OpenSubdiv: Commit of OpenSubdiv integration into Blender

This commit contains all the remained parts needed for initial integration of
OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU
backends which works in the following way:

- When SubSurf modifier is the last in the modifiers stack then GPU pipeline
  of OpenSubdiv is used, making viewport performance as fast as possible.

  This also requires graphscard with GLSL 1.5 support. If this requirement is
  not met, then no GPU pipeline is used at all.

- If SubSurf is not a last modifier or if DerivesMesh is being evaluated for
  rendering then CPU limit evaluation API from OpenSubdiv is used. This only
  replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG
  structures exactly the same as they used to be for ages now.

This integration is fully covered with ifdef and not enabled by default
because there are several TODOs to be solved first:

- Face varying data interpolation is not really cleanly implemented for GPU
  in OpenSubdiv 3.0. It is also not implemented for limit evaluation API.

  This basically means we'll have really hard time supporting UVs.

- Limit evaluation only works with adaptivly subdivided meshes so far, which
  basically means all the points of CCG are pushed to the limit. This gives
  different result from old code.

- There are some serious optimizations possible on the topology refiner
  creation, which would speed up initial OpenSubdiv mesh creation.

- There are some hardcoded asumptions in the GPU and DerivedMesh areas which
  could be generalized.

  That's something where Antony and Campbell can help, making it so the code
  is structured in a way which is reusable by all planned viewport projects.

- There are also some workarounds in the dependency graph to make sure OpenGL
  buffers are only freed from the main thread.

Those who'll be wanting to make experiments with this code should grab dev
branch (NOT master) from

  https://github.com/Nazg-Gul/OpenSubdiv/tree/dev

There are some patches applied in there which we're working on on getting
into upstream.

1 files changed, 169 insertions, 9 deletions

diff --git a/source/blender/gpu/intern/gpu_codegen.c b/source/blender/gpu/intern/gpu_codegen.c
index 335342c7123..68b9e3845f7 100644
--- a/source/blender/gpu/intern/gpu_codegen.c
+++ b/source/blender/gpu/intern/gpu_codegen.c
@@ -56,7 +56,7 @@
 extern char datatoc_gpu_shader_material_glsl[];
 extern char datatoc_gpu_shader_vertex_glsl[];
 extern char datatoc_gpu_shader_vertex_world_glsl[];
-
+extern char datatoc_gpu_shader_geometry_glsl[];
 
 static char *glsl_material_library = NULL;
 
@@ -531,8 +531,19 @@ static int codegen_print_uniforms_functions(DynStr *ds, ListBase *nodes)
 				}
 			}
 			else if (input->source == GPU_SOURCE_ATTRIB && input->attribfirst) {
+#ifdef WITH_OPENSUBDIV
+				bool skip_opensubdiv = input->attribtype == CD_TANGENT;
+				if (skip_opensubdiv) {
+					BLI_dynstr_appendf(ds, "#ifndef USE_OPENSUBDIV\n");
+				}
+#endif
 				BLI_dynstr_appendf(ds, "varying %s var%d;\n",
 					GPU_DATATYPE_STR[input->type], input->attribid);
+#ifdef WITH_OPENSUBDIV
+				if (skip_opensubdiv) {
+					BLI_dynstr_appendf(ds, "#endif\n");
+				}
+#endif
 			}
 		}
 	}
@@ -633,6 +644,12 @@ static char *code_generate_fragment(ListBase *nodes, GPUOutput *output)
 	char *code;
 	int builtins;
 
+#ifdef WITH_OPENSUBDIV
+	GPUNode *node;
+	GPUInput *input;
+#endif
+
+
 #if 0
 	BLI_dynstr_append(ds, FUNCTION_PROTOTYPES);
 #endif
@@ -650,7 +667,35 @@ static char *code_generate_fragment(ListBase *nodes, GPUOutput *output)
 
 	if (builtins & GPU_VIEW_NORMAL)
 		BLI_dynstr_append(ds, "\tvec3 facingnormal = (gl_FrontFacing)? varnormal: -varnormal;\n");
-	
+
+	/* Calculate tangent space. */
+#ifdef WITH_OPENSUBDIV
+	{
+		bool has_tangent = false;
+		for (node = nodes->first; node; node = node->next) {
+			for (input = node->inputs.first; input; input = input->next) {
+				if (input->source == GPU_SOURCE_ATTRIB && input->attribfirst) {
+					if (input->attribtype == CD_TANGENT) {
+						BLI_dynstr_appendf(ds, "#ifdef USE_OPENSUBDIV\n");
+						BLI_dynstr_appendf(ds, "\t%s var%d;\n",
+						                   GPU_DATATYPE_STR[input->type],
+						                   input->attribid);
+						if (has_tangent == false) {
+							BLI_dynstr_appendf(ds, "\tvec3 Q1 = dFdx(inpt.v.position.xyz);\n");
+							BLI_dynstr_appendf(ds, "\tvec3 Q2 = dFdy(inpt.v.position.xyz);\n");
+							BLI_dynstr_appendf(ds, "\tvec2 st1 = dFdx(inpt.v.uv);\n");
+							BLI_dynstr_appendf(ds, "\tvec2 st2 = dFdy(inpt.v.uv);\n");
+							BLI_dynstr_appendf(ds, "\tvec3 T = normalize(Q1 * st2.t - Q2 * st1.t);\n");
+						}
+						BLI_dynstr_appendf(ds, "\tvar%d = vec4(T, 1.0);\n", input->attribid);
+						BLI_dynstr_appendf(ds, "#endif\n");
+					}
+				}
+			}
+		}
+	}
+#endif
+
 	codegen_declare_tmps(ds, nodes);
 	codegen_call_functions(ds, nodes, output);
 
