-
Notifications
You must be signed in to change notification settings - Fork 57
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
* feat: variable rate shading with nvapi extensions * feat: FFR test * feat: adaptive rate shading * feat: motion VRS * feat: finalising VRS
- Loading branch information
Showing
12 changed files
with
634 additions
and
33 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -1,3 +1,6 @@ | ||
[submodule "extern/CommonLibSSE-NG"] | ||
path = extern/CommonLibSSE-NG | ||
url = https://github.com/alandtse/CommonLibVR.git | ||
[submodule "extern/NVAPI"] | ||
path = extern/NVAPI | ||
url = https://github.com/NVIDIA/nvapi.git |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
102 changes: 102 additions & 0 deletions
102
package/Shaders/VariableRateShading/ComputeNASData.hlsl
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,102 @@ | ||
#define GROUP_SIZE (8 * 8) | ||
|
||
Texture2D<float3> InputTexture : register(t0); | ||
Texture2D<float2> InputTextureMotionVectors : register(t1); | ||
RWTexture2D<float4> OutputTexture : register(u0); | ||
|
||
groupshared float4 sampleCache[GROUP_SIZE]; | ||
groupshared float2 sampleCacheMotionVectors[GROUP_SIZE]; | ||
groupshared float errXCache[GROUP_SIZE]; | ||
groupshared float errYCache[GROUP_SIZE]; | ||
|
||
float RgbToLuminance(float3 color) | ||
{ | ||
return dot(color, float3(0.299, 0.587, 0.114)); | ||
} | ||
|
||
[numthreads(8, 8, 1)] | ||
void main(uint3 GroupID : SV_GroupID, uint3 GroupThreadID : SV_GroupThreadID) | ||
{ | ||
const uint threadIndex = GroupThreadID.y * 8 + GroupThreadID.x; | ||
const uint2 sampleIndex = (GroupID.xy * 8 + GroupThreadID.xy) * 2.0; | ||
|
||
// Fetch color (final post-AA) data | ||
// l0.x l0.y | ||
// l0.z l0.w l2.x | ||
// l1.x l1.y | ||
// l1.z l1.w l2.y | ||
// l2.z | ||
float4 l0; | ||
l0.x = RgbToLuminance(InputTexture[sampleIndex + uint2(0, 0)]); | ||
l0.y = RgbToLuminance(InputTexture[sampleIndex + uint2(1, 0)]); | ||
l0.z = RgbToLuminance(InputTexture[sampleIndex + uint2(0, 1)]); | ||
l0.w = RgbToLuminance(InputTexture[sampleIndex + uint2(1, 1)]); | ||
|
||
float4 l1; | ||
l1.x = RgbToLuminance(InputTexture[sampleIndex + uint2(0, 2)]); | ||
l1.y = RgbToLuminance(InputTexture[sampleIndex + uint2(1, 2)]); | ||
l1.z = RgbToLuminance(InputTexture[sampleIndex + uint2(0, 3)]); | ||
l1.w = RgbToLuminance(InputTexture[sampleIndex + uint2(1, 3)]); | ||
|
||
float3 l2; | ||
l2.x = RgbToLuminance(InputTexture[sampleIndex + uint2(2, 1)]); | ||
l2.y = RgbToLuminance(InputTexture[sampleIndex + uint2(2, 3)]); | ||
l2.z = RgbToLuminance(InputTexture[sampleIndex + uint2(1, 4)]); | ||
|
||
sampleCache[threadIndex] = l0 + l1; | ||
|
||
float2 m = 0.0f; | ||
m += InputTextureMotionVectors[sampleIndex + uint2(0, 0)]; | ||
m += InputTextureMotionVectors[sampleIndex + uint2(1, 0)]; | ||
m += InputTextureMotionVectors[sampleIndex + uint2(0, 1)]; | ||
m += InputTextureMotionVectors[sampleIndex + uint2(1, 1)]; | ||
|
||
m += InputTextureMotionVectors[sampleIndex + uint2(0, 2)]; | ||
m += InputTextureMotionVectors[sampleIndex + uint2(1, 2)]; | ||
