Changeset View
Changeset View
Standalone View
Standalone View
intern/cycles/kernel/svm/svm_ramp.h
| Show All 15 Lines | |||||
| #ifndef __SVM_RAMP_H__ | #ifndef __SVM_RAMP_H__ | ||||
| #define __SVM_RAMP_H__ | #define __SVM_RAMP_H__ | ||||
| CCL_NAMESPACE_BEGIN | CCL_NAMESPACE_BEGIN | ||||
| /* NOTE: svm_ramp.h, svm_ramp_util.h and node_ramp_util.h must stay consistent */ | /* NOTE: svm_ramp.h, svm_ramp_util.h and node_ramp_util.h must stay consistent */ | ||||
| ccl_device_inline float fetch_float(ccl_global const KernelGlobals *kg, int offset) | ccl_device_inline float fetch_float(KernelGlobals kg, int offset) | ||||
| { | { | ||||
| uint4 node = kernel_tex_fetch(__svm_nodes, offset); | uint4 node = kernel_tex_fetch(__svm_nodes, offset); | ||||
| return __uint_as_float(node.x); | return __uint_as_float(node.x); | ||||
| } | } | ||||
| ccl_device_inline float float_ramp_lookup(ccl_global const KernelGlobals *kg, | ccl_device_inline float float_ramp_lookup( | ||||
| int offset, | KernelGlobals kg, int offset, float f, bool interpolate, bool extrapolate, int table_size) | ||||
| float f, | |||||
| bool interpolate, | |||||
| bool extrapolate, | |||||
| int table_size) | |||||
| { | { | ||||
| if ((f < 0.0f || f > 1.0f) && extrapolate) { | if ((f < 0.0f || f > 1.0f) && extrapolate) { | ||||
| float t0, dy; | float t0, dy; | ||||
| if (f < 0.0f) { | if (f < 0.0f) { | ||||
| t0 = fetch_float(kg, offset); | t0 = fetch_float(kg, offset); | ||||
| dy = t0 - fetch_float(kg, offset + 1); | dy = t0 - fetch_float(kg, offset + 1); | ||||
| f = -f; | f = -f; | ||||
| } | } | ||||
| Show All 14 Lines | ccl_device_inline float float_ramp_lookup( | ||||
| float a = fetch_float(kg, offset + i); | float a = fetch_float(kg, offset + i); | ||||
| if (interpolate && t > 0.0f) | if (interpolate && t > 0.0f) | ||||
| a = (1.0f - t) * a + t * fetch_float(kg, offset + i + 1); | a = (1.0f - t) * a + t * fetch_float(kg, offset + i + 1); | ||||
| return a; | return a; | ||||
| } | } | ||||
| ccl_device_inline float4 rgb_ramp_lookup(ccl_global const KernelGlobals *kg, | ccl_device_inline float4 rgb_ramp_lookup( | ||||
| int offset, | KernelGlobals kg, int offset, float f, bool interpolate, bool extrapolate, int table_size) | ||||
| float f, | |||||
| bool interpolate, | |||||
| bool extrapolate, | |||||
| int table_size) | |||||
| { | { | ||||
| if ((f < 0.0f || f > 1.0f) && extrapolate) { | if ((f < 0.0f || f > 1.0f) && extrapolate) { | ||||
| float4 t0, dy; | float4 t0, dy; | ||||
| if (f < 0.0f) { | if (f < 0.0f) { | ||||
| t0 = fetch_node_float(kg, offset); | t0 = fetch_node_float(kg, offset); | ||||
| dy = t0 - fetch_node_float(kg, offset + 1); | dy = t0 - fetch_node_float(kg, offset + 1); | ||||
| f = -f; | f = -f; | ||||
| } | } | ||||
| Show All 14 Lines | ccl_device_inline float4 rgb_ramp_lookup( | ||||
| float4 a = fetch_node_float(kg, offset + i); | float4 a = fetch_node_float(kg, offset + i); | ||||
| if (interpolate && t > 0.0f) | if (interpolate && t > 0.0f) | ||||
| a = (1.0f - t) * a + t * fetch_node_float(kg, offset + i + 1); | a = (1.0f - t) * a + t * fetch_node_float(kg, offset + i + 1); | ||||
| return a; | return a; | ||||
| } | } | ||||
| ccl_device_noinline int svm_node_rgb_ramp(ccl_global const KernelGlobals *kg, | ccl_device_noinline int svm_node_rgb_ramp( | ||||
| ccl_private ShaderData *sd, | KernelGlobals kg, ccl_private ShaderData *sd, ccl_private float *stack, uint4 node, int offset) | ||||
| ccl_private float *stack, | |||||
| uint4 node, | |||||
| int offset) | |||||
| { | { | ||||
| uint fac_offset, color_offset, alpha_offset; | uint fac_offset, color_offset, alpha_offset; | ||||
| uint interpolate = node.z; | uint interpolate = node.z; | ||||
| svm_unpack_node_uchar3(node.y, &fac_offset, &color_offset, &alpha_offset); | svm_unpack_node_uchar3(node.y, &fac_offset, &color_offset, &alpha_offset); | ||||
| uint table_size = read_node(kg, &offset).x; | uint table_size = read_node(kg, &offset).x; | ||||
