This repository has been archived on 2023-10-09. You can view files and clone it, but cannot push or open issues or pull requests.
Files
blender-archive/source/blender/modifiers/intern/MOD_nodes.cc

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1143 lines
42 KiB
C++
Raw Normal View History

Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
/*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software Foundation,
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*
* The Original Code is Copyright (C) 2005 by the Blender Foundation.
* All rights reserved.
*/
/** \file
* \ingroup modifiers
*/
#include <cstring>
#include <iostream>
#include <string>
#include "MEM_guardedalloc.h"
#include "BLI_float3.hh"
#include "BLI_listbase.h"
#include "BLI_set.hh"
#include "BLI_string.h"
#include "BLI_utildefines.h"
#include "DNA_collection_types.h"
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
#include "DNA_defaults.h"
#include "DNA_mesh_types.h"
#include "DNA_meshdata_types.h"
#include "DNA_modifier_types.h"
#include "DNA_node_types.h"
#include "DNA_object_types.h"
#include "DNA_pointcloud_types.h"
#include "DNA_scene_types.h"
#include "DNA_screen_types.h"
#include "BKE_customdata.h"
#include "BKE_idprop.h"
#include "BKE_lib_query.h"
#include "BKE_mesh.h"
#include "BKE_modifier.h"
#include "BKE_pointcloud.h"
#include "BKE_screen.h"
#include "BKE_simulation.h"
#include "BLO_read_write.h"
#include "UI_interface.h"
#include "UI_resources.h"
#include "RNA_access.h"
#include "RNA_enum_types.h"
#include "DEG_depsgraph_build.h"
#include "DEG_depsgraph_query.h"
#include "MOD_modifiertypes.h"
#include "MOD_nodes.h"
#include "MOD_ui_common.h"
#include "NOD_derived_node_tree.hh"
#include "NOD_geometry.h"
#include "NOD_geometry_exec.hh"
#include "NOD_node_tree_multi_function.hh"
#include "NOD_type_callbacks.hh"
using blender::float3;
using blender::IndexRange;
using blender::Map;
using blender::Set;
using blender::Span;
using blender::StringRef;
using blender::Vector;
using blender::fn::GMutablePointer;
using blender::fn::GValueMap;
using blender::nodes::GeoNodeExecParams;
using namespace blender::nodes::derived_node_tree_types;
using namespace blender::fn::multi_function_types;
static void initData(ModifierData *md)
{
NodesModifierData *nmd = (NodesModifierData *)md;
BLI_assert(MEMCMP_STRUCT_AFTER_IS_ZERO(nmd, modifier));
MEMCPY_STRUCT_AFTER(nmd, DNA_struct_default_get(NodesModifierData), modifier);
}
static void addIdsUsedBySocket(const ListBase *sockets, Set<ID *> &ids)
{
LISTBASE_FOREACH (const bNodeSocket *, socket, sockets) {
if (socket->type == SOCK_OBJECT) {
Object *object = ((bNodeSocketValueObject *)socket->default_value)->value;
if (object != nullptr) {
ids.add(&object->id);
}
}
else if (socket->type == SOCK_COLLECTION) {
Collection *collection = ((bNodeSocketValueCollection *)socket->default_value)->value;
if (collection != nullptr) {
ids.add(&collection->id);
}
}
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
}
}
static void findUsedIds(const bNodeTree &tree, Set<ID *> &ids)
{
Set<const bNodeTree *> handled_groups;
LISTBASE_FOREACH (const bNode *, node, &tree.nodes) {
addIdsUsedBySocket(&node->inputs, ids);
addIdsUsedBySocket(&node->outputs, ids);
if (node->type == NODE_GROUP) {
const bNodeTree *group = (bNodeTree *)node->id;
if (group != nullptr && handled_groups.add(group)) {
findUsedIds(*group, ids);
}
}
}
}
static void updateDepsgraph(ModifierData *md, const ModifierUpdateDepsgraphContext *ctx)
{
NodesModifierData *nmd = reinterpret_cast<NodesModifierData *>(md);
if (nmd->node_group != nullptr) {
