Commit graph

211 commits

Author SHA1 Message Date
Christoph Oelckers
c773a500f2 - only activate shader storage buffers on modern drivers also supporting GL_ARB_buffer storage. Early adopters of this extension (i.e. older AMD drivers mainly) tend to implement this badly. 2014-09-17 09:01:16 +02:00
Christoph Oelckers
a25ed3c807 - fixed: inverted rendering for fuzzy things with light powerup was switched off due to change of render style. 2014-09-15 12:15:19 +02:00
Christoph Oelckers
8e7e16f73a - fixed: The light uniform buffer may not be mapped with GL_MAP_INVALIDATE_BUFFER_BIT, because it needs to be mapped for each portal in a scene but it must preserve the existing data for the remaining translucent objects. 2014-09-15 10:27:09 +02:00
Christoph Oelckers
32f08adaf3 - moved some code to better places.
- allow GL version 3.0 in Windows, too.
2014-09-14 23:01:57 +02:00
Ralgor
5cc43137a1 Only require OpenGL 3.0 compatibility profile. 2014-09-14 14:43:42 -05:00
Ralgor
cfc8f3dbbf Global GL render context shouldn't be initialized inside of gl_PrintStartupLog, since it's not compiled in non MSC builds. 2014-09-14 14:42:14 -05:00
Ralgor
1a70a6aabc The light buffer should check for shader_storage_buffer_object rather than buffer_storage. 2014-09-14 14:29:13 -05:00
Ralgor
ddf58b43c9 Fix compile errors on linux. 2014-09-14 14:28:05 -05:00
Christoph Oelckers
3c2f1952fd Merge branch 'master' into v2.x 2014-09-12 21:01:55 +02:00
Christoph Oelckers
25951362fc - fixed: When applying Boom's Transfer_Heights effect to a sector, gl_FakeFlat needs to remove all portals from the original sector planes that are being replaced in the sector copy. 2014-09-11 13:02:39 +02:00
Christoph Oelckers
acf6c259d8 - changed the handling of alpha textures. The only special case they need is with palette-less textures and this can be handled far more easily and robustly with a predefined translation instead of passing another parameter through all the layers of the texture management code. This also fixes problems with paletted PNGs that get used as an alpha texture because the old method clobbered the image's palette. 2014-09-09 13:21:36 +02:00
Christoph Oelckers
4bb320a27c - route texture binding through the renderstate class for better control. Currently it's just a direct passthrough but this will change. 2014-09-09 12:00:42 +02:00
Christoph Oelckers
d5633701b4 - swapped order of textures in burn shader to avoid some problems with the texture samplers.
- fixed: texture sampler state for the burn texture was never set.
2014-09-09 10:17:44 +02:00
Christoph Oelckers
c6f4c0b6f0 - fixed: FMaterial's tex pointer could be accessed before it was set.
- allow more than two texture units in shaders.
2014-09-09 10:03:34 +02:00
Christoph Oelckers
62880f113b - fix a render glitch with Back to Saturn X MAP06: Do not flood missing upper and lower textures with the backsector's flat if that backsector is malformed (i.e. has no area.) 2014-09-09 08:47:39 +02:00
Christoph Oelckers
9b8869e78d Merge branch 'master' into v2.x
Conflicts:
	src/gl/textures/gl_material.cpp
2014-09-09 01:30:11 +02:00
Christoph Oelckers
86d9c7ec8e - add some compatibility settings to fix rendering glitches in BTSX_E1 MAP12. 2014-09-09 01:27:41 +02:00
alexey.lysiuk
e29fce6951 Fixed missing transparency on upscaled textures
Textures with diagonal patterns were treated as opaque after resizing
Images upscaled by hqNx were affected mostly by this issue