m += InputTextureMotionVectors[sampleIndex + uint2(0, 3)]; | ||
m += InputTextureMotionVectors[sampleIndex + uint2(1, 3)]; | ||
|
||
sampleCacheMotionVectors[threadIndex] = m; | ||
|
||
// Derivatives X | ||
float4 a = float4(l0.y, l2.x, l1.y, l2.y); | ||
float4 b = float4(l0.x, l0.w, l1.x, l1.w); | ||
float4 dx = abs(a - b); | ||
|
||
// Derivatives Y | ||
a = float4(l0.z, l1.y, l1.z, l2.z); | ||
b = float4(l0.x, l0.w, l1.x, l1.w); | ||
float4 dy = abs(a - b); | ||
|
||
// Compute maximum partial derivative of all 16x16 pixels (256 total) | ||
// this approach is more "sensitive" to individual outliers in a tile, since it takes the max instead of the average | ||
float maxDx = max(max(dx.x, dx.y), max(dx.z, dx.w)); | ||
float maxDy = max(max(dy.x, dy.y), max(dy.z, dy.w)); | ||
|
||
errXCache[threadIndex] = maxDx; | ||
errYCache[threadIndex] = maxDy; | ||
|
||
GroupMemoryBarrierWithGroupSync(); | ||
|
||
// Parallel reduction | ||
[unroll] for (uint s = (64 >> 1); s > 0; s >>= 1) | ||
{ | ||
if(threadIndex < s){ | ||
sampleCache[threadIndex] += sampleCache[threadIndex + s]; | ||
sampleCacheMotionVectors[threadIndex] += sampleCacheMotionVectors[threadIndex + s]; | ||
errXCache[threadIndex] = max(errXCache[threadIndex], errXCache[threadIndex + s]); | ||
errYCache[threadIndex] = max(errYCache[threadIndex], errYCache[threadIndex + s]); | ||
} | ||
|
||
GroupMemoryBarrierWithGroupSync(); | ||
} | ||
|
||
// Average | ||
if(threadIndex == 0){ | ||
float avgLuma = dot(sampleCache[0], 1.0 / 8.0) / GROUP_SIZE + 0.1; | ||
float2 avgMotionVectors = dot(sampleCacheMotionVectors[0], 1.0 / 8.0) / GROUP_SIZE; | ||
float errX = errXCache[0]; | ||
float errY = errYCache[0]; | ||
OutputTexture[GroupID.xy] = float4(float2(errX, errY) / abs(avgLuma), avgMotionVectors); | ||
} | ||
} |
80 changes: 80 additions & 0 deletions
80
package/Shaders/VariableRateShading/ComputeShadingRate.hlsl
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,80 @@ | ||
RWTexture2D<uint> vrsSurface : register(u0); | ||
Texture2D<float4> nasDataSurface : register(t0); | ||
|
||
[numthreads(32, 32, 1)] | ||
void main(uint3 DispatchThreadID : SV_DispatchThreadID, uint3 GroupThreadID : SV_GroupThreadID, uint3 GroupID : SV_GroupID) | ||
{ | ||
float4 nasData = nasDataSurface[DispatchThreadID.xy]; | ||
|
||
float2 mVec = abs(nasData.zw); | ||
|
||
// Error scalers (equations from the I3D 2019 paper) | ||
// bhv for half rate, bqv for quarter rate | ||
float2 bhv = pow(1.0 / (1 + pow(1.05 * mVec, 3.1)), 0.35); | ||
float2 bqv = 2.13 * pow(1.0 / (1 + pow(0.55 * mVec, 2.41)), 0.49); | ||
|
||
// Sample block error data from NAS data pass and apply the error scalars | ||
float2 diff = nasData.xy; | ||
float2 diff2 = diff * bhv; | ||
float2 diff4 = diff * bqv; | ||
|
||
uint screenWidth, screenHeight; | ||
nasDataSurface.GetDimensions(screenWidth, screenHeight); | ||
|
||
float2 uv = DispatchThreadID.xy * rcp(float2(screenWidth, screenHeight)); | ||
float threshold = lerp(0.1, 0.2, distance(float2(0.5, 0.5), uv)); | ||
|
||
/*` | ||
D3D12_SHADING_RATE_1X1 = 0, // 0b0000 | ||
D3D12_SHADING_RATE_1X2 = 0x1, // 0b0001 | ||
D3D12_SHADING_RATE_2X1 = 0x4, // 0b0100 | ||
