Commit Graph

310 Commits

Author SHA1 Message Date
Shyotl
a1eb30ae87 Keep old non-swizzle method around since GL_ARB_texture_swizzle has poor support prior to 3.3 (on osx especially) 2014-05-17 18:25:21 -05:00
Shyotl
0a89503edd Renamed glUniformMatrix3x4fvARB to glUniformMatrix3x4fv, since I don't think it ever had the ARB postfix. Also use its the properly named (although identical to the current) function pointer signature. 2014-05-17 04:04:33 -05:00
Shyotl
67c8ac2b04 Skinned shaders were exceeding maximum amount of vertex uniforms on amd hardware.
-Changed the transform matrix uniform to a 3x4 matrix and packed translation into it to free up uniforms. (3x3 is converted to 3x4 internally, so we were needlessly eating 3*52 extra uniform slots. translationPalette might also have been treated as a vec4 internally too, wasting 52 more slots.)
-matrix3x4 requires opengl2.1 and newer, so added a new featuretable mask.
-Also added a featuretable mask to disable hardware skinning and deferred shading on hardware with less than 1024 vertex uniforms.
NOTE: On old old old amd hardware, evidently a 3x4 matrix might be upgraded to 4x4. I'm unsure, but I doubt such hardware has 1024+ uniform components available to begin with. 4x3 supposedly doesn't do this, but opengl is column-major, so this makes little sense.
2014-05-17 03:31:45 -05:00
Shyotl
117eca3c98 Try some swizzling instead of forcing everything to rgba8 with core profile. Avoids manual conversions with scratch buffers. Works for all pixel datatypes. Cleaner. Uses less vram.
Also, fixed textures when running core pofile.
2014-05-14 03:05:53 -05:00
Shyotl
98f9a09b39 sunlight_color_copy need only be vec3. 2014-05-08 16:39:41 -05:00
Shyotl
7e8e8bdb26 Disable some diagnostic gunk eating memory. 2014-05-08 16:35:12 -05:00
Shyotl
e2a0f801b9 Have LLRender::color4ub fall back to uniform if bound vbo lacks color attribute. Fixed fixed-function call in LLFacePool::LLOverrideFaceColor::setColor. 2014-05-08 16:29:53 -05:00
Shyotl
7158a3c83d Fixed an annoying showstopper regarding rigged attachment vertex corruption. Rigged meshes have can have poor weighting, causing potential div-by-zero issues. 2014-05-08 15:11:12 -05:00
Shyotl
d6fdaa93a4 Added adaptive vsync support (selectable in preferences->graphics->hardware, provided the driver supports it). Also hide 16x fsaa on amd hardware, as it is not supported on said hardware (unsure about intel). 2014-03-28 02:34:03 -05:00
Shyotl
f7fdeeea9b Be super sure not to let rigged vert weights extend beyond actual joint count. 2014-02-10 20:18:39 -06:00
Latif Khalifa
8ea156d128 Merge branch 'master' of git://github.com/Shyotl/SingularityViewer 2014-01-24 02:20:37 +01:00
Shyotl
a0c7b1f8ac Try manually nulling out vbo data if its entire contents are to be overwritten. Rumored intel/nvidia optimization. 2014-01-23 19:06:52 -06:00
Latif Khalifa
269deb56e6 Fitted mesh 2014-01-24 01:11:27 +01:00
Inusaito Sayori
75109e14c3 Italics hotfix
Restores use of the old slant_amt offset for italicizing fonts by slanting quads.
Note that this is the wrong way to fix the issue and should eventually be removed.
2014-01-15 20:15:53 -05:00
Latif Khalifa
bc6348bb34 Merge branch 'master' of git://github.com/Shyotl/SingularityViewer 2013-12-25 18:08:42 +01:00
Shyotl
3da9fe0e94 Avoid calling a few deprecated gl functions when shaders are enabled. 2013-12-25 02:09:36 -06:00
Shyotl
75f94d346d Fixed Alerics favorite assertion. 2013-12-24 20:21:55 -06:00
Shyotl
4eb9dc5191 Use GL_STREAM_DRAW_ARB for post-process fullscreen quad. 2013-12-24 20:17:05 -06:00
Shyotl
033880c8f9 Kill some vertex attribute / shader attribute inconsistencies. (Issue 1287, 1281) Also hopefully improved face edit overlay behavior when editing normal/specular texture/scale/offset/rotation. 2013-12-24 20:15:29 -06:00
Siana Gearz
2bb2e05ef9 Merge branch 'master' of https://github.com/Shyotl/SingularityViewer 2013-12-15 14:56:20 +01:00
Shyotl
7da8b012e6 Pulled SSAO out into its own shader to support running at a lower-than-native framebuffer resolution (defaults at half-resolution [SHRenderSSAOResolutionScale=.5] ). This doesn't improve the slow ssao/shadow softening, however, as that bit remains unchanged. 2013-12-13 02:19:57 -06:00