@@ -678,10 +723,21 @@ static char *code_generate_vertex(ListBase *nodes, const GPUMatType type)
 	for (node = nodes->first; node; node = node->next) {
 		for (input = node->inputs.first; input; input = input->next) {
 			if (input->source == GPU_SOURCE_ATTRIB && input->attribfirst) {
+#ifdef WITH_OPENSUBDIV
+				bool skip_opensubdiv = ELEM(input->attribtype, CD_MTFACE, CD_TANGENT);
+				if (skip_opensubdiv) {
+					BLI_dynstr_appendf(ds, "#ifndef USE_OPENSUBDIV\n");
+				}
+#endif
 				BLI_dynstr_appendf(ds, "attribute %s att%d;\n",
 					GPU_DATATYPE_STR[input->type], input->attribid);
 				BLI_dynstr_appendf(ds, "varying %s var%d;\n",
 					GPU_DATATYPE_STR[input->type], input->attribid);
+#ifdef WITH_OPENSUBDIV
+				if (skip_opensubdiv) {
+					BLI_dynstr_appendf(ds, "#endif\n");
+				}
+#endif
 			}
 		}
 	}
@@ -706,11 +762,29 @@ static char *code_generate_vertex(ListBase *nodes, const GPUMatType type)
 		for (input = node->inputs.first; input; input = input->next)
 			if (input->source == GPU_SOURCE_ATTRIB && input->attribfirst) {
 				if (input->attribtype == CD_TANGENT) { /* silly exception */
+#ifdef WITH_OPENSUBDIV
+					BLI_dynstr_appendf(ds, "#ifndef USE_OPENSUBDIV\n");
+#endif
 					BLI_dynstr_appendf(ds, "\tvar%d.xyz = normalize(gl_NormalMatrix * att%d.xyz);\n", input->attribid, input->attribid);
 					BLI_dynstr_appendf(ds, "\tvar%d.w = att%d.w;\n", input->attribid, input->attribid);
+#ifdef WITH_OPENSUBDIV
+					BLI_dynstr_appendf(ds, "#endif\n");
+#endif
 				}
-				else
+				else {
+#ifdef WITH_OPENSUBDIV
+					bool is_mtface = input->attribtype == CD_MTFACE;
+					if (is_mtface) {
+						BLI_dynstr_appendf(ds, "#ifndef USE_OPENSUBDIV\n");
+					}
+#endif
 					BLI_dynstr_appendf(ds, "\tvar%d = att%d;\n", input->attribid, input->attribid);
+#ifdef WITH_OPENSUBDIV
+					if (is_mtface) {
+						BLI_dynstr_appendf(ds, "#endif\n");
+					}
+#endif
+				}
 			}
 			/* unfortunately special handling is needed here because we abuse gl_Color/gl_SecondaryColor flat shading */
 			else if (input->source == GPU_SOURCE_OPENGL_BUILTIN) {
@@ -738,6 +812,61 @@ static char *code_generate_vertex(ListBase *nodes, const GPUMatType type)
 	return code;
 }
 
+static char *code_generate_geometry(ListBase *nodes, bool use_opensubdiv)
+{
+#ifdef WITH_OPENSUBDIV
+	if (use_opensubdiv) {
+		DynStr *ds = BLI_dynstr_new();
+		GPUNode *node;
+		GPUInput *input;
+		char *code;
+
+		/* Generate varying declarations. */
+		for (node = nodes->first; node; node = node->next) {
+			for (input = node->inputs.first; input; input = input->next) {
+				if (input->source == GPU_SOURCE_ATTRIB && input->attribfirst) {
+					if (input->attribtype == CD_MTFACE) {
+						BLI_dynstr_appendf(ds, "varying %s var%d;\n",
+						                   GPU_DATATYPE_STR[input->type],
+						                   input->attribid);
+						BLI_dynstr_appendf(ds, "uniform int fvar%d_offset;\n",
+						                   input->attribid);
+					}
+				}
+			}
+		}
+
+		BLI_dynstr_append(ds, datatoc_gpu_shader_geometry_glsl);
+
+		/* Generate varying assignments. */
+		for (node = nodes->first; node; node = node->next) {
+			for (input = node->inputs.first; input; input = input->next) {
+				if (input->source == GPU_SOURCE_ATTRIB && input->attribfirst) {
+					if (input->attribtype == CD_MTFACE) {
+						BLI_dynstr_appendf(ds,
+						                   "\tINTERP_FACE_VARYING_2(var%d, "
+						                       "fvar%d_offset, st);\n",
+						                   input->attribid,
+						                   input->attribid);
+					}
+				}
+			}
+		}
+
+		BLI_dynstr_append(ds, "}\n\n");
+		code = BLI_dynstr_get_cstring(ds);
+		BLI_dynstr_free(ds);
+
+		//if (G.debug & G_DEBUG) printf("%s\n", code);
+
+		return code;
+	}
+#else
+	UNUSED_VARS(nodes, use_opensubdiv);
+#endif
+	return NULL;
+}
+
 void GPU_code_generate_glsl_lib(void)
 {
 	DynStr *ds;
@@ -786,8 +915,28 @@ static void gpu_nodes_extract_dynamic_inputs(GPUPass *pass, ListBase *nodes)
 