DEG_add_node_tree_relation(ctx->node, nmd->node_group, "Nodes Modifier");
Set<ID *> used_ids;
findUsedIds(*nmd->node_group, used_ids);
for (ID *id : used_ids) {
if (GS(id->name) == ID_OB) {
Object *object = reinterpret_cast<Object *>(id);
DEG_add_object_relation(ctx->node, object, DEG_OB_COMP_TRANSFORM, "Nodes Modifier");
if (id != &ctx->object->id) {
if (object->type != OB_EMPTY) {
DEG_add_object_relation(
ctx->node, (Object *)id, DEG_OB_COMP_GEOMETRY, "Nodes Modifier");
}
}
}
}
}
/* TODO: Add relations for IDs in settings. */
/* TODO: Add dependency for collection changes. */
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
}
static void foreachIDLink(ModifierData *md, Object *ob, IDWalkFunc walk, void *userData)
{
NodesModifierData *nmd = reinterpret_cast<NodesModifierData *>(md);
walk(userData, ob, (ID **)&nmd->node_group, IDWALK_CB_USER);
struct ForeachSettingData {
IDWalkFunc walk;
void *userData;
Object *ob;
} settings = {walk, userData, ob};
IDP_foreach_property(
nmd->settings.properties,
IDP_TYPE_FILTER_ID,
[](IDProperty *id_prop, void *user_data) {
ForeachSettingData *settings = (ForeachSettingData *)user_data;
settings->walk(
settings->userData, settings->ob, (ID **)&id_prop->data.pointer, IDWALK_CB_USER);
},
&settings);
}
static bool isDisabled(const struct Scene *UNUSED(scene),
ModifierData *md,
bool UNUSED(useRenderParams))
{
NodesModifierData *nmd = reinterpret_cast<NodesModifierData *>(md);
if (nmd->node_group == nullptr) {
return true;
}
return false;
}
class GeometryNodesEvaluator {
private:
blender::LinearAllocator<> allocator_;
Map<const DInputSocket *, GMutablePointer> value_by_input_;
Vector<const DInputSocket *> group_outputs_;
blender::nodes::MultiFunctionByNode &mf_by_node_;
const blender::nodes::DataTypeConversions &conversions_;
const blender::bke::PersistentDataHandleMap &handle_map_;
const Object *self_object_;
public:
GeometryNodesEvaluator(const Map<const DOutputSocket *, GMutablePointer> &group_input_data,
Vector<const DInputSocket *> group_outputs,
blender::nodes::MultiFunctionByNode &mf_by_node,
const blender::bke::PersistentDataHandleMap &handle_map,
const Object *self_object)
: group_outputs_(std::move(group_outputs)),
mf_by_node_(mf_by_node),
conversions_(blender::nodes::get_implicit_type_conversions()),
handle_map_(handle_map),
self_object_(self_object)
{
for (auto item : group_input_data.items()) {
this->forward_to_inputs(*item.key, item.value);
}
}
Vector<GMutablePointer> execute()
{
Vector<GMutablePointer> results;
for (const DInputSocket *group_output : group_outputs_) {
GMutablePointer result = this->get_input_value(*group_output);
results.append(result);
}
for (GMutablePointer value : value_by_input_.values()) {
value.destruct();
}
return results;
}
private:
GMutablePointer get_input_value(const DInputSocket &socket_to_compute)
{
std::optional<GMutablePointer> value = value_by_input_.pop_try(&socket_to_compute);
if (value.has_value()) {
/* This input has been computed before, return it directly. */
return *value;
}
Span<const DOutputSocket *> from_sockets = socket_to_compute.linked_sockets();
Span<const DGroupInput *> from_group_inputs = socket_to_compute.linked_group_inputs();
const int total_inputs = from_sockets.size() + from_group_inputs.size();
BLI_assert(total_inputs <= 1);
if (total_inputs == 0) {
/* The input is not connected, use the value from the socket itself. */
return get_unlinked_input_value(socket_to_compute);
}
if (from_group_inputs.size() == 1) {
/* The input gets its value from the input of a group that is not further connected. */