http://forum.drdteam.org/viewtopic.php?f=24&t=5370
http://zandronum.com/tracker/view.php?id=269
http://zandronum.com/tracker/view.php?id=315
2014-09-07 11:52:51 +03:00
Christoph Oelckers
b96dd6c421 - fixed: The missing fourth component of the texture coordinate must be filled with 1.0, not 0.0 before applying the texture matrix. Not doing so will cancel out the translation part of the matrix. 2014-09-02 10:31:48 +02:00
Christoph Oelckers
fa3a62e954 - fix a render glitch with Back to Saturn X MAP06: Do not flood missing upper and lower textures with the backsector's flat if that backsector is malformed (i.e. has no area.) 2014-08-31 23:01:53 +02:00
Christoph Oelckers
a280c20b4e - fixed: If we want to cache texture binding state we have to reset it in all places where a texture becomes unbound. 2014-08-31 19:00:17 +02:00
Christoph Oelckers
12160bd29c - remove some obsolete code from decal rendering.
- avoid rebinding the same texture multiple times, as there's considerable overhead in the texture manager.
- check gl_sort_textures only once per scene, not per draw list.
2014-08-30 15:34:14 +02:00
Christoph Oelckers
6a3cd6378a - found out that reading the CPU's real time clock costs a not insignificant amount of time so this is now only done when either the benchmark command is running or the rendertimes are shown. 2014-08-30 14:33:06 +02:00
Christoph Oelckers
49ec7beb8f - separate draw lists for walls and flats. This makes the sorting much more efficient because draw types no longer need to be checked in the compare function. This is a lot more important than having perfect texture order. 2014-08-30 13:04:41 +02:00
Christoph Oelckers
a903cbe12e - sorting draw items by light level no longer makes sense so remove all corresponding code from dicmp. 2014-08-24 13:10:45 +02:00
Christoph Oelckers
bf03d02228 - print OpenGL profile type in startup log. 2014-08-24 01:09:44 +02:00
Christoph Oelckers
904cc2e158 - some code cleanup. 2014-08-23 18:54:24 +02:00
Christoph Oelckers
bf6079af46 - fixed incorrect check for overrideshader. 2014-08-23 00:47:05 +02:00
Christoph Oelckers
1050013017 major cleanup of the texture manager:
- use sampler objects to avoid creating up to 4 different system textures for one game texture just because of different clamping settings.
- avoids flushing all textures for change of texture filter mode.
- separate sprite and regular dimensions on the material level to have better control over which one gets used. It's now an explicit parameter of ValidateTexture. The main reason for this change is better handling of wall sprites which may not be subjected to such handling.
- create mipmaps based on use case, not texture type.
- allows removal of FCloneTexture hack for proper sharing of the same sprite for decals and other purposes.
- better precaching of skyboxes.
2014-08-22 23:50:38 +02:00
Christoph Oelckers
274a4216ea - disabling inlining in the GL loader produces an executable that's 8kb smaller. 2014-08-21 11:47:53 +02:00
Christoph Oelckers
e132fc5eed - replaced GLEW with GLLoadGen for GL access. This allows to have a header that only contains what's actually required, namely OpenGL 3.3 plus glBegin and glEnd which are the only compatibility functions needed for the fallback render path.
GLEW has two major problems:

- it always includes everything, there is no way to restrict the header to a specific GL version
- it is mostly broken with a core profile and only works if all sanity checks get switched off.
2014-08-21 11:02:46 +02:00
Christoph Oelckers
6f65bccf1c - reinstated the far superior assembly HQnX version for Visual C++. 2014-08-20 12:45:33 +02:00
Christoph Oelckers
86d37e06f9 - lowered requirements of GL 2.x to OpenGL 3.3.
There was one issue preventing the previous 2.0 betas from running under GL 3.x: The lack of persistently mapped buffers.
For the dynamic light buffer today's changes take care of that problem.
For the vertex buffer there is no good workaround but we can use immediate mode render calls instead which have been reinstated.

To handle the current setup, the engine first tries to get a core profile context and checks for presence of GL 4.4 or the GL_ARB_buffer_storage extension.
If this fails the context is deleted again and a compatibility context retrieved which is then used for 'old style' rendering which does work on older GL versions.

This new version does not support GL 3.2 or lower, meaning that Intel GMA 3000 or lower is not supported. The reason for this is that the engine uses a few GL 3.3 features which are not present in the latest Intel driver.
In general the Intel GMA 3000 is far too weak, though, to run the demanding shader of GZDoom 2.x, so this is no real loss. Performance would be far from satisfying.

