Cycles: Support half and half4 textures.

This is an initial commit for half texture support in Cycles.
It adds the basic infrastructure inside of the ImageManager and support for these textures on CPU.

Supported:
* Half Float OpenEXR images (can be used for e.g HDRs or Normalmaps) now use 1/2 the memory, when loaded via disk (OIIO).

ToDo:
Various things like support for inbuilt half textures, GPU... will come later, step by step.

Part of my GSoC 2016.

4 files changed, 62 insertions, 3 deletions

diff --git a/intern/cycles/kernel/kernel_compat_cpu.h b/intern/cycles/kernel/kernel_compat_cpu.h
index bb303b32705..c882b477c35 100644
--- a/intern/cycles/kernel/kernel_compat_cpu.h
+++ b/intern/cycles/kernel/kernel_compat_cpu.h
@@ -122,6 +122,17 @@ template<typename T> struct texture_image  {
 		return make_float4(r, r, r, 1.0f);
 	}
 
+	ccl_always_inline float4 read(half4 r)
+	{
+		return half4_to_float4(r);
+	}
+
+	ccl_always_inline float4 read(half r)
+	{
+		float f = half_to_float(r);
+		return make_float4(f, f, f, 1.0f);
+	}
+
 	ccl_always_inline int wrap_periodic(int x, int width)
 	{
 		x %= width;
@@ -486,8 +497,10 @@ typedef texture<uint4> texture_uint4;
 typedef texture<uchar4> texture_uchar4;
 typedef texture_image<float> texture_image_float;
 typedef texture_image<uchar> texture_image_uchar;
+typedef texture_image<half> texture_image_half;
 typedef texture_image<float4> texture_image_float4;
 typedef texture_image<uchar4> texture_image_uchar4;
+typedef texture_image<half4> texture_image_half4;
 
 /* Macros to handle different memory storage on different devices */
 
diff --git a/intern/cycles/kernel/kernel_globals.h b/intern/cycles/kernel/kernel_globals.h
index e06c68f2fc9..8e66a3a0340 100644
--- a/intern/cycles/kernel/kernel_globals.h
+++ b/intern/cycles/kernel/kernel_globals.h
@@ -37,8 +37,10 @@ struct VolumeStep;
 typedef struct KernelGlobals {
 	texture_image_uchar4 texture_byte4_images[TEX_NUM_BYTE4_CPU];
 	texture_image_float4 texture_float4_images[TEX_NUM_FLOAT4_CPU];
+	texture_image_half4 texture_half4_images[TEX_NUM_HALF4_CPU];
 	texture_image_float texture_float_images[TEX_NUM_FLOAT_CPU];
 	texture_image_uchar texture_byte_images[TEX_NUM_BYTE_CPU];
+	texture_image_half texture_half_images[TEX_NUM_HALF_CPU];
 
 #  define KERNEL_TEX(type, ttype, name) ttype name;
 #  define KERNEL_IMAGE_TEX(type, ttype, name)
diff --git a/intern/cycles/kernel/kernels/cpu/kernel.cpp b/intern/cycles/kernel/kernels/cpu/kernel.cpp
index d8a83f69685..f11c85d5f6a 100644
--- a/intern/cycles/kernel/kernels/cpu/kernel.cpp
+++ b/intern/cycles/kernel/kernels/cpu/kernel.cpp
@@ -154,6 +154,38 @@ void kernel_tex_copy(KernelGlobals *kg,
 			tex->extension = extension;
 		}
 	}
+	else if(strstr(name, "__tex_image_half4")) {
+		texture_image_half4 *tex = NULL;
+		int id = atoi(name + strlen("__tex_image_half4_"));
+		int array_index = id - TEX_START_HALF4_CPU;
+
+		if(array_index >= 0 && array_index < TEX_NUM_HALF4_CPU) {
+			tex = &kg->texture_half4_images[array_index];
+		}
+
+		if(tex) {
+			tex->data = (half4*)mem;
+			tex->dimensions_set(width, height, depth);
+			tex->interpolation = interpolation;
+			tex->extension = extension;
+		}
+	}
+	else if(strstr(name, "__tex_image_half")) {
+		texture_image_half *tex = NULL;
+		int id = atoi(name + strlen("__tex_image_half_"));
+		int array_index = id - TEX_START_HALF_CPU;
+
+		if(array_index >= 0 && array_index < TEX_NUM_HALF_CPU) {
+			tex = &kg->texture_half_images[array_index];
+		}
+
+		if(tex) {
+			tex->data = (half*)mem;
+			tex->dimensions_set(width, height, depth);
+			tex->interpolation = interpolation;
+			tex->extension = extension;
+		}
+	}
 	else
 		assert(0);
 }
diff --git a/intern/cycles/kernel/kernels/cpu/kernel_cpu_image.h b/intern/cycles/kernel/kernels/cpu/kernel_cpu_image.h
index b10861ab857..47383140170 100644
--- a/intern/cycles/kernel/kernels/cpu/kernel_cpu_image.h
+++ b/intern/cycles/kernel/kernels/cpu/kernel_cpu_image.h
@@ -23,7 +23,11 @@ CCL_NAMESPACE_BEGIN
 
