Update to r3

Add 16 bit depth support
WolframRhodium · Jun 3, 2017 · f64d32a · f64d32a
1 parent e2e7293
commit f64d32a
Show file tree

Hide file tree

Showing 7 changed files with 28 additions and 319 deletions.
diff --git a/Source/dpid.cpp b/Source/dpid.cpp
@@ -2,8 +2,9 @@
 #include <cmath>
 #include <string>
 
-#include <vapoursynth\VapourSynth.h>
-#include <vapoursynth\VSHelper.h>
+#include "vapoursynth\VapourSynth.h"
+#include "vapoursynth\VSHelper.h"
+#include "dpid.h"
 
 struct DpidData {
 	VSNodeRef * node;
@@ -13,13 +14,8 @@ struct DpidData {
 	uint16_t * src16Interleaved, *dst16Interleaved;
 };
 
-struct Params {
-	uint32_t oWidth, oHeight, iWidth, iHeight, pixel_max;
-	float pWidth, pHeight, lambda;
-};
-
-void run8(Params& i, const void* hInput, void* hOutput);
-void run16(Params& i, const void* hInput, void* hOutput);
+extern void run8(Params& i, const void* hInput, void* hOutput);
+extern void run16(Params& i, const void* hInput, void* hOutput);
 
 static void process8(const VSFrameRef * src, VSFrameRef * dst, DpidData * d, const VSAPI * vsapi)
 {
@@ -39,8 +35,8 @@ static void process8(const VSFrameRef * src, VSFrameRef * dst, DpidData * d, con
 	args.iHeight = height;
 	args.oWidth = d->vi.width;
 	args.oHeight = d->vi.height;
-	//args.pWidth = args.iWidth / (float)args.oWidth;
-	//args.pHeight = args.iHeight / (float)args.oHeight; // This args broke when the struct "Params" contained "pixel_max".
+	args.pWidth = args.iWidth / (float)args.oWidth;
+	args.pHeight = args.iHeight / (float)args.oHeight;
 	args.lambda = d->lambda;
 	args.pixel_max = (1 << d->vi.format->bitsPerSample) - 1;
 
@@ -95,8 +91,8 @@ static void process16(const VSFrameRef * src, VSFrameRef * dst, DpidData * d, co
 	args.iHeight = height;
 	args.oWidth = d->vi.width;
 	args.oHeight = d->vi.height;
-	//args.pWidth = args.iWidth / (float)args.oWidth;
-	//args.pHeight = args.iHeight / (float)args.oHeight;  // This args broke when the struct "Params" contained "pixel_max".
+	args.pWidth = args.iWidth / (float)args.oWidth;
+	args.pHeight = args.iHeight / (float)args.oHeight;
 	args.lambda = d->lambda;
 	args.pixel_max = (1 << d->vi.format->bitsPerSample) - 1;
 

diff --git a/Source/dpid.h b/Source/dpid.h
@@ -0,0 +1,11 @@
+#ifndef DPID_H_
+#define DPID_H_
+
+#include <cstdint>
+
+struct Params {
+	uint32_t oWidth, oHeight, iWidth, iHeight, pixel_max;
+	float pWidth, pHeight, lambda;
+};
+
+#endif
diff --git a/Source/kernel.cu b/Source/kernel.cu
diff --git a/Source/kernel16.cu b/Source/kernel16.cu
@@ -297,9 +297,6 @@ void run16(Params& i, const void* hInput, void* hOutput)
 	check(cudaMalloc(&dOutput, sOutput));
 	check(cudaMalloc(&dGuidance, sGuidance));
 
-	i.pHeight = i.iHeight / (float)i.oHeight;
-	i.pWidth = i.iWidth / (float)i.oWidth;
-
 	// copy data
 	check(cudaMemcpy(dInput, hInput, sInput, cudaMemcpyHostToDevice));
 

diff --git a/Source/kernel8.cu b/Source/kernel8.cu
@@ -297,9 +297,6 @@ void run8(Params& i, const void* hInput, void* hOutput)
 	check(cudaMalloc(&dOutput, sOutput));
 	check(cudaMalloc(&dGuidance, sGuidance));
 
-	i.pHeight = i.iHeight / (float)i.oHeight;
-	i.pWidth = i.iWidth / (float)i.oWidth;
-
 	// copy data
 	check(cudaMemcpy(dInput, hInput, sInput, cudaMemcpyHostToDevice));
 

diff --git a/Source/vapoursynth-dpid/vapoursynth-dpid.vcxproj b/Source/vapoursynth-dpid/vapoursynth-dpid.vcxproj
@@ -100,6 +100,9 @@
     <CudaCompile Include="..\kernel16.cu" />
     <CudaCompile Include="..\kernel8.cu" />
   </ItemGroup>
+  <ItemGroup>
+    <ClInclude Include="..\dpid.h" />
+  </ItemGroup>
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
   <ImportGroup Label="ExtensionTargets">
     <Import Project="$(VCTargetsPath)\BuildCustomizations\CUDA 8.0.targets" />