| View Issue Details [ Jump to Notes ] | [ Issue History ] [ Print ] | ||||||||
| ID | Project | Category | View Status | Date Submitted | Last Update | ||||
|---|---|---|---|---|---|---|---|---|---|
| 0005637 | Spring engine | General | public | 2017-07-08 06:00 | 2020-05-14 15:24 | ||||
| Reporter | Google_Frog | ||||||||
| Assigned To | Kloot | ||||||||
| Priority | normal | Severity | minor | Reproducibility | always | ||||
| Status | closed | Resolution | reopened | ||||||
| Product Version | 103.0 +git | ||||||||
| Target Version | Fixed in Version | ||||||||
| Summary | 0005637: 103.0.1-1166 no shadows or dynlights with win64 | ||||||||
| Description | Works with 32 bit spring. | ||||||||
| Tags | No tags attached. | ||||||||
| Checked infolog.txt for Errors | |||||||||
| Attached Files |
| ||||||||
 Relationships |
 Notes |
|
|
Kloot (developer) 2017-07-08 14:08 |
win64 buildbot GLEW (1.13.0) should be updated; it pretends to support clipctrl whereas the more recent win32 1.5.8 does not. |
|
hokomoko (developer) 2017-07-08 20:56 |
I wonder why the win32 one used is so old |
|
hokomoko (developer) 2017-07-08 21:11 |
I think there was some header-binary inconsistency, tried to fix that, will see. |
|
Kloot (developer) 2017-07-08 21:40 |
ah, nasty. I think win32 glew wasn't touched since the static builds were introduced. |
|
hokomoko (developer) 2017-07-09 00:30 Last edited: 2017-07-09 00:30 |
Should be fixed in 103.0.1-1168-g70c3340 I'm going to update win32 glew as well and see what happens |
|
hokomoko (developer) 2017-07-09 11:37 |
This didn't help it seems. I'm starting to suspect it's not glew lying but something else. Any ideas of how to test? |
|
Kloot (developer) 2017-07-09 12:14 |
what's in the infolog? one other possibility is that ENABLE_CLIP_CONTROL isn't defined in Camera.cpp which would be weird. |
|
hokomoko (developer) 2017-07-09 12:16 |
It's worth mentioning that on my machine I have lights and shadows and clip control in infolog |
|
Kloot (developer) 2017-07-09 12:32 |
could there be a similar header-binary mismatch with buildbot's GL version? |
|
hokomoko (developer) 2017-07-09 12:43 |
what headers/binaries do you mean? https://github.com/spring/mingwlibs64 |
|
Kloot (developer) 2017-07-09 13:01 |
nm, mingwlibs doesn't include custom GL headers. it would be useful to know the actual GL version installed. |
|
hokomoko (developer) 2017-07-09 13:07 |
I still don't understand why it works on my machine and not on GoogleFrog's I'll attach my infolog. |
|
hokomoko (developer) 2017-07-09 13:12 |
Also attached an infolog given by google on latest engine |
|
Kloot (developer) 2017-07-09 13:26 |
think I see the problem now, standby |
|
Kloot (developer) 2017-07-09 13:33 Last edited: 2017-07-09 13:34 |
to be absolutely sure my hypothesis is correct: does it still not work for GF with the 1169 build from which infolog_google64.txt was generated? |
|
hokomoko (developer) 2017-07-09 13:51 |
http://paste.springfiles.com/view/c95a43eb |
|
Google_Frog (reporter) 2017-07-09 15:20 |
Also, the dynlights patch breaks them for me with win64. Patch: https://github.com/ZeroK-RTS/Zero-K/commit/0693d9c187f5b544a881de9975731ea0200511d8 |
|
Kloot (developer) 2017-07-09 15:28 |
Fix 0c38a9bce3144895a117f18fe66699f53d3e10e3 committed to develop branch: fix 0005637 ext-check issues, repo: spring changeset id: 8477 |
|
Kloot (developer) 2017-07-09 17:06 |