return get_unlinked_input_value(socket_to_compute);
}
/* Compute the socket now. */
const DOutputSocket &from_socket = *from_sockets[0];
this->compute_output_and_forward(from_socket);
return value_by_input_.pop(&socket_to_compute);
}
void compute_output_and_forward(const DOutputSocket &socket_to_compute)
{
const DNode &node = socket_to_compute.node();
const bNode &bnode = *node.bnode();
if (!socket_to_compute.is_available()) {
/* If the output is not available, use a default value. */
const CPPType &type = *blender::nodes::socket_cpp_type_get(*socket_to_compute.typeinfo());
void *buffer = allocator_.allocate(type.size(), type.alignment());
type.copy_to_uninitialized(type.default_value(), buffer);
this->forward_to_inputs(socket_to_compute, {type, buffer});
return;
}
/* Prepare inputs required to execute the node. */
GValueMap<StringRef> node_inputs_map{allocator_};
for (const DInputSocket *input_socket : node.inputs()) {
if (input_socket->is_available()) {
GMutablePointer value = this->get_input_value(*input_socket);
node_inputs_map.add_new_direct(input_socket->identifier(), value);
}
}
/* Execute the node. */
GValueMap<StringRef> node_outputs_map{allocator_};
GeoNodeExecParams params{bnode, node_inputs_map, node_outputs_map, handle_map_, self_object_};
this->execute_node(node, params);
/* Forward computed outputs to linked input sockets. */
for (const DOutputSocket *output_socket : node.outputs()) {
if (output_socket->is_available()) {
GMutablePointer value = node_outputs_map.extract(output_socket->identifier());
this->forward_to_inputs(*output_socket, value);
}
}
}
void execute_node(const DNode &node, GeoNodeExecParams params)
{
const bNode &bnode = params.node();
/* Use the geometry-node-execute callback if it exists. */
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
if (bnode.typeinfo->geometry_node_execute != nullptr) {
bnode.typeinfo->geometry_node_execute(params);
return;
}
/* Use the multi-function implementation if it exists. */
const MultiFunction *multi_function = mf_by_node_.lookup_default(&node, nullptr);
if (multi_function != nullptr) {
this->execute_multi_function_node(node, params, *multi_function);
return;
}
/* Just output default values if no implementation exists. */
this->execute_unknown_node(node, params);
}
void execute_multi_function_node(const DNode &node,
GeoNodeExecParams params,
const MultiFunction &fn)
{
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
MFContextBuilder fn_context;
MFParamsBuilder fn_params{fn, 1};
Vector<GMutablePointer> input_data;
for (const DInputSocket *dsocket : node.inputs()) {
if (dsocket->is_available()) {
GMutablePointer data = params.extract_input(dsocket->identifier());
fn_params.add_readonly_single_input(GSpan(*data.type(), data.get(), 1));
input_data.append(data);
}
}
Vector<GMutablePointer> output_data;
for (const DOutputSocket *dsocket : node.outputs()) {
if (dsocket->is_available()) {
const CPPType &type = *blender::nodes::socket_cpp_type_get(*dsocket->typeinfo());
void *buffer = allocator_.allocate(type.size(), type.alignment());
fn_params.add_uninitialized_single_output(GMutableSpan(type, buffer, 1));
output_data.append(GMutablePointer(type, buffer));
}
}
fn.call(IndexRange(1), fn_params, fn_context);
for (GMutablePointer value : input_data) {
value.destruct();
}
int output_index = 0;
for (const int i : node.outputs().index_range()) {
if (node.output(i).is_available()) {
GMutablePointer value = output_data[output_index];
params.set_output_by_move(node.output(i).identifier(), value);
value.destruct();
output_index++;
}
}