A command line option '-gl3' exists to force the fallback render path. On my Geforce 550Ti there's approx. 10% performance loss on this path.
2014-08-19 15:56:33 +02:00
Christoph Oelckers
a2dc4afe3f - screwed by the editor's autocompletion... (wrong GL flag was used...) 2014-08-19 14:25:47 +02:00
Christoph Oelckers
00d7707aef - allow reallocation of light buffer if more lights are needed.
- added a light preprocessing pass to the renderer so that a non-persistent buffer can be used with minimal mapping/unmapping. This only gets used if necessary because it adds some overhead to the renderer.
2014-08-19 14:18:21 +02:00
Christoph Oelckers
38796e7714 - removed some obsolete and useless GL calls. 2014-08-17 11:41:03 +02:00
Christoph Oelckers
b9ffb51d0c - small but important optimization: Two-sided lines with both sides in the same sector don't really require vertex splitting for precise rendering. 2014-08-04 23:00:40 +02:00
Christoph Oelckers
09a68e8c93 - fixed incorrect function call in shader compiler. 2014-08-03 18:14:19 +02:00
Christoph Oelckers
19cfffebb3 - fixed: the WallTypes enum contained a value that no longer was valid. This was fixed orignally last week but it seems to have gotten lost. 2014-08-03 12:21:05 +02:00
Christoph Oelckers
c98e3ca99d Merge branch 'master' into v2.x 2014-08-03 10:58:41 +02:00
Christoph Oelckers
b3e0d93385 - implemented rendering of wall sprites. 2014-08-03 10:57:58 +02:00
Christoph Oelckers
398d902e7e - split off sprite clipping adjustment code into its own function. 2014-08-03 10:22:12 +02:00
Christoph Oelckers
a63871d170 - at least for Intel GMA we need shaders without 'discard' to render non-transparent stuff. The performance penalty is rather hefty here. 2014-08-02 21:06:34 +02:00
Christoph Oelckers
a97b58fa27 - added check for light uniform buffer overflows, because uniform buffers on Intel are rather small. 2014-08-02 20:41:13 +02:00
Christoph Oelckers
e35fefdc06 - better rebind the active shader after updating the matrices. 2014-08-02 11:59:04 +02:00
Christoph Oelckers
b2860a1d63 - it looks like glProgramUniform is not working correctly with Intel drivers, so better forget about it for setting the view and projection matrices. Even on NVidia the time difference can only be measured in microseconds per frame so it's not a big loss. 2014-08-02 11:57:42 +02:00
Christoph Oelckers
a8e9c1832f - decided to restrict the 2.0 beta to OpenGL 4.x with GL_ARB_buffer_storage extension and removed all code for supporting older versions.
Sadly, anything else makes no sense.
All the recently made changes live or die, depending on this extension's presence.
Without it, there are major performance issues with the buffer uploads. All of the traditional buffer upload methods are without exception horrendously slow, especially in the context of a Doom engine where frequent small updates are required.
It could be solved with a complete restructuring of the engine, of course, but that's hardly worth the effort, considering it's only for legacy hardware whose market share will inevitably shrink considerably over the next years.
And even then, under the best circumstances I'd still get the same performance as the old immediate mode renderer in GZDoom 1.x and still couldn't implement the additions I'd like to make.

So, since I need to keep GZDoom 1.x around anyway for older GL 2.x hardware, it may as well serve for 3.x hardware, too. It's certainly less work than constantly trying to find workarounds for the older hardware's limitations that cost more time than working on future-proofing the engine.

This new, trimmed down 4.x renderer runs on a core profile configuration and uses persistently mapped buffers for nearly everything that is getting transferred to the GPU. (The global uniforms are still being used as such but they'll be phased out after the first beta release.
2014-08-01 22:42:39 +02:00
Christoph Oelckers
7967082e60 - use the light buffer to handle dynamic lighting. 2014-08-01 20:59:39 +02:00
Christoph Oelckers
1ec58011d2 - start of light buffer implementation so that we don't have to use uniform arrays which appear to be broken on AMD. 2014-07-31 00:44:22 +02:00
Christoph Oelckers
ef8f66c9a1 - removed the code for hardware alpha testing again because it didn't work anymore with how things are set up now.
- we need to check all GL versions when trying to get a context because some drivers only give us the version we request, leaving out newer features that are not exposed via extension.
- added some status info about uniform blocks.
2014-07-30 23:13:16 +02:00