 			/* attributes don't need to be bound, they already have
 			 * an id that the drawing functions will use */
-			if (input->source == GPU_SOURCE_ATTRIB ||
-			    input->source == GPU_SOURCE_BUILTIN ||
+			if (input->source == GPU_SOURCE_ATTRIB) {
+#ifdef WITH_OPENSUBDIV
+				/* We do need mtface attributes for later, so we can
+				 * update face-varuing variables offset in the texture
+				 * buffer for proper sampling from the shader.
+				 *
+				 * We don't do anything about attribute itself, we
+				 * only use it to learn which uniform name is to be
+				 * updated.
+				 *
+				 * TODO(sergey): We can add ad extra uniform input
+				 * for the offset, which will be purely internal and
+				 * which would avoid having such an exceptions.
+				 */
+				if (input->attribtype != CD_MTFACE) {
+					continue;
+				}
+#else
+				continue;
+#endif
+			}
+			if (input->source == GPU_SOURCE_BUILTIN ||
 			    input->source == GPU_SOURCE_OPENGL_BUILTIN)
 			{
 				continue;
@@ -811,6 +960,14 @@ static void gpu_nodes_extract_dynamic_inputs(GPUPass *pass, ListBase *nodes)
 			if (extract)
 				input->shaderloc = GPU_shader_get_uniform(shader, input->shadername);
 
+#ifdef WITH_OPENSUBDIV
+			if (input->source == GPU_SOURCE_ATTRIB &&
+			    input->attribtype == CD_MTFACE)
+			{
+				extract = 1;
+			}
+#endif
+
 			/* extract nodes */
 			if (extract) {
 				BLI_remlink(&node->inputs, input);
@@ -1432,11 +1589,11 @@ static void gpu_nodes_prune(ListBase *nodes, GPUNodeLink *outlink)
 
 GPUPass *GPU_generate_pass(ListBase *nodes, GPUNodeLink *outlink,
 						   GPUVertexAttribs *attribs, int *builtins,
-						   const GPUMatType type, const char *UNUSED(name))
+						   const GPUMatType type, const char *UNUSED(name), const bool use_opensubdiv)
 {
 	GPUShader *shader;
 	GPUPass *pass;
-	char *vertexcode, *fragmentcode;
+	char *vertexcode, *geometrycode, *fragmentcode;
 
 #if 0
 	if (!FUNCTION_LIB) {
@@ -1454,7 +1611,8 @@ GPUPass *GPU_generate_pass(ListBase *nodes, GPUNodeLink *outlink,
 	/* generate code and compile with opengl */
 	fragmentcode = code_generate_fragment(nodes, outlink->output);
 	vertexcode = code_generate_vertex(nodes, type);
-	shader = GPU_shader_create(vertexcode, fragmentcode, NULL, glsl_material_library, NULL, 0, 0, 0);
+	geometrycode = code_generate_geometry(nodes, use_opensubdiv);
+	shader = GPU_shader_create(vertexcode, fragmentcode, geometrycode, glsl_material_library, NULL, 0, 0, 0);
 
 	/* failed? */
 	if (!shader) {
@@ -1474,6 +1632,7 @@ GPUPass *GPU_generate_pass(ListBase *nodes, GPUNodeLink *outlink,
 	pass->output = outlink->output;
 	pass->shader = shader;
 	pass->fragmentcode = fragmentcode;
+	pass->geometrycode = geometrycode;
 	pass->vertexcode = vertexcode;
 	pass->libcode = glsl_material_library;
 
@@ -1490,8 +1649,9 @@ void GPU_pass_free(GPUPass *pass)
 	gpu_inputs_free(&pass->inputs);
 	if (pass->fragmentcode)
 		MEM_freeN(pass->fragmentcode);
+	if (pass->geometrycode)
+		MEM_freeN(pass->geometrycode);
 	if (pass->vertexcode)
 		MEM_freeN(pass->vertexcode);
 	MEM_freeN(pass);
 }
-