Differential D12370 Diff 41359 intern/cycles/integrator/pass_accessor_cpu.cpp

Changeset View

Standalone View

intern/cycles/integrator/pass_accessor_cpu.cpp

	Show First 20 Lines • Show All 91 Lines • ▼ Show 20 Lines
	template<typename Processor>			template<typename Processor>
	inline void PassAccessorCPU::run_get_pass_kernel_processor_float(			inline void PassAccessorCPU::run_get_pass_kernel_processor_float(
	const KernelFilmConvert *kfilm_convert,			const KernelFilmConvert *kfilm_convert,
	const RenderBuffers *render_buffers,			const RenderBuffers *render_buffers,
	const BufferParams &buffer_params,			const BufferParams &buffer_params,
	const Destination &destination,			const Destination &destination,
	const Processor &processor) const			const Processor &processor) const
	{			{
				DCHECK_EQ(destination.stride, 0) << "Custom stride for float destination is not implemented.";

	const float *buffer_data = render_buffers->buffer.data();			const float *buffer_data = render_buffers->buffer.data();
	const int pixel_stride = destination.pixel_stride ? destination.pixel_stride :			const int pixel_stride = destination.pixel_stride ? destination.pixel_stride :
	destination.num_components;			destination.num_components;

	tbb::parallel_for(0, buffer_params.height, [&](int y) {			tbb::parallel_for(0, buffer_params.height, [&](int64_t y) {
	int64_t pixel_index = int64_t(y) * buffer_params.width;			int64_t pixel_index = y * buffer_params.width;
	for (int x = 0; x < buffer_params.width; ++x, ++pixel_index) {			for (int64_t x = 0; x < buffer_params.width; ++x, ++pixel_index) {
	const int64_t input_pixel_offset = pixel_index * buffer_params.pass_stride;			const int64_t input_pixel_offset = pixel_index * buffer_params.pass_stride;
	const float *buffer = buffer_data + input_pixel_offset;			const float *buffer = buffer_data + input_pixel_offset;
	float pixel = destination.pixels + (pixel_index + destination.offset) pixel_stride;			float pixel = destination.pixels + (pixel_index + destination.offset) pixel_stride;

	processor(kfilm_convert, buffer, pixel);			processor(kfilm_convert, buffer, pixel);
	}			}
	});			});
	}			}

	template<typename Processor>			template<typename Processor>
	inline void PassAccessorCPU::run_get_pass_kernel_processor_half_rgba(			inline void PassAccessorCPU::run_get_pass_kernel_processor_half_rgba(
	const KernelFilmConvert *kfilm_convert,			const KernelFilmConvert *kfilm_convert,
	const RenderBuffers *render_buffers,			const RenderBuffers *render_buffers,
	const BufferParams &buffer_params,			const BufferParams &buffer_params,
	const Destination &destination,			const Destination &destination,
	const Processor &processor) const			const Processor &processor) const
	{			{
	const float *buffer_data = render_buffers->buffer.data();			const float *buffer_data = render_buffers->buffer.data();

	tbb::parallel_for(0, buffer_params.height, [&](int y) {			half4 *dst_start = destination.pixels_half_rgba + destination.offset;
				const int destination_stride = destination.stride != 0 ? destination.stride :
				buffer_params.width;

				tbb::parallel_for(0, buffer_params.height, [&](int64_t y) {
	int64_t pixel_index = y * buffer_params.width;			int64_t pixel_index = y * buffer_params.width;
	for (int x = 0; x < buffer_params.width; ++x, ++pixel_index) {			half4 dst_row_start = dst_start + y destination_stride;
				for (int64_t x = 0; x < buffer_params.width; ++x, ++pixel_index) {
	const int64_t input_pixel_offset = pixel_index * buffer_params.pass_stride;			const int64_t input_pixel_offset = pixel_index * buffer_params.pass_stride;
	const float *buffer = buffer_data + input_pixel_offset;			const float *buffer = buffer_data + input_pixel_offset;

	float pixel[4];			float pixel[4];
	processor(kfilm_convert, buffer, pixel);			processor(kfilm_convert, buffer, pixel);

	film_apply_pass_pixel_overlays_rgba(kfilm_convert, buffer, pixel);			film_apply_pass_pixel_overlays_rgba(kfilm_convert, buffer, pixel);

	half4 *pixel_half_rgba = destination.pixels_half_rgba + pixel_index + destination.offset;			half4 *pixel_half_rgba = dst_row_start + x;
	float4_store_half(&pixel_half_rgba->x, make_float4(pixel[0], pixel[1], pixel[2], pixel[3]));			float4_store_half(&pixel_half_rgba->x, make_float4(pixel[0], pixel[1], pixel[2], pixel[3]));
	}			}
	});			});
	}			}

	/* --------------------------------------------------------------------			/* --------------------------------------------------------------------
	* Pass accessors.			* Pass accessors.
	*/			*/
	Show All 31 Lines