D3D12_SHADING_RATE_2X2 = 0x5, // 0b0101 | ||
D3D12_SHADING_RATE_2X4 = 0x6, // 0b0110 | ||
D3D12_SHADING_RATE_4X2 = 0x9, // 0b1001 | ||
D3D12_SHADING_RATE_4X4 = 0xa // 0b1010 | ||
*/ | ||
|
||
// Compute block shading rate based on if the error computation goes over the threshold | ||
// shading rates in D3D are purposely designed to be able to combined, e.g. 2x1 | 1x2 = 2x2 | ||
uint ShadingRate = 0; | ||
ShadingRate |= ((diff2.x >= threshold) ? 0 : ((diff4.x > threshold) ? 0x4 : 0x8)); | ||
ShadingRate |= ((diff2.y >= threshold) ? 0 : ((diff4.y > threshold) ? 0x1 : 0x2)); | ||
|
||
// Disable 4x4 shading rate (low quality, limited perf gain) | ||
if (ShadingRate == 0xa) | ||
{ | ||
ShadingRate = (diff2.x > diff2.y) ? 0x6 : 0x9; // use 2x4 or 4x2 based on directional gradient | ||
} | ||
// Disable 4x1 or 1x4 shading rate (unsupported) | ||
else if (ShadingRate == 0x8) | ||
{ | ||
ShadingRate = 0x4; | ||
} | ||
else if (ShadingRate == 0x2) | ||
{ | ||
ShadingRate = 0x1; | ||
} | ||
|
||
// vsrd[i].shadingRateTable[0] = NV_PIXEL_X1_PER_RASTER_PIXEL; | ||
// vsrd[i].shadingRateTable[1] = NV_PIXEL_X1_PER_2X1_RASTER_PIXELS; | ||
// vsrd[i].shadingRateTable[2] = NV_PIXEL_X1_PER_1X2_RASTER_PIXELS; | ||
// vsrd[i].shadingRateTable[3] = NV_PIXEL_X1_PER_2X2_RASTER_PIXELS; | ||
// vsrd[i].shadingRateTable[4] = NV_PIXEL_X1_PER_4X2_RASTER_PIXELS; | ||
// vsrd[i].shadingRateTable[5] = NV_PIXEL_X1_PER_2X4_RASTER_PIXELS; | ||
// vsrd[i].shadingRateTable[6] = NV_PIXEL_X1_PER_4X4_RASTER_PIXELS; | ||
|
||
if (ShadingRate == 0x1) | ||
ShadingRate = 2; | ||
else if (ShadingRate == 0x4) | ||
ShadingRate = 1; | ||
else if (ShadingRate == 0x5) | ||
ShadingRate = 3; | ||
else if (ShadingRate == 0x6) | ||
ShadingRate = 5; | ||
else if (ShadingRate == 0x9) | ||
ShadingRate = 4; | ||
else if (ShadingRate == 0xa) | ||
ShadingRate = 6; | ||
|
||
vrsSurface[DispatchThreadID.xy] = ShadingRate; | ||
} |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,19 @@ | ||
Copyright (c) 2014-2021, NVIDIA CORPORATION. All rights reserved. | ||
|
||
Permission is hereby granted, free of charge, to any person obtaining a | ||
copy of this software and associated documentation files (the "Software"), | ||
to deal in the Software without restriction, including without limitation | ||
the rights to use, copy, modify, merge, publish, distribute, sublicense, | ||
and/or sell copies of the Software, and to permit persons to whom the | ||
Software is furnished to do so, subject to the following conditions: | ||
|
||
The above copyright notice and this permission notice shall be included in | ||
all copies or substantial portions of the Software. | ||
|
||
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR | ||
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, | ||
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL | ||
THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER | ||
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING | ||
FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER | ||
DEALINGS IN THE SOFTWARE. |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.