias, are you sure your regression test was right?
http://bugs.winehq.org/show_bug.cgi?id=14751
--- Comment #8 from Tobias Jakobi liquid.acid@gmx.net 2008-08-22 04:58:46 --- Hmm, I did the bisect with the "character shadows"="low" setting, probably it's a different commit with medium and high setting. I think I have to recheck this, but I won't have the time for it now and from next week on I won't have access to my NV based system (for six weeks) so I think this has to wait.
http://bugs.winehq.org/show_bug.cgi?id=14751
--- Comment #9 from Tobias Jakobi liquid.acid@gmx.net 2008-10-13 16:44:00 --- Currently taking a (new) look at it.
It still happens with wine-1.1.6 (reconfirming there).
However Alexander was right. It doesn't depend on NP2 settings, but on GLSL settings (and ORM, but since backbuffer isn't usuable for me I have to stay with FBO here).
So ORM=FBO for all checks done here. Post processing effects are set to medium.
Enabling GLSL does indeed fix the issues (not the GL errors though). However the PP effects now render fine again. Disable GLSL and everything breaks.
I'm going to do a new bisect in the next few days.
http://bugs.winehq.org/show_bug.cgi?id=14751
--- Comment #10 from Jan Buecken jb.faq@gmx.de 2008-10-16 12:05:11 --- Hi, tested the follwing with git today on ati:
MP2-Demo:
ORM=fbo or backbuffer (no difference!), PPE=medium
All seems to be good (bullet-time) but a completly black screen occurs if motion-blur is used. For example the first video sequence after you start controlling the character in the hospital is black. I want to draw attention to the fact that this is independent if fbo or backbuffer is used.
And this doesn't append on the the GLSL setting in the registry.
http://bugs.winehq.org/show_bug.cgi?id=14751
Jan Buecken jb.faq@gmx.de changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |NEW Ever Confirmed|0 |1
--- Comment #11 from Jan Buecken jb.faq@gmx.de 2008-10-16 12:05:43 --- *** This bug has been confirmed by popular vote. ***
http://bugs.winehq.org/show_bug.cgi?id=14751
--- Comment #12 from Tobias Jakobi liquid.acid@gmx.net 2008-10-20 13:39:28 --- Now bisecting between wine-1.1.2 and wine-1.1.3 again.
wine settings: GLSL=disabled ORM=fbo
ingame settings: PPE=medium PSS=off mirrors=off
The interesting option is 'character shadows'.
When CS=low, then the issue only appears when enabling PP through e.g. bullettime or motion blur (in cutscenes). When switching it to medium (or high), the issue becomes different. Now transparent surfaces like the windows in the hospital room seem to trigger the "blackout" effect. In fact switching into bullettime mode restores the vision while the effect is active. Switch out of bullettime and the screen is black again.
So, I'm going to bisect once with CS=low between 1.1.2 and 1.1.3 and again with CS=medium.
http://bugs.winehq.org/show_bug.cgi?id=14751
--- Comment #13 from Tobias Jakobi liquid.acid@gmx.net 2008-10-23 15:56:41 --- Getting weird results for this bisect.
Bisect again shows commit bc4435e4063c70d0ef2fcc2ab7bb6df3efcfc0e7 from 2008-07-31 (wined3d: Use a hashmap to store the ffp shaders). But manually checking out this version, testing it, then reverting the patch doesn't change a bit. So this commit is GOOD.
Now manually checking: commit 2d6d8795624b95e654d712e1958dd631df46dcfc ( wined3d: Handle CTXUSAGE_BLIT with offscreen targets and FBO ORM) That one is BAD.
Bisecting between those and this commit is the faulty one: [14b24058d69d73ebf6b70bc36c8aa62993351079] wined3d: GL_ARB_fragment_program ffp implementation.
Verified this by checking out 14b24058d69d73ebf6b70bc36c8aa62993351079, testing, reverting the patch and testing again.
Another verification done with git master and patching select_fragment_implementation in directx.c such that always nvts_fragment_pipeline is returned (NV35 hw here). Also fixes the issues.
So the bug is inside the (relatively new) arbfp_fragment_pipeline implementation. @stefand: Any idea how to track this down?
Greets, Tobias
http://bugs.winehq.org/show_bug.cgi?id=14751
Tobias Jakobi liquid.acid@gmx.net changed:
What |Removed |Added ---------------------------------------------------------------------------- Summary|Max Payne 2: PP effects |Max Payne 2: PP effects |produce (nearly) black |produce black screen in ARB |screen |mode
http://bugs.winehq.org/show_bug.cgi?id=14751
Tobias Jakobi liquid.acid@gmx.net changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED
--- Comment #14 from Tobias Jakobi liquid.acid@gmx.net 2008-11-03 13:28:46 --- This seems to be fixed with latest git master.
I suspect it's this commit by stefand: http://source.winehq.org/git/wine.git/?a=commit;h=8b38533f691154b889f92ab872...
Thanks Stefan! :)
Resolving to FIXED.
http://bugs.winehq.org/show_bug.cgi?id=14751
Alexandre Julliard julliard@winehq.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #15 from Alexandre Julliard julliard@winehq.org 2008-11-07 10:46:59 --- Closing bugs fixed in 1.1.8.
http://bugs.winehq.org/show_bug.cgi?id=14751
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Version|CVS/GIT |unspecified