| float fac = stack_load_float(stack, fac_offset); | float fac = stack_load_float(stack, fac_offset); | ||||
| float4 color = rgb_ramp_lookup(kg, offset, fac, interpolate, false, table_size); | float4 color = rgb_ramp_lookup(kg, offset, fac, interpolate, false, table_size); | ||||
| if (stack_valid(color_offset)) | if (stack_valid(color_offset)) | ||||
| stack_store_float3(stack, color_offset, float4_to_float3(color)); | stack_store_float3(stack, color_offset, float4_to_float3(color)); | ||||
| if (stack_valid(alpha_offset)) | if (stack_valid(alpha_offset)) | ||||
| stack_store_float(stack, alpha_offset, color.w); | stack_store_float(stack, alpha_offset, color.w); | ||||
| offset += table_size; | offset += table_size; | ||||
| return offset; | return offset; | ||||
| } | } | ||||
| ccl_device_noinline int svm_node_curves(ccl_global const KernelGlobals *kg, | ccl_device_noinline int svm_node_curves( | ||||
| ccl_private ShaderData *sd, | KernelGlobals kg, ccl_private ShaderData *sd, ccl_private float *stack, uint4 node, int offset) | ||||
| ccl_private float *stack, | |||||
| uint4 node, | |||||
| int offset) | |||||
| { | { | ||||
| uint fac_offset, color_offset, out_offset; | uint fac_offset, color_offset, out_offset; | ||||
| svm_unpack_node_uchar3(node.y, &fac_offset, &color_offset, &out_offset); | svm_unpack_node_uchar3(node.y, &fac_offset, &color_offset, &out_offset); | ||||
| uint table_size = read_node(kg, &offset).x; | uint table_size = read_node(kg, &offset).x; | ||||
| float fac = stack_load_float(stack, fac_offset); | float fac = stack_load_float(stack, fac_offset); | ||||
| float3 color = stack_load_float3(stack, color_offset); | float3 color = stack_load_float3(stack, color_offset); | ||||
| const float min_x = __int_as_float(node.z), max_x = __int_as_float(node.w); | const float min_x = __int_as_float(node.z), max_x = __int_as_float(node.w); | ||||
| const float range_x = max_x - min_x; | const float range_x = max_x - min_x; | ||||
| const float3 relpos = (color - make_float3(min_x, min_x, min_x)) / range_x; | const float3 relpos = (color - make_float3(min_x, min_x, min_x)) / range_x; | ||||
| float r = rgb_ramp_lookup(kg, offset, relpos.x, true, true, table_size).x; | float r = rgb_ramp_lookup(kg, offset, relpos.x, true, true, table_size).x; | ||||
| float g = rgb_ramp_lookup(kg, offset, relpos.y, true, true, table_size).y; | float g = rgb_ramp_lookup(kg, offset, relpos.y, true, true, table_size).y; | ||||
| float b = rgb_ramp_lookup(kg, offset, relpos.z, true, true, table_size).z; | float b = rgb_ramp_lookup(kg, offset, relpos.z, true, true, table_size).z; | ||||
| color = (1.0f - fac) * color + fac * make_float3(r, g, b); | color = (1.0f - fac) * color + fac * make_float3(r, g, b); | ||||
| stack_store_float3(stack, out_offset, color); | stack_store_float3(stack, out_offset, color); | ||||
| offset += table_size; | offset += table_size; | ||||
| return offset; | return offset; | ||||
| } | } | ||||
| ccl_device_noinline int svm_node_curve(ccl_global const KernelGlobals *kg, | ccl_device_noinline int svm_node_curve( | ||||
| ccl_private ShaderData *sd, | KernelGlobals kg, ccl_private ShaderData *sd, ccl_private float *stack, uint4 node, int offset) | ||||
| ccl_private float *stack, | |||||
| uint4 node, | |||||
| int offset) | |||||
| { | { | ||||
| uint fac_offset, value_in_offset, out_offset; | uint fac_offset, value_in_offset, out_offset; | ||||
| svm_unpack_node_uchar3(node.y, &fac_offset, &value_in_offset, &out_offset); | svm_unpack_node_uchar3(node.y, &fac_offset, &value_in_offset, &out_offset); | ||||
| uint table_size = read_node(kg, &offset).x; | uint table_size = read_node(kg, &offset).x; | ||||
| float fac = stack_load_float(stack, fac_offset); | float fac = stack_load_float(stack, fac_offset); | ||||
| float in = stack_load_float(stack, value_in_offset); | float in = stack_load_float(stack, value_in_offset); | ||||
| Show All 17 Lines | |||||