 ccl_device float4 kernel_tex_image_interp_impl(KernelGlobals *kg, int tex, float x, float y)
 {
-	if(tex >= TEX_START_BYTE_CPU)
+	if(tex >= TEX_START_HALF_CPU)
+		return kg->texture_half_images[tex - TEX_START_HALF_CPU].interp(x, y);
+	else if(tex >= TEX_START_HALF4_CPU)
+		return kg->texture_half4_images[tex - TEX_START_HALF4_CPU].interp(x, y);
+	else if(tex >= TEX_START_BYTE_CPU)
 		return kg->texture_byte_images[tex - TEX_START_BYTE_CPU].interp(x, y);
 	else if(tex >= TEX_START_FLOAT_CPU)
 		return kg->texture_float_images[tex - TEX_START_FLOAT_CPU].interp(x, y);
@@ -35,7 +39,11 @@ ccl_device float4 kernel_tex_image_interp_impl(KernelGlobals *kg, int tex, float
 
 ccl_device float4 kernel_tex_image_interp_3d_impl(KernelGlobals *kg, int tex, float x, float y, float z)
 {
-	if(tex >= TEX_START_BYTE_CPU)
+	if(tex >= TEX_START_HALF_CPU)
+		return kg->texture_half_images[tex - TEX_START_HALF_CPU].interp_3d(x, y, z);
+	else if(tex >= TEX_START_HALF4_CPU)
+		return kg->texture_half4_images[tex - TEX_START_HALF4_CPU].interp_3d(x, y, z);
+	else if(tex >= TEX_START_BYTE_CPU)
 		return kg->texture_byte_images[tex - TEX_START_BYTE_CPU].interp_3d(x, y, z);
 	else if(tex >= TEX_START_FLOAT_CPU)
 		return kg->texture_float_images[tex - TEX_START_FLOAT_CPU].interp_3d(x, y, z);
@@ -48,7 +56,11 @@ ccl_device float4 kernel_tex_image_interp_3d_impl(KernelGlobals *kg, int tex, fl
 
 ccl_device float4 kernel_tex_image_interp_3d_ex_impl(KernelGlobals *kg, int tex, float x, float y, float z, int interpolation)
 {
-	if(tex >= TEX_START_BYTE_CPU)
+	if(tex >= TEX_START_HALF_CPU)
+		return kg->texture_half4_images[tex - TEX_START_HALF_CPU].interp_3d_ex(x, y, z, interpolation);
+	else if(tex >= TEX_START_HALF4_CPU)
+		return kg->texture_half_images[tex - TEX_START_HALF4_CPU].interp_3d_ex(x, y, z, interpolation);
+	else if(tex >= TEX_START_BYTE_CPU)
 		return kg->texture_byte_images[tex - TEX_START_BYTE_CPU].interp_3d_ex(x, y, z, interpolation);
 	else if(tex >= TEX_START_FLOAT_CPU)
 		return kg->texture_float_images[tex - TEX_START_FLOAT_CPU].interp_3d_ex(x, y, z, interpolation);