diff options
25 files changed, 904 insertions, 4 deletions
diff --git a/intern/cycles/blender/blender_object.cpp b/intern/cycles/blender/blender_object.cpp index d8c4ce9c5df..5e99759fdac 100644 --- a/intern/cycles/blender/blender_object.cpp +++ b/intern/cycles/blender/blender_object.cpp @@ -234,6 +234,10 @@ Object *BlenderSync::sync_object(BL::Depsgraph &b_depsgraph, /* special case not tracked by object update flags */ + if (sync_object_attributes(b_instance, object)) { + object_updated = true; + } + /* holdout */ if (use_holdout != object->use_holdout) { object->use_holdout = use_holdout; @@ -343,6 +347,132 @@ Object *BlenderSync::sync_object(BL::Depsgraph &b_depsgraph, return object; } +/* This function mirrors drw_uniform_property_lookup in draw_instance_data.cpp */ +static bool lookup_property(BL::ID b_id, const string &name, float4 *r_value) +{ + PointerRNA ptr; + PropertyRNA *prop; + + if (!RNA_path_resolve(&b_id.ptr, name.c_str(), &ptr, &prop)) { + return false; + } + + PropertyType type = RNA_property_type(prop); + int arraylen = RNA_property_array_length(&ptr, prop); + + if (arraylen == 0) { + float value; + + if (type == PROP_FLOAT) + value = RNA_property_float_get(&ptr, prop); + else if (type == PROP_INT) + value = RNA_property_int_get(&ptr, prop); + else + return false; + + *r_value = make_float4(value, value, value, 1.0f); + return true; + } + else if (type == PROP_FLOAT && arraylen <= 4) { + *r_value = make_float4(0.0f, 0.0f, 0.0f, 1.0f); + RNA_property_float_get_array(&ptr, prop, &r_value->x); + return true; + } + + return false; +} + +/* This function mirrors drw_uniform_attribute_lookup in draw_instance_data.cpp */ +static float4 lookup_instance_property(BL::DepsgraphObjectInstance &b_instance, + const string &name, + bool use_instancer) +{ + string idprop_name = string_printf("[\"%s\"]", name.c_str()); + float4 value; + + /* If requesting instance data, check the parent particle system and object. */ + if (use_instancer && b_instance.is_instance()) { + BL::ParticleSystem b_psys = b_instance.particle_system(); + + if (b_psys) { + if (lookup_property(b_psys.settings(), idprop_name, &value) || + lookup_property(b_psys.settings(), name, &value)) { + return value; + } + } + if (lookup_property(b_instance.parent(), idprop_name, &value) || + lookup_property(b_instance.parent(), name, &value)) { + return value; + } + } + + /* Check the object and mesh. */ + BL::Object b_ob = b_instance.object(); + BL::ID b_data = b_ob.data(); + + if (lookup_property(b_ob, idprop_name, &value) || lookup_property(b_ob, name, &value) || + lookup_property(b_data, idprop_name, &value) || lookup_property(b_data, name, &value)) { + return value; + } + + return make_float4(0.0f); +} + +bool BlenderSync::sync_object_attributes(BL::DepsgraphObjectInstance &b_instance, Object *object) +{ + /* Find which attributes are needed. */ + AttributeRequestSet requests = object->geometry->needed_attributes(); + + /* Delete attributes that became unnecessary. */ + vector<ParamValue> &attributes = object->attributes; + bool changed = false; + + for (int i = attributes.size() - 1; i >= 0; i--) { + if (!requests.find(attributes[i].name())) { + attributes.erase(attributes.begin() + i); + changed = true; + } + } + + /* Update attribute values. */ + foreach (AttributeRequest &req, requests.requests) { + ustring name = req.name; + + std::string real_name; + BlenderAttributeType type = blender_attribute_name_split_type(name, &real_name); + + if (type != BL::ShaderNodeAttribute::attribute_type_GEOMETRY) { + bool use_instancer = (type == BL::ShaderNodeAttribute::attribute_type_INSTANCER); + float4 value = lookup_instance_property(b_instance, real_name, use_instancer); + + /* Try finding the existing attribute value. */ + ParamValue *param = NULL; + + for (size_t i = 0; i < attributes.size(); i++) { + if (attributes[i].name() == name) { + param = &attributes[i]; + break; + } + } + + /* Replace or add the value. */ + ParamValue new_param(name, TypeDesc::TypeFloat4, 1, &value); + assert(new_param.datasize() == sizeof(value)); + + if (!param) { + changed = true; + attributes.push_back(new_param); + } + else if (memcmp(param->data(), &value, sizeof(value)) != 0) { + changed = true; + *param = new_param; + } + } + } + + return changed; +} + /* Object Loop */ void BlenderSync::sync_objects(BL::Depsgraph &b_depsgraph, diff --git a/intern/cycles/blender/blender_shader.cpp b/intern/cycles/blender/blender_shader.cpp index c171982b29d..0c1b240c593 100644 --- a/intern/cycles/blender/blender_shader.cpp +++ b/intern/cycles/blender/blender_shader.cpp @@ -97,6 +97,53 @@ static ImageAlphaType get_image_alpha_type(BL::Image &b_image) return (ImageAlphaType)validate_enum_value(value, IMAGE_ALPHA_NUM_TYPES, IMAGE_ALPHA_AUTO); } +/* Attribute name translation utilities */ + +/* Since Eevee needs to know whether the attribute is uniform or varying + * at the time it compiles the shader for the material, Blender had to + * introduce different namespaces (types) in its attribute node. However, + * Cycles already has object attributes that form a uniform namespace with + * the more common varying attributes. Without completely reworking the + * attribute handling in Cycles to introduce separate namespaces (this could + * be especially hard for OSL which directly uses the name string), the + * space identifier has to be added to the attribute name as a prefix. + * + * The prefixes include a control character to ensure the user specified + * name can't accidentally include a special prefix. + */ + +static const string_view object_attr_prefix("\x01object:"); +static const string_view instancer_attr_prefix("\x01instancer:"); + +static ustring blender_attribute_name_add_type(const string &name, BlenderAttributeType type) +{ + switch (type) { + case BL::ShaderNodeAttribute::attribute_type_OBJECT: + return ustring::concat(object_attr_prefix, name); + case BL::ShaderNodeAttribute::attribute_type_INSTANCER: + return ustring::concat(instancer_attr_prefix, name); + default: + return ustring(name); + } +} + +BlenderAttributeType blender_attribute_name_split_type(ustring name, string *r_real_name) +{ + string_view sname(name); + + if (sname.substr(0, object_attr_prefix.size()) == object_attr_prefix) { + *r_real_name = sname.substr(object_attr_prefix.size()); + return BL::ShaderNodeAttribute::attribute_type_OBJECT; + } + + if (sname.substr(0, instancer_attr_prefix.size()) == instancer_attr_prefix) { + *r_real_name = sname.substr(instancer_attr_prefix.size()); + return BL::ShaderNodeAttribute::attribute_type_INSTANCER; + } + + return BL::ShaderNodeAttribute::attribute_type_GEOMETRY; +} + /* Graph */ static BL::NodeSocket get_node_output(BL::Node &b_node, const string &name) @@ -369,7 +416,8 @@ static ShaderNode *add_node(Scene *scene, else if (b_node.is_a(&RNA_ShaderNodeAttribute)) { BL::ShaderNodeAttribute b_attr_node(b_node); AttributeNode *attr = graph->create_node<AttributeNode>(); - attr->attribute = b_attr_node.attribute_name(); + attr->attribute = blender_attribute_name_add_type(b_attr_node.attribute_name(), + b_attr_node.attribute_type()); node = attr; } else if (b_node.is_a(&RNA_ShaderNodeBackground)) { diff --git a/intern/cycles/blender/blender_sync.h b/intern/cycles/blender/blender_sync.h index 360468da0ef..8ca021ff0d2 100644 --- a/intern/cycles/blender/blender_sync.h +++ b/intern/cycles/blender/blender_sync.h @@ -149,6 +149,8 @@ class BlenderSync { bool *use_portal, TaskPool *geom_task_pool); + bool sync_object_attributes(BL::DepsgraphObjectInstance &b_instance, Object *object); + /* Volume */ void sync_volume(BL::Object &b_ob, Volume *volume); diff --git a/intern/cycles/blender/blender_util.h b/intern/cycles/blender/blender_util.h index 1ea34b41aa2..5185ebae789 100644 --- a/intern/cycles/blender/blender_util.h +++ b/intern/cycles/blender/blender_util.h @@ -40,6 +40,9 @@ float *BKE_image_get_float_pixels_for_frame(void *image, int frame, int tile); CCL_NAMESPACE_BEGIN +typedef BL::ShaderNodeAttribute::attribute_type_enum BlenderAttributeType; +BlenderAttributeType blender_attribute_name_split_type(ustring name, string *r_real_name); + void python_thread_state_save(void **python_thread_state); void python_thread_state_restore(void **python_thread_state); diff --git a/source/blender/blenlib/BLI_bitmap.h b/source/blender/blenlib/BLI_bitmap.h index 960ce44c58c..c97be6eed3c 100644 --- a/source/blender/blenlib/BLI_bitmap.h +++ b/source/blender/blenlib/BLI_bitmap.h @@ -106,7 +106,7 @@ typedef unsigned int BLI_bitmap; #define BLI_BITMAP_RESIZE(_bitmap, _tot) \ { \ CHECK_TYPE(_bitmap, BLI_bitmap *); \ - (_bitmap) = MEM_reallocN(_bitmap, BLI_BITMAP_SIZE(_tot)); \ + (_bitmap) = MEM_recallocN(_bitmap, BLI_BITMAP_SIZE(_tot)); \ } \ (void)0 diff --git a/source/blender/draw/DRW_engine.h b/source/blender/draw/DRW_engine.h index ca5c2c94b40..2d5b93f4272 100644 --- a/source/blender/draw/DRW_engine.h +++ b/source/blender/draw/DRW_engine.h @@ -36,6 +36,7 @@ struct ARegion; struct DRWInstanceDataList; struct Depsgraph; struct DrawEngineType; +struct GHash; struct GPUMaterial; struct GPUOffScreen; struct GPUViewport; @@ -140,6 +141,7 @@ void DRW_render_gpencil(struct RenderEngine *engine, struct Depsgraph *depsgraph /* This is here because GPUViewport needs it */ struct DRWInstanceDataList *DRW_instance_data_list_create(void); void DRW_instance_data_list_free(struct DRWInstanceDataList *idatalist); +void DRW_uniform_attrs_pool_free(struct GHash *table); void DRW_render_context_enable(struct Render *render); void DRW_render_context_disable(struct Render *render); diff --git a/source/blender/draw/intern/draw_instance_data.c b/source/blender/draw/intern/draw_instance_data.c index 4050a5f8b69..47710c32916 100644 --- a/source/blender/draw/intern/draw_instance_data.c +++ b/source/blender/draw/intern/draw_instance_data.c @@ -30,9 +30,20 @@ */ #include "draw_instance_data.h" +#include "draw_manager.h" + #include "DRW_engine.h" #include "DRW_render.h" /* For DRW_shgroup_get_instance_count() */ +#include "GPU_material.h" + +#include "DNA_particle_types.h" + +#include "BKE_duplilist.h" + +#include "RNA_access.h" + +#include "BLI_bitmap.h" #include "BLI_memblock.h" #include "BLI_mempool.h" #include "BLI_utildefines.h" @@ -408,3 +419,359 @@ void DRW_instance_data_list_resize(DRWInstanceDataList *idatalist) } /** \} */ +/* -------------------------------------------------------------------- */ +/** \name Sparse Uniform Buffer + * \{ */ + +#define CHUNK_LIST_STEP (1 << 4) + +/** A chunked UBO manager that doesn't actually allocate unneeded chunks. */ +typedef struct DRWSparseUniformBuf { + /* Memory buffers used to stage chunk data before transfer to UBOs. */ + char **chunk_buffers; + /* Uniform buffer objects with flushed data. */ + struct GPUUniformBuf **chunk_ubos; + /* True if the relevant chunk contains data (distinct from simply being allocated). */ + BLI_bitmap *chunk_used; + + int num_chunks; + unsigned int item_size, chunk_size, chunk_bytes; +} DRWSparseUniformBuf; + +static void drw_sparse_uniform_buffer_init(DRWSparseUniformBuf *buffer, + unsigned int item_size, + unsigned int chunk_size) +{ + buffer->chunk_buffers = NULL; + buffer->chunk_used = NULL; + buffer->chunk_ubos = NULL; + buffer->num_chunks = 0; + buffer->item_size = item_size; + buffer->chunk_size = chunk_size; + buffer->chunk_bytes = item_size * chunk_size; +} + +/** Allocate a chunked UBO with the specified item and chunk size. */ +DRWSparseUniformBuf *DRW_sparse_uniform_buffer_new(unsigned int item_size, unsigned int chunk_size) +{ + DRWSparseUniformBuf *buf = MEM_mallocN(sizeof(DRWSparseUniformBuf), __func__); + drw_sparse_uniform_buffer_init(buf, item_size, chunk_size); + return buf; +} + +/** Flush data from ordinary memory to UBOs. */ +void DRW_sparse_uniform_buffer_flush(DRWSparseUniformBuf *buffer) +{ + for (int i = 0; i < buffer->num_chunks; i++) { + if (BLI_BITMAP_TEST(buffer->chunk_used, i)) { + if (buffer->chunk_ubos[i] == NULL) { + buffer->chunk_ubos[i] = GPU_uniformbuf_create(buffer->chunk_bytes); + } + GPU_uniformbuf_update(buffer->chunk_ubos[i], buffer->chunk_buffers[i]); + } + } +} + +/** Clean all buffers and free unused ones. */ +void DRW_sparse_uniform_buffer_clear(DRWSparseUniformBuf *buffer, bool free_all) +{ + int max_used_chunk = 0; + + for (int i = 0; i < buffer->num_chunks; i++) { + /* Delete buffers that were not used since the last clear call. */ + if (free_all || !BLI_BITMAP_TEST(buffer->chunk_used, i)) { + MEM_SAFE_FREE(buffer->chunk_buffers[i]); + + if (buffer->chunk_ubos[i]) { + GPU_uniformbuf_free(buffer->chunk_ubos[i]); + buffer->chunk_ubos[i] = NULL; + } + } + else { + max_used_chunk = i + 1; + } + } + + /* Shrink the chunk array if appropriate. */ + const int old_num_chunks = buffer->num_chunks; + + buffer->num_chunks = (max_used_chunk + CHUNK_LIST_STEP - 1) & ~(CHUNK_LIST_STEP - 1); + + if (buffer->num_chunks == 0) { + /* Ensure that an empty pool holds no memory allocations. */ + MEM_SAFE_FREE(buffer->chunk_buffers); + MEM_SAFE_FREE(buffer->chunk_used); + MEM_SAFE_FREE(buffer->chunk_ubos); + return; + } + + if (buffer->num_chunks != old_num_chunks) { + buffer->chunk_buffers = MEM_recallocN(buffer->chunk_buffers, + buffer->num_chunks * sizeof(void *)); + buffer->chunk_ubos = MEM_recallocN(buffer->chunk_ubos, buffer->num_chunks * sizeof(void *)); + BLI_BITMAP_RESIZE(buffer->chunk_used, buffer->num_chunks); + } + + BLI_bitmap_set_all(buffer->chunk_used, false, buffer->num_chunks); +} + +/** Frees the buffer. */ +void DRW_sparse_uniform_buffer_free(DRWSparseUniformBuf *buffer) +{ + DRW_sparse_uniform_buffer_clear(buffer, true); + MEM_freeN(buffer); +} + +/** Checks if the buffer contains any allocated chunks. */ +bool DRW_sparse_uniform_buffer_is_empty(DRWSparseUniformBuf *buffer) +{ + return buffer->num_chunks == 0; +} + +static GPUUniformBuf *drw_sparse_uniform_buffer_get_ubo(DRWSparseUniformBuf *buffer, int chunk) +{ + if (buffer && chunk < buffer->num_chunks && BLI_BITMAP_TEST(buffer->chunk_used, chunk)) { + return buffer->chunk_ubos[chunk]; + } + else { + return NULL; + } +} + +/** Bind the UBO for the given chunk, if present. A NULL buffer pointer is handled as empty. */ +void DRW_sparse_uniform_buffer_bind(DRWSparseUniformBuf *buffer, int chunk, int location) +{ + GPUUniformBuf *ubo = drw_sparse_uniform_buffer_get_ubo(buffer, chunk); + if (ubo) { + GPU_uniformbuf_bind(ubo, location); + } +} + +/** Unbind the UBO for the given chunk, if present. A NULL buffer pointer is handled as empty. */ +void DRW_sparse_uniform_buffer_unbind(DRWSparseUniformBuf *buffer, int chunk) +{ + GPUUniformBuf *ubo = drw_sparse_uniform_buffer_get_ubo(buffer, chunk); + if (ubo) { + GPU_uniformbuf_unbind(ubo); + } +} + +/** Returns a pointer to the given item of the given chunk, allocating memory if necessary. */ +void *DRW_sparse_uniform_buffer_ensure_item(DRWSparseUniformBuf *pool, int chunk, int item) +{ + if (chunk >= pool->num_chunks) { + pool->num_chunks = (chunk + CHUNK_LIST_STEP) & ~(CHUNK_LIST_STEP - 1); + pool->chunk_buffers = MEM_recallocN(pool->chunk_buffers, pool->num_chunks * sizeof(void *)); + pool->chunk_ubos = MEM_recallocN(pool->chunk_ubos, pool->num_chunks * sizeof(void *)); + BLI_BITMAP_RESIZE(pool->chunk_used, pool->num_chunks); + } + + char *buffer = pool->chunk_buffers[chunk]; + + if (buffer == NULL) { + pool->chunk_buffers[chunk] = buffer = MEM_callocN(pool->chunk_bytes, __func__); + } + else if (!BLI_BITMAP_TEST(pool->chunk_used, chunk)) { + memset(buffer, 0, pool->chunk_bytes); + } + + BLI_BITMAP_ENABLE(pool->chunk_used, chunk); + + return buffer + pool->item_size * item; +} + +/** \} */ + +/* -------------------------------------------------------------------- */ +/** \name Uniform Attribute Buffers + * \{ */ + +/** Sparse UBO buffer for a specific uniform attribute list. */ +typedef struct DRWUniformAttrBuf { + /* Attribute list (also used as hash table key) handled by this buffer. */ + GPUUniformAttrList key; + /* Sparse UBO buffer containing the attribute values. */ + DRWSparseUniformBuf ubos; + /* Last handle used to update the buffer, checked for avoiding redundant updates. */ + DRWResourceHandle last_handle; + /* Linked list pointer used for freeing the empty unneeded buffers. */ + struct DRWUniformAttrBuf *next_empty; +} DRWUniformAttrBuf; + +static DRWUniformAttrBuf *drw_uniform_attrs_pool_ensure(GHash *table, GPUUniformAttrList *key) +{ + void **pkey, **pval; + + if (!BLI_ghash_ensure_p_ex(table, key, &pkey, &pval)) { + DRWUniformAttrBuf *buffer = MEM_callocN(sizeof(*buffer), __func__); + + *pkey = &buffer->key; + *pval = buffer; + + GPU_uniform_attr_list_copy(&buffer->key, key); + drw_sparse_uniform_buffer_init( + &buffer->ubos, key->count * sizeof(float[4]), DRW_RESOURCE_CHUNK_LEN); + + buffer->last_handle = (DRWResourceHandle)-1; + } + + return (DRWUniformAttrBuf *)*pval; +} + +/* This function mirrors lookup_property in cycles/blender/blender_object.cpp */ +static bool drw_uniform_property_lookup(ID *id, const char *name, float r_data[4]) +{ + PointerRNA ptr, id_ptr; + PropertyRNA *prop; + + if (!id) { + return false; + } + + RNA_id_pointer_create(id, &id_ptr); + + if (!RNA_path_resolve(&id_ptr, name, &ptr, &prop)) { + return false; + } + + PropertyType type = RNA_property_type(prop); + int arraylen = RNA_property_array_length(&ptr, prop); + + if (arraylen == 0) { + float value; + + if (type == PROP_FLOAT) + value = RNA_property_float_get(&ptr, prop); + else if (type == PROP_INT) + value = RNA_property_int_get(&ptr, prop); + else + return false; + + copy_v4_fl4(r_data, value, value, value, 1); + return true; + } + else if (type == PROP_FLOAT && arraylen <= 4) { + copy_v4_fl4(r_data, 0, 0, 0, 1); + RNA_property_float_get_array(&ptr, prop, r_data); + return true; + } + + return false; +} + +/* This function mirrors lookup_instance_property in cycles/blender/blender_object.cpp */ +static void drw_uniform_attribute_lookup(GPUUniformAttr *attr, + Object *ob, + Object *dupli_parent, + DupliObject *dupli_source, + float r_data[4]) +{ + char idprop_name[sizeof(attr->name) + 4]; + + copy_v4_fl(r_data, 0); + sprintf(idprop_name, "[\"%s\"]", attr->name); + + /* If requesting instance data, check the parent particle system and object. */ + if (attr->use_dupli) { + if (dupli_source && dupli_source->particle_system) { + ParticleSettings *settings = dupli_source->particle_system->part; + if (drw_uniform_property_lookup((ID *)settings, idprop_name, r_data) || + drw_uniform_property_lookup((ID *)settings, attr->name, r_data)) { + return; + } + } + if (drw_uniform_property_lookup((ID *)dupli_parent, idprop_name, r_data) || + drw_uniform_property_lookup((ID *)dupli_parent, attr->name, r_data)) { + return; + } + } + + /* Check the object and mesh. */ + if (ob) { + if (drw_uniform_property_lookup((ID *)ob, idprop_name, r_data) || + drw_uniform_property_lookup((ID *)ob, attr->name, r_data) || + drw_uniform_property_lookup((ID *)ob->data, idprop_name, r_data) || + drw_uniform_property_lookup((ID *)ob->data, attr->name, r_data)) { + return; + } + } +} + +void drw_uniform_attrs_pool_update(GHash *table, + GPUUniformAttrList *key, + DRWResourceHandle *handle, + Object *ob, + Object *dupli_parent, + DupliObject *dupli_source) +{ + DRWUniformAttrBuf *buffer = drw_uniform_attrs_pool_ensure(table, key); + + if (buffer->last_handle != *handle) { + buffer->last_handle = *handle; + + int chunk = DRW_handle_chunk_get(handle); + int item = DRW_handle_id_get(handle); + float(*values)[4] = DRW_sparse_uniform_buffer_ensure_item(&buffer->ubos, chunk, item); + + LISTBASE_FOREACH (GPUUniformAttr *, attr, &buffer->key.list) { + drw_uniform_attribute_lookup(attr, ob, dupli_parent, dupli_source, *values++); + } + } +} + +DRWSparseUniformBuf *DRW_uniform_attrs_pool_find_ubo(GHash *table, struct GPUUniformAttrList *key) +{ + DRWUniformAttrBuf *buffer = BLI_ghash_lookup(table, key); + return buffer ? &buffer->ubos : NULL; +} + +GHash *DRW_uniform_attrs_pool_new() +{ + return GPU_uniform_attr_list_hash_new("obattr_hash"); +} + +void DRW_uniform_attrs_pool_flush_all(GHash *table) +{ + GHASH_FOREACH_BEGIN (DRWUniformAttrBuf *, buffer, table) { + DRW_sparse_uniform_buffer_flush(&buffer->ubos); + } + GHASH_FOREACH_END(); +} + +static void drw_uniform_attrs_pool_free_cb(void *ptr) +{ + DRWUniformAttrBuf *buffer = ptr; + + GPU_uniform_attr_list_free(&buffer->key); + DRW_sparse_uniform_buffer_clear(&buffer->ubos, true); + MEM_freeN(buffer); +} + +void DRW_uniform_attrs_pool_clear_all(GHash *table) +{ + DRWUniformAttrBuf *remove_list = NULL; + + GHASH_FOREACH_BEGIN (DRWUniformAttrBuf *, buffer, table) { + buffer->last_handle = (DRWResourceHandle)-1; + DRW_sparse_uniform_buffer_clear(&buffer->ubos, false); + + if (DRW_sparse_uniform_buffer_is_empty(&buffer->ubos)) { + buffer->next_empty = remove_list; + remove_list = buffer; + } + } + GHASH_FOREACH_END(); + + while (remove_list) { + DRWUniformAttrBuf *buffer = remove_list; + remove_list = buffer->next_empty; + BLI_ghash_remove(table, &buffer->key, NULL, drw_uniform_attrs_pool_free_cb); + } +} + +void DRW_uniform_attrs_pool_free(GHash *table) +{ + BLI_ghash_free(table, NULL, drw_uniform_attrs_pool_free_cb); +} + +/** \} */ diff --git a/source/blender/draw/intern/draw_instance_data.h b/source/blender/draw/intern/draw_instance_data.h index e562d99097e..c959a9e19d6 100644 --- a/source/blender/draw/intern/draw_instance_data.h +++ b/source/blender/draw/intern/draw_instance_data.h @@ -31,8 +31,12 @@ #define DRW_BUFFER_VERTS_CHUNK 128 +struct GHash; +struct GPUUniformAttrList; + typedef struct DRWInstanceData DRWInstanceData; typedef struct DRWInstanceDataList DRWInstanceDataList; +typedef struct DRWSparseUniformBuf DRWSparseUniformBuf; void *DRW_instance_data_next(DRWInstanceData *idata); DRWInstanceData *DRW_instance_data_request(DRWInstanceDataList *idatalist, uint attr_size); @@ -54,3 +58,21 @@ void DRW_instance_buffer_finish(DRWInstanceDataList *idatalist); void DRW_instance_data_list_reset(DRWInstanceDataList *idatalist); void DRW_instance_data_list_free_unused(DRWInstanceDataList *idatalist); void DRW_instance_data_list_resize(DRWInstanceDataList *idatalist); + +/* Sparse chunked UBO manager. */ +DRWSparseUniformBuf *DRW_sparse_uniform_buffer_new(unsigned int item_size, + unsigned int chunk_size); +void DRW_sparse_uniform_buffer_flush(DRWSparseUniformBuf *buffer); +void DRW_sparse_uniform_buffer_clear(DRWSparseUniformBuf *buffer, bool free_all); +void DRW_sparse_uniform_buffer_free(DRWSparseUniformBuf *buffer); +bool DRW_sparse_uniform_buffer_is_empty(DRWSparseUniformBuf *buffer); +void DRW_sparse_uniform_buffer_bind(DRWSparseUniformBuf *buffer, int chunk, int location); +void DRW_sparse_uniform_buffer_unbind(DRWSparseUniformBuf *buffer, int chunk); +void *DRW_sparse_uniform_buffer_ensure_item(DRWSparseUniformBuf *buffer, int chunk, int item); + +/* Uniform attribute UBO management. */ +struct GHash *DRW_uniform_attrs_pool_new(void); +void DRW_uniform_attrs_pool_flush_all(struct GHash *table); +void DRW_uniform_attrs_pool_clear_all(struct GHash *table); +struct DRWSparseUniformBuf *DRW_uniform_attrs_pool_find_ubo(struct GHash *table, + struct GPUUniformAttrList *key); diff --git a/source/blender/draw/intern/draw_manager.c b/source/blender/draw/intern/draw_manager.c index b60583deda3..f51328e9bc9 100644 --- a/source/blender/draw/intern/draw_manager.c +++ b/source/blender/draw/intern/draw_manager.c @@ -467,6 +467,8 @@ static void drw_viewport_cache_resize(void) BLI_memblock_clear(DST.vmempool->passes, NULL); BLI_memblock_clear(DST.vmempool->views, NULL); BLI_memblock_clear(DST.vmempool->images, NULL); + + DRW_uniform_attrs_pool_clear_all(DST.vmempool->obattrs_ubo_pool); } DRW_instance_data_list_free_unused(DST.idatalist); @@ -593,6 +595,9 @@ static void drw_viewport_var_init(void) if (DST.vmempool->images == NULL) { DST.vmempool->images = BLI_memblock_create(sizeof(GPUTexture *)); } + if (DST.vmempool->obattrs_ubo_pool == NULL) { + DST.vmempool->obattrs_ubo_pool = DRW_uniform_attrs_pool_new(); + } DST.resource_handle = 0; DST.pass_handle = 0; diff --git a/source/blender/draw/intern/draw_manager.h b/source/blender/draw/intern/draw_manager.h index 8e505d5df71..dae204ad9f7 100644 --- a/source/blender/draw/intern/draw_manager.h +++ b/source/blender/draw/intern/draw_manager.h @@ -43,6 +43,9 @@ #include "draw_instance_data.h" +struct Object; +struct DupliObject; + /* Use draw manager to call GPU_select, see: DRW_draw_select_loop */ #define USE_GPU_SELECT @@ -286,6 +289,7 @@ typedef enum { /** Per drawcall uniforms/UBO */ DRW_UNIFORM_BLOCK_OBMATS, DRW_UNIFORM_BLOCK_OBINFOS, + DRW_UNIFORM_BLOCK_OBATTRS, DRW_UNIFORM_RESOURCE_CHUNK, DRW_UNIFORM_RESOURCE_ID, /** Legacy / Fallback */ @@ -317,6 +321,8 @@ struct DRWUniform { float fvalue[4]; /* DRW_UNIFORM_INT_COPY */ int ivalue[4]; + /* DRW_UNIFORM_BLOCK_OBATTRS */ + struct GPUUniformAttrList *uniform_attrs; }; int location; /* Uniform location or binding point for textures and ubos. */ uint8_t type; /* DRWUniformType */ @@ -340,6 +346,9 @@ struct DRWShadingGroup { struct { int objectinfo; /* Equal to 1 if the shader needs obinfos. */ DRWResourceHandle pass_handle; /* Memblock key to parent pass. */ + + /* Set of uniform attributes used by this shader. */ + struct GPUUniformAttrList *uniform_attrs; }; /* This struct is used after cache populate if using the Z sorting. * It will not conflict with the above struct. */ @@ -598,3 +607,10 @@ void drw_resource_buffer_finish(ViewportMemoryPool *vmempool); GPUBatch *drw_cache_procedural_points_get(void); GPUBatch *drw_cache_procedural_lines_get(void); GPUBatch *drw_cache_procedural_triangles_get(void); + +void drw_uniform_attrs_pool_update(struct GHash *table, + struct GPUUniformAttrList *key, + DRWResourceHandle *handle, + struct Object *ob, + struct Object *dupli_parent, + struct DupliObject *dupli_source); diff --git a/source/blender/draw/intern/draw_manager_data.c b/source/blender/draw/intern/draw_manager_data.c index 5f394804bcf..f1d3f8c8d5a 100644 --- a/source/blender/draw/intern/draw_manager_data.c +++ b/source/blender/draw/intern/draw_manager_data.c @@ -128,6 +128,8 @@ void drw_resource_buffer_finish(ViewportMemoryPool *vmempool) GPU_uniformbuf_update(vmempool->obinfos_ubo[i], data_infos); } + DRW_uniform_attrs_pool_flush_all(vmempool->obattrs_ubo_pool); + /* Aligned alloc to avoid unaligned memcpy. */ DRWCommandChunk *chunk_tmp = MEM_mallocN_aligned(sizeof(DRWCommandChunk), 16, "tmp call chunk"); DRWCommandChunk *chunk; @@ -209,6 +211,9 @@ static void drw_shgroup_uniform_create_ex(DRWShadingGroup *shgroup, uni->texture_ref = (GPUTexture **)value; uni->sampler_state = sampler_state; break; + case DRW_UNIFORM_BLOCK_OBATTRS: + uni->uniform_attrs = (GPUUniformAttrList *)value; + break; default: uni->pvalue = (const float *)value; break; @@ -611,6 +616,15 @@ static DRWResourceHandle drw_resource_handle(DRWShadingGroup *shgroup, } } + if (shgroup->uniform_attrs) { + drw_uniform_attrs_pool_update(DST.vmempool->obattrs_ubo_pool, + shgroup->uniform_attrs, + &DST.ob_handle, + ob, + DST.dupli_parent, + DST.dupli_source); + } + return DST.ob_handle; } @@ -1184,6 +1198,7 @@ void DRW_buffer_add_entry_array(DRWCallBuffer *callbuf, const void *attr[], uint static void drw_shgroup_init(DRWShadingGroup *shgroup, GPUShader *shader) { shgroup->uniforms = NULL; + shgroup->uniform_attrs = NULL; int view_ubo_location = GPU_shader_get_builtin_block(shader, GPU_UNIFORM_BLOCK_VIEW); int model_ubo_location = GPU_shader_get_builtin_block(shader, GPU_UNIFORM_BLOCK_MODEL); @@ -1329,6 +1344,13 @@ void DRW_shgroup_add_material_resources(DRWShadingGroup *grp, struct GPUMaterial if (ubo != NULL) { DRW_shgroup_uniform_block(grp, GPU_UBO_BLOCK_NAME, ubo); } + + GPUUniformAttrList *uattrs = GPU_material_uniform_attributes(material); + if (uattrs != NULL) { + int loc = GPU_shader_get_uniform_block_binding(grp->shader, GPU_ATTRIBUTE_UBO_BLOCK_NAME); + drw_shgroup_uniform_create_ex(grp, loc, DRW_UNIFORM_BLOCK_OBATTRS, uattrs, 0, 0, 1); + grp->uniform_attrs = uattrs; + } } GPUVertFormat *DRW_shgroup_instance_format_array(const DRWInstanceAttrFormat attrs[], diff --git a/source/blender/draw/intern/draw_manager_exec.c b/source/blender/draw/intern/draw_manager_exec.c index 808b5cc675b..21bcaa2d2b2 100644 --- a/source/blender/draw/intern/draw_manager_exec.c +++ b/source/blender/draw/intern/draw_manager_exec.c @@ -57,12 +57,15 @@ typedef struct DRWCommandsState { /* Resource location. */ int obmats_loc; int obinfos_loc; + int obattrs_loc; int baseinst_loc; int chunkid_loc; int resourceid_loc; /* Legacy matrix support. */ int obmat_loc; int obinv_loc; + /* Uniform Attributes. */ + DRWSparseUniformBuf *obattrs_ubo; /* Selection ID state. */ GPUVertBuf *select_buf; uint select_id; @@ -648,6 +651,12 @@ static void draw_update_uniforms(DRWShadingGroup *shgroup, state->obinfos_loc = uni->location; GPU_uniformbuf_bind(DST.vmempool->obinfos_ubo[0], uni->location); break; + case DRW_UNIFORM_BLOCK_OBATTRS: + state->obattrs_loc = uni->location; + state->obattrs_ubo = DRW_uniform_attrs_pool_find_ubo(DST.vmempool->obattrs_ubo_pool, + uni->uniform_attrs); + DRW_sparse_uniform_buffer_bind(state->obattrs_ubo, 0, uni->location); + break; case DRW_UNIFORM_RESOURCE_CHUNK: state->chunkid_loc = uni->location; GPU_shader_uniform_int(shgroup->shader, uni->location, 0); @@ -762,6 +771,10 @@ static void draw_call_resource_bind(DRWCommandsState *state, const DRWResourceHa GPU_uniformbuf_unbind(DST.vmempool->obinfos_ubo[state->resource_chunk]); GPU_uniformbuf_bind(DST.vmempool->obinfos_ubo[chunk], state->obinfos_loc); } + if (state->obattrs_loc != -1) { + DRW_sparse_uniform_buffer_unbind(state->obattrs_ubo, state->resource_chunk); + DRW_sparse_uniform_buffer_bind(state->obattrs_ubo, chunk, state->obattrs_loc); + } state->resource_chunk = chunk; } @@ -884,6 +897,9 @@ static void draw_call_batching_finish(DRWShadingGroup *shgroup, DRWCommandsState if (state->obinfos_loc != -1) { GPU_uniformbuf_unbind(DST.vmempool->obinfos_ubo[state->resource_chunk]); } + if (state->obattrs_loc != -1) { + DRW_sparse_uniform_buffer_unbind(state->obattrs_ubo, state->resource_chunk); + } } static void draw_shgroup(DRWShadingGroup *shgroup, DRWState pass_state) @@ -893,11 +909,13 @@ static void draw_shgroup(DRWShadingGroup *shgroup, DRWState pass_state) DRWCommandsState state = { .obmats_loc = -1, .obinfos_loc = -1, + .obattrs_loc = -1, .baseinst_loc = -1, .chunkid_loc = -1, .resourceid_loc = -1, .obmat_loc = -1, .obinv_loc = -1, + .obattrs_ubo = NULL, .drw_state_enabled = 0, .drw_state_disabled = 0, }; diff --git a/source/blender/editors/space_node/drawnode.c b/source/blender/editors/space_node/drawnode.c index 66da2887427..16bbca2a0c3 