Inusaito Sayori
be369b5500 Merge branch 'master' of https://github.com/singularity-viewer/SingularityViewer 2013-12-04 14:05:46 -05:00
Inusaito Sayori
23958965aa LLInitParam/LLUI Params enhancement
Specialized param handling for common linden types
And pre-viewer-interesting modernized llinitparam.* (includes llsdparam.cpp update)
2013-12-02 00:27:50 -05:00
Siana Gearz
9c29488044 If the text is too skinny to render, don't crash 2013-11-20 10:33:15 +01:00
Siana Gearz
060628397b Merge branch 'master' of https://github.com/Shyotl/SingularityViewer 2013-11-20 05:50:03 +01:00
Shyotl
3715b7439d Doer fonts fast(er)! (better batching for font glyph rendering) 2013-11-19 19:40:29 -06:00
Siana Gearz
07736106b1 Suppress asserts in fallback font loading 2013-11-20 02:20:58 +01:00
Shyotl
5d2511e39e More font cleanup and updating. 2013-11-19 17:56:27 -06:00
Shyotl
b9d2ad3132 Merge branch 'master' of https://github.com/singularity-viewer/SingularityViewer.git 2013-11-18 10:38:39 -06:00
Shyotl
8aa4d0fd27 Removed fasttimer that was hurting perf a tiny bit. 2013-11-18 09:58:21 -06:00
Shyotl
be249ac8a1 Font update. Minor refactor, and removed some pointless code that was leaking. 2013-11-18 09:56:49 -06:00
Siana Gearz
a5103a1fc1 Sprinkle some Intel GPU support on Linux 2013-11-18 04:22:21 +01:00
Siana Gearz
749d0eec28 Merge https://github.com/abaph/SingularityViewer
Brings in support for MESA accelerators.
2013-11-17 16:38:06 +01:00
Shyotl
01e27eac76 Kill a tiny bit of dead code. 2013-11-13 13:06:47 -06:00
Andros Baphomet
997fdebb9f Re-fix shader compilation errors 2013-11-03 13:12:27 -05:00
Andros Baphomet
fe0a761ce6 Merge remote-tracking branch 'upstream/master'
Conflicts:
	indra/newview/app_settings/shaders/class1/deferred/alphaNonIndexedNoColorF.glsl
	indra/newview/app_settings/shaders/class1/deferred/fullbrightShinyF.glsl
	indra/newview/app_settings/shaders/class1/deferred/fxaaF.glsl
	indra/newview/app_settings/shaders/class1/deferred/giF.glsl
	indra/newview/app_settings/shaders/class1/deferred/multiSpotLightF.glsl
	indra/newview/app_settings/shaders/class1/deferred/softenLightF.glsl
	indra/newview/app_settings/shaders/class1/deferred/spotLightF.glsl
	indra/newview/app_settings/shaders/class2/deferred/alphaF.glsl
	indra/newview/app_settings/shaders/class2/deferred/alphaNonIndexedF.glsl
	indra/newview/app_settings/shaders/class2/deferred/alphaNonIndexedNoColorF.glsl
	indra/newview/app_settings/shaders/class2/deferred/multiSpotLightF.glsl
	indra/newview/app_settings/shaders/class2/deferred/spotLightF.glsl
2013-11-03 12:17:50 -05:00
Shyotl
77eb928380 Add LLGLManager::mIsMobileGF 2013-11-02 03:33:21 -05:00
Shyotl
af8f8bb040 Mats. 2013-11-02 01:20:05 -05:00
Andros Baphomet
c009866d86 Merge remote-tracking branch 'upstream/master' 2013-10-30 09:42:58 -04:00
Shyotl
5e8c7264da Fix artifacts from local stash merge. 2013-10-26 19:58:54 -05:00
Shyotl
30c5b9514e Let the rendertarget dictate internal format when resizing. 2013-10-26 18:09:34 -05:00
Shyotl
8e55d5499e Occlusion FBOs in deferred. 2013-10-26 18:07:56 -05:00
Shyotl
fc04f6b2e1 Added handling of GL_LUMINANCE pixel format to setManualImage. Was missing, although GL_LUMINANCE_ALPHA and GL_ALPHA were handled.
Also added missing file hurr.
2013-10-26 16:42:41 -05:00
Shyotl
2a432f73da Try and squelch some warnings regarding nightvision post shader on amd gpus. 2013-10-26 16:36:38 -05:00
Shyotl
f4863527aa Migrate ui scale/translation transform matrix stacks to LLAlignedArray. 2013-10-26 16:29:38 -05:00
Shyotl
8441bc80a4 Use LLStaticHashedString for faster uniform lookup. 2013-10-26 16:26:46 -05:00
Andros Baphomet
c264607207 Merge remote-tracking branch 'upstream/master' 2013-10-19 10:03:17 -04:00
Shyotl
6fd73ee8bb Use GL_RED instead of GL_LUMINANCE for post-process noise texture. 2013-10-17 22:07:58 -05:00
Shyotl
19d7186508 Merge branch 'master' of git://github.com/siana/SingularityViewer.git 2013-10-17 21:08:30 -05:00
Shyotl
d18fa5ba42 Modified LLRenderTarget to handle sCurFBO and mPreviousFBO as pointers instead of arbiturary indexes. 2013-10-16 21:35:04 -05:00