GoogleFrog: please test https://springrts.com/dl/buildbot/default/develop/103.0.1-1170-g0c38a9b/win64/ |
|
Google_Frog (reporter) 2017-07-10 01:58 |
No shadows or dynlights with 103.0.1-1170-g0c38a9b. |
|
Kloot (developer) 2017-07-10 02:14 Last edited: 2017-07-10 03:10 |
I'm out of ideas for a while; in the meantime there is a ForceDisableClipCtrl option in bce9931d. if you feel like it you might also try changing to hokomoko's driver version (384.76). just out of curiosity, why did you set Shadows=4? |
|
Google_Frog (reporter) 2017-07-10 02:35 |
I was toggling through the shadows options. |
|
Google_Frog (reporter) 2017-07-10 07:37 |
Setting ForceDisableClipCtrl = 1 works for both dynlights and shadows. |
|
hokomoko (developer) 2017-07-10 10:09 |
Can you try running with DebugGL=1 and ForceDisableClipCtrl=0 and posting your infolog? |
|
Kloot (developer) 2017-07-10 14:34 |
hmm, "GLEW version: 1.12.0"? |
|
hokomoko (developer) 2017-07-10 14:36 |
It shouldn't be the issue IMO. GoogleFrog has a more recent card than I do and everything works on my machine. Maybe it's related to some ZK settings which are different for me and for GoogleFrog? different widgets? |
|
Kloot (developer) 2017-07-10 14:41 Last edited: 2017-07-10 14:43 |
IMO it's either a library or a driver difference at this point, no shader errors (or any relevant GL debug output) and afaik there are no widgets in ZK that could be of influence. also, the 32-bit build shouldn't work for GF either in case of a game-related setting. |
|
hokomoko (developer) 2017-07-10 14:47 Last edited: 2017-07-10 14:47 |
btw, I'm not entirely sure how clip control affects deferred rendering and/or shadows. Also, I think that GF had dynamic lights (but not shadows) working when using a ZK version from before https://github.com/ZeroK-RTS/Zero-K/commit/0693d9c187f5b544a881de9975731ea0200511d8 (v1.5.7.1 I think?) GoogleFrog can you verify this? |
|
Google_Frog (reporter) 2017-07-10 15:11 |
Yes, the dynlights patch for ZK breaks dynlights. |
|
Kloot (developer) 2017-07-10 15:51 Last edited: 2017-07-10 15:56 |
correction: it fixes them based on the assumption clipctrl works. the math behind CC is trivial: A*B has to map to [0,1], where A is a projected z-value in [-1,1] in standard GL and a value in [0,1] otherwise. B is the part controlled by the driver and equal to *0.5+0.5 in standard GL. Spring adjusts all A's if CC is available by factoring B into them, and tells the driver to set B to 1. this last part fails for reasons unknown. |
|
hokomoko (developer) 2017-07-10 22:09 |
Yes, I'm familiar with the clipping. Why does it only affect shadows and dynamic lights though? Are they mapped into [-1,0)? Why is everything else rendered correctly? |
|
Kloot (developer) 2017-07-11 01:45 |
because those two techniques involve inverse-projecting z's which are only in the proper range if the driver sets B as it is told. I'm putting this in suspended animation until the glew mess is cleared up. |
| Notes |
| Issue History |
|||
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2017-07-08 06:00 | Google_Frog | New Issue | |
| 2017-07-08 06:00 | Google_Frog | File Added: infolog32.txt | |
| 2017-07-08 06:00 | Google_Frog | File Added: infolog64.txt | |
| 2017-07-08 14:08 | Kloot | Note Added: 0017955 | |
| 2017-07-08 20:56 | hokomoko | Note Added: 0017956 | |
| 2017-07-08 21:11 | hokomoko | Note Added: 0017957 | |
| 2017-07-08 21:40 | Kloot | Note Added: 0017958 | |
