Deve
cc92ee6ef3
One more tweak in SSAO.
...
Just use mod(x, 6.283185307179586) as suggested by mesa developer, instead converting it to degrees and back to radians.
2016-05-26 11:48:57 +02:00
Deve
cee0de3af3
Fixed a bug in my previous commit.
...
I forgot that sin and cos functions want angle value in radians.
Now the code looks a bit ugly, but generates proper values.
2016-05-26 02:39:35 +02:00
Deve
cdae595724
Fixed SSAO on intel.
...
In ssao.frag we are computing theta valaue and it looks that intel driver doesn't like too big values for sinus and cosinus computations.
I just used modulo 360 to store lower angle values in theta variable.
2016-05-25 23:16:01 +02:00
Vincent Lejeune
cfbbeef441
Clamp lod level in SSAO shader later.
2015-01-12 22:47:53 +01:00
Vincent Lejeune
95e2213b1e
SSAO can me 0 sometimes
2015-01-12 20:13:11 +01:00
Vincent Lejeune
b8ab68b31a
Fix precision issue in SSAO
2015-01-12 20:05:40 +01:00
vlj
496269e73b
Make SSAO param changeable.
2014-08-09 18:51:52 +02:00
Vincent Lejeune
58c094917e
Use an header included in every shaders.
2014-08-02 00:50:56 +02:00
vlj
40a7ab0780
Raise one of ssao tolerance value.
2014-07-17 00:12:54 +02:00
Vincent Lejeune
d7139920ed
Fix position reconstruction formula in ssao
2014-07-12 19:54:52 +02:00
vlj
d2fc258124
Tweak ssao (again)
...
Random factor is less perdicable, no more spiral.
2014-06-23 01:03:55 +02:00
vlj
8f5dd6edae
Some tweak to ssao.
2014-06-22 18:04:46 +02:00
vlj
e5a5c78045
Use a big triangle instead of 2 to cover full screen.
...
This allows to rasterize a single primitive instead of two, and avoid
trashing the cache between the 2 triangles drawing.
2014-06-03 19:56:59 +02:00
vlj
e571afc43b
Speed up SSAO again
...
Use log2(xy) = log2(x) + log2(y), and approximate log2(alpha) =
log2(previous alpha) + .5.
Log2 is indeed a costly instruction.
2014-06-03 03:41:05 +02:00
vlj
164ffa2b4f
Spped up SSAO
...
Avoid using cos/sin too much, use sum of cos/sin formula instead.
2014-06-03 03:14:39 +02:00
Stephen Just
f4992b80fc
Fix errors when compiling shaders with UBOs disabled
2014-05-19 09:57:31 -06:00
vlj
883b39591f
Width and height are passed by UBO
2014-05-17 02:39:55 +02:00
vlj
229b461a6c
SSAO: Less abrupt transition
2014-05-14 03:26:43 +02:00
vlj
1d4bcd5ade
SSAO: Increase beta constant to avoid error
2014-05-14 02:27:17 +02:00
vlj
86c8b9f83d
Some tweaking to make ssao more apparent
2014-05-14 02:12:18 +02:00
Vincent Lejeune
59484ea09a
DS buffer is not linear and cant be mipmapped
...
Use the method provided by SAO paper (ie convert to linear and then
mipmap)
2014-05-13 19:40:30 +02:00
vlj
26d48cdad0
SSAO: Stronger blur + tweak param
...
Use another algorithm to have gaussian blur that lets us customise
radius more easily.
2014-05-12 23:59:17 +02:00
vlj
8e17ae90b6
Resolution independent ssao
2014-05-12 20:13:12 +02:00
vlj
23ba597a1c
SSAO: Use int rounding
2014-05-12 18:39:33 +02:00
vlj
229b72f84a
Render SSAO at fullscreen resolution.
2014-05-12 01:06:33 +02:00
vlj
4fcec02123
Remove ssao depth guard.
2014-05-12 00:34:44 +02:00
vlj
7e1625da5e
Tweak SSAO params.
2014-05-12 00:30:36 +02:00
vlj
f8544493b6
Improve SSAO perf
2014-05-12 00:14:50 +02:00
vlj
11b7d20ad8
Use the mipmap trick to speed up ssao.
2014-05-12 00:03:42 +02:00
Vincent Lejeune
567974aeed
Use dFdX/dFdY to compute normals in SSAO.
2014-05-11 18:09:48 +02:00
Vincent Lejeune
e5410e23a2
Use Alchemy SSAO
2014-05-11 18:00:30 +02:00
vlj
2a1623d8ca
Use a more efficient screen to view conversion
...
Also use the opportunity to use UBO where possible.
2014-05-02 18:11:34 +02:00
Vincent Lejeune
da15aebcda
SSAO: Clamp to 0.
2014-04-22 18:22:39 +02:00
vlj
878a639a84
Another ssao tweak.
2014-04-20 02:56:07 +02:00
Vincent Lejeune
123a192963
Use half res FBO for SSAO
...
SSAO is a medium frequency effect, using half res fbo is fine and
improve performance and look by doing an additionnal bilinear filtering.
Also tune up some values.
2014-04-14 01:01:36 +02:00
Vincent Lejeune
01b88fde9d
Factorize DecodeNormal from misc shaders.
2014-03-21 18:17:21 +01:00
Vincent Lejeune
da4896c3d8
SSAO: tweak fade distance.
2014-03-01 22:50:22 +01:00
Vincent Lejeune
af862cb6c5
#version autoadded to shaders and turn some v120
2014-02-28 17:29:05 +01:00
Vincent Lejeune
0243546e4d
Use an openGL 3.3 compatibility context when available.
2014-02-12 22:16:10 +01:00
Vincent Lejeune
f50e6f817d
Use another normal encoding method from a crytek slide.
2014-01-27 23:29:46 +01:00
Vincent Lejeune
0b531b20f7
SSAO: Use depth buffer.
2014-01-27 20:24:57 +01:00
Vincent Lejeune
ac8a729bcd
SSAO: Actually simplifying code was a mistake.
2014-01-27 00:56:58 +01:00
Vincent Lejeune
684d265924
SSAO: Disable alpha test and simplify code
2014-01-27 00:53:52 +01:00
Vincent Lejeune
993d1e26dc
Revert "SSAO: Remove the strenght parameter"
...
This reverts commit b92f0b6ecc
2014-01-27 00:52:20 +01:00
Vincent Lejeune
63882318a7
SSAO: Fix typo and use 16 samples
2014-01-27 00:47:35 +01:00
Vincent Lejeune
b92f0b6ecc
SSAO: Remove the strenght parameter
2014-01-27 00:41:41 +01:00
Vincent Lejeune
31991ecd7e
SSAO: Improve randomness using NEAREST filtering
2014-01-27 00:09:36 +01:00
Marianne Gagnon
30011be115
My tweaks to SSAO, to improve randomness and final blending
2014-01-25 22:03:47 -05:00
Vincent Lejeune
ceb1fdc88a
SSAO: Use a single texture fetch for random.
2014-01-20 20:43:03 +01:00
Vincent Lejeune
32836cf93d
Use 8 samples instead of 16
2014-01-19 20:57:58 +01:00