100644 --- a/source/blender/editors/space_node/drawnode.c +++ b/source/blender/editors/space_node/drawnode.c @@ -748,6 +748,7 @@ static void node_shader_buts_vect_transform(uiLayout *layout, bContext *UNUSED(C static void node_shader_buts_attribute(uiLayout *layout, bContext *UNUSED(C), PointerRNA *ptr) { + uiItemR(layout, ptr, "attribute_type", DEFAULT_FLAGS, IFACE_("Type"), ICON_NONE); uiItemR(layout, ptr, "attribute_name", DEFAULT_FLAGS, IFACE_("Name"), ICON_NONE); } diff --git a/source/blender/gpu/GPU_material.h b/source/blender/gpu/GPU_material.h index 2f12625acac..67cd1a61aed 100644 --- a/source/blender/gpu/GPU_material.h +++ b/source/blender/gpu/GPU_material.h @@ -34,6 +34,7 @@ extern "C" { #endif +struct GHash; struct GPUMaterial; struct GPUNode; struct GPUNodeLink; @@ -143,6 +144,7 @@ typedef void (*GPUMaterialEvalCallbackFn)(GPUMaterial *mat, GPUNodeLink *GPU_constant(const float *num); GPUNodeLink *GPU_uniform(const float *num); GPUNodeLink *GPU_attribute(GPUMaterial *mat, CustomDataType type, const char *name); +GPUNodeLink *GPU_uniform_attribute(GPUMaterial *mat, const char *name, bool use_dupli); GPUNodeLink *GPU_image(GPUMaterial *mat, struct Image *ima, struct ImageUser *iuser, @@ -259,6 +261,31 @@ ListBase GPU_material_attributes(GPUMaterial *material); ListBase GPU_material_textures(GPUMaterial *material); ListBase GPU_material_volume_grids(GPUMaterial *material); +typedef struct GPUUniformAttr { + struct GPUUniformAttr *next, *prev; + + /* Meaningful part of the attribute set key. */ + char name[64]; /* MAX_CUSTOMDATA_LAYER_NAME */ + bool use_dupli; + + /* Helper fields used by code generation. */ + short id; + int users; +} GPUUniformAttr; + +typedef struct GPUUniformAttrList { + ListBase list; /* GPUUniformAttr */ + + /* List length and hash code precomputed for fast lookup and comparison. */ + unsigned int count, hash_code; +} GPUUniformAttrList; + +GPUUniformAttrList *GPU_material_uniform_attributes(GPUMaterial *material); + +struct GHash *GPU_uniform_attr_list_hash_new(const char *info); +void GPU_uniform_attr_list_copy(GPUUniformAttrList *dest, GPUUniformAttrList *src); +void GPU_uniform_attr_list_free(GPUUniformAttrList *set); + #ifdef __cplusplus } #endif diff --git a/source/blender/gpu/GPU_shader.h b/source/blender/gpu/GPU_shader.h index 9aaa5d4cae8..27a7ea1e6a5 100644 --- a/source/blender/gpu/GPU_shader.h +++ b/source/blender/gpu/GPU_shader.h @@ -413,6 +413,9 @@ void GPU_shader_free_builtin_shaders(void); * This makes sure the GPUVertexFormat name buffer does not overflow. */ #define GPU_MAX_ATTR 15 +/* Determined by the maximum uniform buffer size divided by chunk size. */ +#define GPU_MAX_UNIFORM_ATTR 8 + #ifdef __cplusplus } #endif diff --git a/source/blender/gpu/GPU_uniform_buffer.h b/source/blender/gpu/GPU_uniform_buffer.h index ebcaa80e6f6..4efac0a8c00 100644 --- a/source/blender/gpu/GPU_uniform_buffer.h +++ b/source/blender/gpu/GPU_uniform_buffer.h @@ -53,6 +53,7 @@ void GPU_uniformbuf_unbind(GPUUniformBuf *ubo); void GPU_uniformbuf_unbind_all(void); #define GPU_UBO_BLOCK_NAME "nodeTree" +#define GPU_ATTRIBUTE_UBO_BLOCK_NAME "uniformAttrs" #ifdef __cplusplus } diff --git a/source/blender/gpu/GPU_viewport.h b/source/blender/gpu/GPU_viewport.h index 7b0d8c274d3..d8e4c5377b0 100644 --- a/source/blender/gpu/GPU_viewport.h +++ b/source/blender/gpu/GPU_viewport.h @@ -38,6 +38,7 @@ extern "C" { #define GPU_INFO_SIZE 512 /* IMA_MAX_RENDER_TEXT */ #define GLA_PIXEL_OFS 0.375f +typedef struct GHash GHash; typedef struct GPUViewport GPUViewport; struct GPUFrameBuffer; @@ -57,6 +58,7 @@ typedef struct ViewportMemoryPool { struct BLI_memblock *images; struct GPUUniformBuf **matrices_ubo; struct GPUUniformBuf **obinfos_ubo; + struct GHash *obattrs_ubo_pool; uint ubo_len; } ViewportMemoryPool; diff --git a/source/blender/gpu/intern/gpu_codegen.c b/source/blender/gpu/intern/gpu_codegen.c index 39c3119cc39..c16fc08d1aa 100644 --- a/source/blender/gpu/intern/gpu_codegen.c +++ b/source/blender/gpu/intern/gpu_codegen.c @@ -377,6 +377,19 @@ static int codegen_process_uniforms_functions(GPUMaterial *material, BLI_freelistN(&ubo_inputs); } + /* Generate the uniform attribute UBO if necessary. */ + if (!BLI_listbase_is_empty(&graph->uniform_attrs.list)) { + BLI_dynstr_append(ds, "\nstruct UniformAttributes {\n"); + LISTBASE_FOREACH (GPUUniformAttr *, attr, &graph->uniform_attrs.list) { + BLI_dynstr_appendf(ds, " vec4 attr%d;\n", attr->id); + } + BLI_dynstr_append(ds, "};\n"); + BLI_dynstr_appendf(ds, "layout (std140) uniform %s {\n", GPU_ATTRIBUTE_UBO_BLOCK_NAME); + BLI_dynstr_append(ds, " UniformAttributes uniform_attrs[DRW_RESOURCE_CHUNK_LEN];\n"); + BLI_dynstr_append(ds, "};\n"); + BLI_dynstr_append(ds, "#define GET_UNIFORM_ATTR(name) (uniform_attrs[resource_id].name)\n"); + } + BLI_dynstr_append(ds, "\n"); return builtins; @@ -480,6 +493,9 @@ static void codegen_call_functions(DynStr *ds, GPUNodeGraph *graph, GPUOutput *f else if (input->source == GPU_SOURCE_ATTR) { codegen_convert_datatype(ds, input->attr->gputype, input->type, "var", input->attr->id); } + else if (input->source == GPU_SOURCE_UNIFORM_ATTR) { + BLI_dynstr_appendf(ds, "GET_UNIFORM_ATTR(attr%d)", input->uniform_attr->id); + } BLI_dynstr_append(ds, ", "); } @@ -799,6 +815,7 @@ GPUPass *GPU_generate_pass(GPUMaterial *material, /* Prune the unused nodes and extract attributes before compiling so the * generated VBOs are ready to accept the future shader. */ gpu_node_graph_prune_unused(graph); + gpu_node_graph_finalize_uniform_attrs(graph); int builtins = 0; LISTBASE_FOREACH (GPUNode *, node, &graph->nodes) { diff --git a/source/blender/gpu/intern/gpu_material.c b/source/blender/gpu/intern/gpu_material.c index 011d14673b4..e676c0b33cd 100644 --- a/source/blender/gpu/intern/gpu_material.c +++ b/source/blender/gpu/intern/gpu_material.c @@ -584,6 +584,12 @@ ListBase GPU_material_volume_grids(GPUMaterial *material) return material->graph.volume_grids; } +GPUUniformAttrList *GPU_material_uniform_attributes(GPUMaterial *material) +{ + GPUUniformAttrList *attrs = &material->graph.uniform_attrs; + return attrs->count > 0 ? attrs : NULL; +} + void GPU_material_output_link(GPUMaterial *material, GPUNodeLink *link) { if (!material->graph.outlink) { diff --git a/source/blender/gpu/intern/gpu_node_graph.c b/source/blender/gpu/intern/gpu_node_graph.c index fdf8ba172cb..2a2a51e32b3 100644 --- a/source/blender/gpu/intern/gpu_node_graph.c +++ b/source/blender/gpu/intern/gpu_node_graph.c @@ -137,6 +137,10 @@ static void gpu_node_input_link(GPUNode *node, GPUNodeLink *link, const eGPUType * can happen if mixing the generic Attribute node with specialized ones. */ CLAMP_MIN(input->attr->gputype, type); break; + case GPU_NODE_LINK_UNIFORM_ATTR: + input->source = GPU_SOURCE_UNIFORM_ATTR; + input->uniform_attr = link->uniform_attr; + break; case GPU_NODE_LINK_CONSTANT: input->source = (type == GPU_CLOSURE) ? GPU_SOURCE_STRUCT : GPU_SOURCE_CONSTANT; break; @@ -262,8 +266,90 @@ static void gpu_node_output(GPUNode *node, const eGPUType type, GPUNodeLink **li BLI_addtail(&node->outputs, output); } +/* Uniform Attribute Functions */ + +static int uniform_attr_sort_cmp(const void *a, const void *b) +{ + const GPUUniformAttr *attr_a = a, *attr_b = b; + + int cmps = strcmp(attr_a->name, attr_b->name); + if (cmps != 0) { + return cmps > 0 ? 1 : 0; + } + + return (attr_a->use_dupli && !attr_b->use_dupli); +} + +static unsigned int uniform_attr_list_hash(const void *key) +{ + const GPUUniformAttrList *attrs = key; + return attrs->hash_code; +} + +static bool uniform_attr_list_cmp(const void *a, const void *b) +{ + const GPUUniformAttrList *set_a = a, *set_b = b; + + if (set_a->hash_code != set_b->hash_code || set_a->count != set_b->count) { + return true; + } + + GPUUniformAttr *attr_a = set_a->list.first, *attr_b = set_b->list.first; + + for (; attr_a && attr_b; attr_a = attr_a->next, attr_b = attr_b->next) { + if (!STREQ(attr_a->name, attr_b->name) || attr_a->use_dupli != attr_b->use_dupli) { + return true; + } + } + + return attr_a || attr_b; +} + +struct GHash *GPU_uniform_attr_list_hash_new(const char *info) +{ + return BLI_ghash_new(uniform_attr_list_hash, uniform_attr_list_cmp, info); +} + +void GPU_uniform_attr_list_copy(GPUUniformAttrList *dest, GPUUniformAttrList *src) +{ + dest->count = src->count; + dest->hash_code = src->hash_code; + BLI_duplicatelist(&dest->list, &src->list); +} + +void GPU_uniform_attr_list_free(GPUUniformAttrList *set) +{ + set->count = 0; + set->hash_code = 0; + BLI_freelistN(&set->list); +} + +void gpu_node_graph_finalize_uniform_attrs(GPUNodeGraph *graph) +{ + GPUUniformAttrList *attrs = &graph->uniform_attrs; + BLI_assert(attrs->count == BLI_listbase_count(&attrs->list)); + + /* Sort the attributes by name to ensure a stable order. */ + BLI_listbase_sort(&attrs->list, uniform_attr_sort_cmp); + + /* Compute the indices and the hash code. */ + int next_id = 0; + attrs->hash_code = 0; + + LISTBASE_FOREACH (GPUUniformAttr *, attr, &attrs->list) { + attr->id = next_id++; + + attrs->hash_code ^= BLI_ghashutil_strhash_p(attr->name); + + if (attr->use_dupli) { + attrs->hash_code ^= BLI_ghashutil_uinthash(attr->id); + } + } +} + /* Attributes and Textures */ +/** Add a new varying attribute of given type and name. Returns NULL if out of slots. */ static GPUMaterialAttribute *gpu_node_graph_add_attribute(GPUNodeGraph *graph, CustomDataType type, const char *name) @@ -299,6 +385,38 @@ static GPUMaterialAttribute *gpu_node_graph_add_attribute(GPUNodeGraph *graph, return attr; } +/** Add a new uniform attribute of given type and name. Returns NULL if out of slots. */ +static GPUUniformAttr *gpu_node_graph_add_uniform_attribute(GPUNodeGraph *graph, + const char *name, + bool use_dupli) +{ + /* Find existing attribute. */ + GPUUniformAttrList *attrs = &graph->uniform_attrs; + GPUUniformAttr *attr = attrs->list.first; + + for (; attr; attr = attr->next) { + if (STREQ(attr->name, name) && attr->use_dupli == use_dupli) { + break; + } + } + + /* Add new requested attribute if it's within GPU limits. */ + if (attr == NULL && attrs->count < GPU_MAX_UNIFORM_ATTR) { + attr = MEM_callocN(sizeof(*attr), __func__); + STRNCPY(attr->name, name); + attr->use_dupli = use_dupli; + attr->id = -1; + BLI_addtail(&attrs->list, attr); + attrs->count++; + } + + if (attr != NULL) { + attr->users++; + } + + return attr; +} + static GPUMaterialTexture *gpu_node_graph_add_texture(GPUNodeGraph *graph, Image *ima, ImageUser *iuser, @@ -372,6 +490,7 @@ GPUNodeLink *GPU_attribute(GPUMaterial *mat, const CustomDataType type, const ch GPUNodeGraph *graph = gpu_material_node_graph(mat); GPUMaterialAttribute *attr = gpu_node_graph_add_attribute(graph, type, name); + /* Dummy fallback if out of slots. */ if (attr == NULL) { static const float zero_data[GPU_MAX_CONSTANT_DATA] = {0.0f}; return GPU_constant(zero_data); @@ -383,6 +502,23 @@ GPUNodeLink *GPU_attribute(GPUMaterial *mat, const CustomDataType type, const ch return link; } +GPUNodeLink *GPU_uniform_attribute(GPUMaterial *mat, const char *name, bool use_dupli) +{ + GPUNodeGraph *graph = gpu_material_node_graph(mat); + GPUUniformAttr *attr = gpu_node_graph_add_uniform_attribute(graph, name, use_dupli); + + /* Dummy fallback if out of slots. */ + if (attr == NULL) { + static const float zero_data[GPU_MAX_CONSTANT_DATA] = {0.0f}; + return GPU_constant(zero_data); + } + + GPUNodeLink *link = gpu_node_link_create(); + link->link_type = GPU_NODE_LINK_UNIFORM_ATTR; + link->uniform_attr = attr; + return link; +} + GPUNodeLink *GPU_constant(const float *num) { GPUNodeLink *link = gpu_node_link_create(); @@ -619,6 +755,9 @@ static void gpu_inputs_free(ListBase *inputs) if (input->source == GPU_SOURCE_ATTR) { input->attr->users--; } + else if (input->source == GPU_SOURCE_UNIFORM_ATTR) { + input->uniform_attr->users--; + } else if (ELEM(input->source, GPU_SOURCE_TEX, GPU_SOURCE_TEX_TILED_MAPPING)) { input->texture->users--; } @@ -674,6 +813,7 @@ void gpu_node_graph_free(GPUNodeGraph *graph) BLI_freelistN(&graph->volume_grids); BLI_freelistN(&graph->textures); BLI_freelistN(&graph->attributes); + GPU_uniform_attr_list_free(&graph->uniform_attrs); } /* Prune Unused Nodes */ @@ -738,4 +878,13 @@ void gpu_node_graph_prune_unused(GPUNodeGraph *graph) BLI_freelinkN(&graph->volume_grids, grid); } } + + GPUUniformAttrList *uattrs = &graph->uniform_attrs; + + LISTBASE_FOREACH_MUTABLE (GPUUniformAttr *, attr, &uattrs->list) { + if (attr->users == 0) { + BLI_freelinkN(&uattrs->list, attr); + uattrs->count--; + } + } } diff --git a/source/blender/gpu/intern/gpu_node_graph.h b/source/blender/gpu/intern/gpu_node_graph.h index 7265abf4d65..a0e6298cd92 100644 --- a/source/blender/gpu/intern/gpu_node_graph.h +++ b/source/blender/gpu/intern/gpu_node_graph.h @@ -42,6 +42,7 @@ typedef enum eGPUDataSource { GPU_SOURCE_CONSTANT, GPU_SOURCE_UNIFORM, GPU_SOURCE_ATTR, + GPU_SOURCE_UNIFORM_ATTR, GPU_SOURCE_BUILTIN, GPU_SOURCE_STRUCT, GPU_SOURCE_TEX, @@ -53,6 +54,7 @@ typedef enum eGPUDataSource { typedef enum { GPU_NODE_LINK_NONE = 0, GPU_NODE_LINK_ATTR, + GPU_NODE_LINK_UNIFORM_ATTR, GPU_NODE_LINK_BUILTIN, GPU_NODE_LINK_COLORBAND, GPU_NODE_LINK_CONSTANT, @@ -96,6 +98,8 @@ struct GPUNodeLink { struct GPUOutput *output; /* GPU_NODE_LINK_ATTR */ struct GPUMaterialAttribute *attr; + /* GPU_NODE_LINK_UNIFORM_ATTR */ + struct GPUUniformAttr *uniform_attr; /* GPU_NODE_LINK_IMAGE_BLENDER */ struct GPUMaterialTexture *texture; }; @@ -130,6 +134,8 @@ typedef struct GPUInput { struct GPUMaterialTexture *texture; /* GPU_SOURCE_ATTR */ struct GPUMaterialAttribute *attr; + /* GPU_SOURCE_UNIFORM_ATTR */ + struct GPUUniformAttr *uniform_attr; /* GPU_SOURCE_VOLUME_GRID | GPU_SOURCE_VOLUME_GRID_TRANSFORM */ struct GPUMaterialVolumeGrid *volume_grid; }; @@ -146,11 +152,15 @@ typedef struct GPUNodeGraph { ListBase attributes; ListBase textures; ListBase volume_grids; + + /* The list of uniform attributes. */ + GPUUniformAttrList uniform_attrs; } GPUNodeGraph; /* Node Graph */ void gpu_node_graph_prune_unused(GPUNodeGraph *graph); +void gpu_node_graph_finalize_uniform_attrs(GPUNodeGraph *graph); void gpu_node_graph_free_nodes(GPUNodeGraph *graph); void gpu_node_graph_free(GPUNodeGraph *graph); diff --git a/source/blender/gpu/intern/gpu_viewport.c b/source/blender/gpu/intern/gpu_viewport.c index 9063c8bdbce..188c8786665 100644 --- a/source/blender/gpu/intern/gpu_viewport.c +++ b/source/blender/gpu/intern/gpu_viewport.c @@ -1023,6 +1023,9 @@ void GPU_viewport_free(GPUViewport *viewport) } BLI_memblock_destroy(viewport->vmempool.images, NULL); } + if (viewport->vmempool.obattrs_ubo_pool != NULL) { + DRW_uniform_attrs_pool_free(viewport->vmempool.obattrs_ubo_pool); + } for (int i = 0; i < viewport->vmempool.ubo_len; i++) { GPU_uniformbuf_free(viewport->vmempool.matrices_ubo[i]); diff --git a/source/blender/makesdna/DNA_node_types.h b/source/blender/makesdna/DNA_node_types.h index cba93bbadfc..6f6c9029d07 100644 --- a/source/blender/makesdna/DNA_node_types.h +++ b/source/blender/makesdna/DNA_node_types.h @@ -923,6 +923,8 @@ typedef struct NodeTexMagic { typedef struct NodeShaderAttribute { char name[64]; + int type; + char _pad[4]; } NodeShaderAttribute; typedef struct NodeShaderVectTransform { @@ -1092,6 +1094,13 @@ typedef struct NodeDenoise { #define SHD_VECT_TRANSFORM_SPACE_OBJECT 1 #define SHD_VECT_TRANSFORM_SPACE_CAMERA 2 +/* attribute */ +enum { + SHD_ATTRIBUTE_GEOMETRY = 0, + SHD_ATTRIBUTE_OBJECT = 1, + SHD_ATTRIBUTE_INSTANCER = 2, +}; + /* toon modes */ #define SHD_TOON_DIFFUSE 0 #define SHD_TOON_GLOSSY 1 diff --git a/source/blender/makesrna/intern/rna_nodetree.c b/source/blender/makesrna/intern/rna_nodetree.c index 78d02795325..bc13601879f 100644 --- a/source/blender/makesrna/intern/rna_nodetree.c +++ b/source/blender/makesrna/intern/rna_nodetree.c @@ -4362,10 +4362,38 @@ static void def_sh_vector_rotate(StructRNA *srna) static void def_sh_attribute(StructRNA *srna) { + static const EnumPropertyItem prop_attribute_type[] = { + {SHD_ATTRIBUTE_GEOMETRY, + "GEOMETRY", + 0, + "Geometry", + "The attribute is associated with the object geometry, and its value " + "varies from vertex to vertex, or within the object volume"}, + {SHD_ATTRIBUTE_OBJECT, + "OBJECT", + 0, + "Object", + "The attribute is associated with the object or mesh datablock itself, " + "and its value is uniform"}, + {SHD_ATTRIBUTE_INSTANCER, + "INSTANCER", + 0, + "Instancer", + "The attribute is associated with the instancer particle system or object, " + "falling back to the Object mode if the attribute isn't found, or the object " + "is not instanced"}, + {0, NULL, 0, NULL, NULL}, + }; PropertyRNA *prop; RNA_def_struct_sdna_from(srna, "NodeShaderAttribute", "storage"); + prop = RNA_def_property(srna, "attribute_type", PROP_ENUM, PROP_NONE); + RNA_def_property_enum_sdna(prop, NULL, "type"); + RNA_def_property_enum_items(prop, prop_attribute_type); + RNA_def_property_ui_text(prop, "Attribute Type", "General type of the attribute"); + RNA_def_property_update(prop, NC_NODE | NA_EDITED, "rna_Node_update"); + prop = RNA_def_property(srna, "attribute_name", PROP_STRING, PROP_NONE); RNA_def_property_string_sdna(prop, NULL, "name"); RNA_def_property_ui_text(prop, "Attribute Name", ""); diff --git a/source/blender/nodes/shader/nodes/node_shader_attribute.c b/source/blender/nodes/shader/nodes/node_shader_attribute.c index 37472aeec2e..9b3122e38e0 100644 --- a/source/blender/nodes/shader/nodes/node_shader_attribute.c +++ b/source/blender/nodes/shader/nodes/node_shader_attribute.c @@ -42,8 +42,9 @@ static int node_shader_gpu_attribute(GPUMaterial *mat, GPUNodeStack *out) { NodeShaderAttribute *attr = node->storage; + bool is_varying = attr->type == SHD_ATTRIBUTE_GEOMETRY; - if (GPU_material_is_volume_shader(mat)) { + if (GPU_material_is_volume_shader(mat) && is_varying) { if (out[0].hasoutput) { out[0].link = GPU_volume_grid(mat, attr->name, GPU_VOLUME_DEFAULT_0); } @@ -61,7 +62,15 @@ static int node_shader_gpu_attribute(GPUMaterial *mat, return 1; } - GPUNodeLink *cd_attr = GPU_attribute(mat, CD_AUTO_FROM_NAME, attr->name); + GPUNodeLink *cd_attr; + + if (is_varying) { + cd_attr = GPU_attribute(mat, CD_AUTO_FROM_NAME, attr->name); + } + else { + cd_attr = GPU_uniform_attribute(mat, attr->name, attr->type == SHD_ATTRIBUTE_INSTANCER); + } + GPU_stack_link(mat, node, "node_attribute", in, out, cd_attr); /* for each output. */ |