| 2017-07-08 23:33 | hokomoko | Assigned To | => hokomoko |
| 2017-07-08 23:33 | hokomoko | Status | new => resolved |
| 2017-07-08 23:33 | hokomoko | Resolution | open => fixed |
| 2017-07-09 00:30 | hokomoko | Note Added: 0017959 | |
| 2017-07-09 00:30 | hokomoko | Note Edited: 0017959 | View Revisions |
| 2017-07-09 11:37 | hokomoko | Assigned To | hokomoko => |
| 2017-07-09 11:37 | hokomoko | Status | resolved => new |
| 2017-07-09 11:37 | hokomoko | Resolution | fixed => reopened |
| 2017-07-09 11:37 | hokomoko | Note Added: 0017962 | |
| 2017-07-09 12:14 | Kloot | Note Added: 0017964 | |
| 2017-07-09 12:16 | hokomoko | Note Added: 0017965 | |
| 2017-07-09 12:32 | Kloot | Note Added: 0017966 | |
| 2017-07-09 12:43 | hokomoko | Note Added: 0017967 | |
| 2017-07-09 13:01 | Kloot | Note Added: 0017968 | |
| 2017-07-09 13:07 | hokomoko | Note Added: 0017969 | |
| 2017-07-09 13:10 | hokomoko | File Added: infolog_hoko64.txt | |
| 2017-07-09 13:11 | hokomoko | File Added: infolog_google64.txt | |
| 2017-07-09 13:12 | hokomoko | Note Added: 0017970 | |
| 2017-07-09 13:26 | Kloot | Note Added: 0017971 | |
| 2017-07-09 13:33 | Kloot | Note Added: 0017972 | |
| 2017-07-09 13:34 | Kloot | Note Edited: 0017972 | View Revisions |
| 2017-07-09 13:51 | hokomoko | Note Added: 0017973 | |
| 2017-07-09 14:01 | Kloot | Assigned To | => Kloot |
| 2017-07-09 14:01 | Kloot | Status | new => assigned |
| 2017-07-09 15:20 | Google_Frog | File Added: infolog_1169_win64.txt | |
| 2017-07-09 15:20 | Google_Frog | Note Added: 0017974 | |
| 2017-07-09 15:28 | Kloot | Changeset attached | => spring develop 0c38a9bc |
| 2017-07-09 15:28 | Kloot | Note Added: 0017975 | |
| 2017-07-09 15:28 | Kloot | Status | assigned => resolved |
| 2017-07-09 15:28 | Kloot | Resolution | reopened => fixed |
| 2017-07-09 17:06 | Kloot | Note Added: 0017977 | |
| 2017-07-09 17:10 | Kloot | Status | resolved => feedback |
| 2017-07-09 17:10 | Kloot | Resolution | fixed => reopened |
| 2017-07-10 01:58 | Google_Frog | File Added: infolog.txt | |
| 2017-07-10 01:58 | Google_Frog | Note Added: 0017979 | |
| 2017-07-10 01:58 | Google_Frog | Status | feedback => assigned |
| 2017-07-10 02:14 | Kloot | Note Added: 0017980 | |
| 2017-07-10 02:27 | Kloot | Note Edited: 0017980 | View Revisions |
| 2017-07-10 02:35 | Google_Frog | Note Added: 0017981 | |
| 2017-07-10 03:10 | Kloot | Note Edited: 0017980 | View Revisions |
| 2017-07-10 07:37 | Google_Frog | Note Added: 0017982 | |
| 2017-07-10 10:09 | hokomoko | Note Added: 0017985 | |
| 2017-07-10 10:35 | Google_Frog | File Added: infologDebugGL.txt | |
| 2017-07-10 14:34 | Kloot | Note Added: 0017990 | |
| 2017-07-10 14:36 | hokomoko | Note Added: 0017991 | |
| 2017-07-10 14:41 | Kloot | Note Added: 0017992 | |
| 2017-07-10 14:43 | Kloot | Note Edited: 0017992 | View Revisions |
| 2017-07-10 14:47 | hokomoko | Note Added: 0017993 | |
| 2017-07-10 14:47 | hokomoko | Note Edited: 0017993 | View Revisions |
| 2017-07-10 15:11 | Google_Frog | Note Added: 0017994 | |
| 2017-07-10 15:51 | Kloot | Note Added: 0017995 | |
| 2017-07-10 15:53 | Kloot | Note Edited: 0017995 | View Revisions |
| 2017-07-10 15:56 | Kloot | Note Edited: 0017995 | View Revisions |
| 2017-07-10 22:09 | hokomoko | Note Added: 0018006 | |
| 2017-07-11 01:45 | Kloot | Status | assigned => closed |
| 2017-07-11 01:45 | Kloot | Note Added: 0018007 | |
| 2020-05-14 15:24 | abma | Relationship added | related to 0006394 |
| Issue History |