Azaezel
a9cb78b67f
straight convert of RGBA16F saving support for probes. should note gl side seems to be having an issue with loading probes thats resulting in color channel swapping.
2018-12-09 18:09:45 -06:00
Azaezel
9d560d9b12
skylight conversion
2018-12-09 16:57:14 -06:00
Azaezel
78f81e2b83
pointlight: forgot a secondary ssPos.xy/ssPos.w entry. GL rightly doesn't you to modify an input variable with impunity and reuse it. if we waant to go back to the precalculated .xyz/w route, pick a different varname (one that makes real sense, not just IN_ tacked on), or better yet precalculate the var itsself rather than going perpixel.
...
also converted spotlight.
2018-12-09 15:59:48 -06:00
Azaezel
89c534ce21
vectorlight conversion cleanups, pointlight port
2018-12-09 15:26:14 -06:00
Azaezel
c8b76d82c3
gl conversion WIP. general notes: mSamplerNames[#]/samplerNames[#] entry explicitly corresponds to the order of definition GL side.
...
shifted the colorbuffer slot over to S1 in keeping with the gbuffer layout for consistency
completed converts: brdf, lighting, torque.
nonvisually verified convert: vectorlight
noncompiling due to tripping on deferredUncondition: reflectionprobe
2018-12-08 01:41:06 -06:00
Azaezel
1766e291ba
Merge branch 'PBR_PR' of https://github.com/Azaezel/Torque3D into PBR_PR
2018-12-05 10:55:37 -06:00
Azaezel
74ca8d106c
retweaked filter for wether ot not to supress the zonelight culling methods when not in the diffuse pass
2018-12-05 10:55:29 -06:00
Tim Barnes
6937989d61
RGBA16F saving support for probes
2018-12-05 20:01:58 +10:00
Azaezel
10b7630312
put the z=w trick back for the cloud layer options. those parallaxing a bit is actually desireable (and fixes it rendering over everything else)
2018-12-04 14:25:49 -06:00
Azaezel
bd742bbf75
registergloballights now taks a (by default on) cullSceneLights. off for reflection capture as that is presently acting up
2018-12-04 12:41:15 -06:00
Azaezel
0d915b31ae
bump effective LOD (Level Of Detail) up during the capture phase to work arround things like trying to capture within a building with agressive optimization killing off the internals due to the smaller effective 'screen' being rendered.
2018-12-03 11:15:17 -06:00
Azaezel
a2bd0e0fcd
simplification
2018-11-29 18:19:27 -06:00
Azaezel
fcef7061c8
in keeping with a more integrated setup, used multiplicative ao for darkening existing composite map ao contributions.
2018-11-29 17:47:07 -06:00
Azaezel
fa41ea3e16
puts a form of ssao back. as this is intended to replicate shadows, bolted that to the shadow portion of the calc. as it was previously addative, did so in an addative manner.
2018-11-29 17:30:51 -06:00
Azaezel
6248b427bf
'better' (read here less completely craptacular) blending for boxes
2018-11-29 12:21:28 -06:00
Azaezel
47c36ff59d
THIS IS AN ABOMINATION UNTO THE CODEBASE AND SHOULD BE REMOVED THE SECOND WE CAN SORT OUT WHAT THE HECK IS GOING ON WITH THAT QSORT
...
stops flickering by itterating through the probe list twice. once looking for the skylight, then doing the rest
2018-11-28 18:12:12 -06:00
Azaezel
0974902cc8
put lerp back after verifying we were indeed getting values passed along.
2018-11-28 14:25:48 -06:00
Azaezel
9bf2378e70
simpler falloff calc for probes
2018-11-28 13:12:59 -06:00
Azaezel
4ef1a25610
put sphereical influence back via the new spherical attenuation methods, shifted the stateblock for probes specifically to max of either source or desitination alpha (though not entirely convinced that is in fact the case)
2018-11-28 12:00:06 -06:00
Tim Barnes
521bd2375d
lighting shader fixes
2018-11-28 20:01:35 +10:00
Tim Barnes
ab10cc0c87
timmy merge work
2018-11-28 17:51:52 +10:00
Azaezel
f1e584ca69
stateblocks: skylight uses one+(dest)zero. probes use (src)alpha+(dst(1.0-alpha) for a lerp blend. skylight writes out 0 alpha to ensure probes always win if even only a little.
2018-11-27 08:50:44 -06:00
Azaezel
65ab39a2d2
cast to pointer to pointer, not just pointer. (confused yet?)
2018-11-27 08:28:33 -06:00
Azaezel
0f0fc5279b
missed mscore pasalong in .set
2018-11-27 08:03:04 -06:00
Azaezel
2c1a5f5647
qsort compilation, though still not quite correct
2018-11-25 12:35:35 -06:00
Azaezel
b2ca85b988
couple additinal places for scroe to be set
2018-11-25 12:35:02 -06:00
Azaezel
79d506d439
1)use standard setupPass inheritance chain. already checks for !mprocessedmaterial and a few other things
...
2)pass along mIsSkylight from probes to matinstances
3) stubbs in a seperate setupPass for skylightmatinstance
2018-11-23 02:05:36 -06:00
Azaezel
96101a0e56
correction for folks supporting older codebases outside of this engine: don't try and include the older directx library unless youre also deliberately using the older directsound end of things, and for dx11, reference the #include <d3d11_1.h> lib instead of the <d3d11.h> one to pair with the reference used by SDL2
2018-11-23 00:30:23 -06:00
Tim Barnes
9df1338d4b
probe shader fix
2018-11-21 20:00:59 +10:00
Tim Barnes
9e65e940d0
lighting single buffer
2018-11-21 15:53:02 +10:00
Areloch
c4a4fe5304
Implemented registration of probes to avoid rendering all probes when unneeded.
2018-11-19 01:18:09 -06:00
Azaezel
ecd47830ca
shift sky render to last of non-translucents
2018-11-18 18:54:02 -06:00
Azaezel
518bcaaabf
from @rextimmy: shifts sybox rendering to last (kils a bit of overdraw)
2018-11-18 06:36:16 -06:00
Tim Barnes
c64aee9dcc
point light WIP & moved lighting position/direction to WS
2018-11-14 20:58:47 +10:00
Azaezel
da2c536daa
emissive fix: return the surface.basecolor in place of zeros emissive is at minimum non-shadowed. TODO: consult folks on if they want a multiplier, mask, or mix for further customization there.
2018-11-12 22:55:52 -06:00
Azaezel
338e66b7e6
add back in brightness and contrast controls, as well as the capacity to shut tonemapping off
2018-11-12 22:54:00 -06:00
Azaezel
29e78d2034
mUseHDRCaptures needs more work. shut that back off till we've time to circle back
2018-11-10 00:34:23 -06:00
Azaezel
9ec243cb87
fix to display that second box
2018-11-10 00:33:42 -06:00
Azaezel
b19fe23473
from timmy: fix for seperated out RGB and A blends
2018-11-10 00:32:24 -06:00
Azaezel
79e2d5d459
REVIEW LATER: invert the world transform for probes to shove them into cam space as far as rotation goes. keep position.
2018-11-09 20:16:15 -06:00
Azaezel
daef7aa128
debug spew cleanup
2018-11-08 01:27:42 -06:00
Azaezel
452c35f11c
further wipwork on the pointlight conversion. filled CreateSurfaceToLight out in a theoretically correct manner (spoiler, not actually correct)
2018-11-06 20:07:09 -06:00
Azaezel
b8fb90ab94
added the additional colorbuffer entry to the probe projection shader so we get output. TODO: leverage the new CreateSurface methodologies to match with the skylight
2018-11-06 20:03:22 -06:00
Azaezel
73a7d205e2
add in secondary bounds box display for probes as that is not actually linked to the radius/box-based-on-radius
2018-11-06 20:00:34 -06:00
Azaezel
913fbef66d
shift surface calcs to post-attenuation. no point if we're just gonna throw the vals out in the end
2018-11-02 17:41:30 -05:00
Azaezel
508e9b4f27
account for attenuation
2018-11-02 16:28:50 -05:00
Azaezel
18c204b341
add animated spotlight and pointlight to test scene
2018-11-02 16:09:02 -05:00
Azaezel
7c05e3d33d
pointlight conversion rev1- filled in creatrsurface, used pre-exisitng calcs for nDotL to aproximate the light direction needed by CreateSurfaceToLight(surface, dot( lightVec, normDepth.xyz )); NOTE: compiles, but artifacts badly.
2018-11-02 16:07:52 -05:00
Azaezel
1169ad05f1
spotlight conversion
2018-11-02 16:04:53 -05:00
Azaezel
b992446fff
we're no longer making use of a singular lightbuffer to double up as a baked lighting/ibl channel passed to lights themselves, so save a screens worth of transmission
2018-11-02 16:04:04 -05:00