}
void execute_unknown_node(const DNode &node, GeoNodeExecParams params)
{
for (const DOutputSocket *socket : node.outputs()) {
if (socket->is_available()) {
const CPPType &type = *blender::nodes::socket_cpp_type_get(*socket->typeinfo());
params.set_output_by_copy(socket->identifier(), {type, type.default_value()});
}
}
}
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
void forward_to_inputs(const DOutputSocket &from_socket, GMutablePointer value_to_forward)
{
Span<const DInputSocket *> to_sockets_all = from_socket.linked_sockets();
const CPPType &from_type = *value_to_forward.type();
Vector<const DInputSocket *> to_sockets_same_type;
for (const DInputSocket *to_socket : to_sockets_all) {
const CPPType &to_type = *blender::nodes::socket_cpp_type_get(*to_socket->typeinfo());
if (from_type == to_type) {
to_sockets_same_type.append(to_socket);
}
else {
void *buffer = allocator_.allocate(to_type.size(), to_type.alignment());
if (conversions_.is_convertible(from_type, to_type)) {
conversions_.convert(from_type, to_type, value_to_forward.get(), buffer);
}
else {
to_type.copy_to_uninitialized(to_type.default_value(), buffer);
}
value_by_input_.add_new(to_socket, GMutablePointer{to_type, buffer});
}
}
if (to_sockets_same_type.size() == 0) {
/* This value is not further used, so destruct it. */
value_to_forward.destruct();
}
else if (to_sockets_same_type.size() == 1) {
/* This value is only used on one input socket, no need to copy it. */
const DInputSocket *to_socket = to_sockets_same_type[0];
value_by_input_.add_new(to_socket, value_to_forward);
}
else {
/* Multiple inputs use the value, make a copy for every input except for one. */
const DInputSocket *first_to_socket = to_sockets_same_type[0];
Span<const DInputSocket *> other_to_sockets = to_sockets_same_type.as_span().drop_front(1);
const CPPType &type = *value_to_forward.type();
value_by_input_.add_new(first_to_socket, value_to_forward);
for (const DInputSocket *to_socket : other_to_sockets) {
void *buffer = allocator_.allocate(type.size(), type.alignment());
type.copy_to_uninitialized(value_to_forward.get(), buffer);
value_by_input_.add_new(to_socket, GMutablePointer{type, buffer});
}
}
}
GMutablePointer get_unlinked_input_value(const DInputSocket &socket)
{
bNodeSocket *bsocket;
if (socket.linked_group_inputs().size() == 0) {
bsocket = socket.bsocket();
}
else {
bsocket = socket.linked_group_inputs()[0]->bsocket();
}
const CPPType &type = *blender::nodes::socket_cpp_type_get(*socket.typeinfo());
void *buffer = allocator_.allocate(type.size(), type.alignment());
if (bsocket->type == SOCK_OBJECT) {
Object *object = ((bNodeSocketValueObject *)bsocket->default_value)->value;
blender::bke::PersistentObjectHandle object_handle = handle_map_.lookup(object);
new (buffer) blender::bke::PersistentObjectHandle(object_handle);
}
else if (bsocket->type == SOCK_COLLECTION) {
Collection *collection = ((bNodeSocketValueCollection *)bsocket->default_value)->value;
blender::bke::PersistentCollectionHandle collection_handle = handle_map_.lookup(collection);
new (buffer) blender::bke::PersistentCollectionHandle(collection_handle);
}
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
else {
blender::nodes::socket_cpp_value_get(*bsocket, buffer);
}
return {type, buffer};
}
};
/**
* This code is responsible for creating the new property and also creating the group of
* properties in the prop_ui_container group for the UI info, the mapping for which is
* scattered about in RNA_access.c.
*
* TODO(Hans): Codify this with some sort of table or refactor IDProperty use in RNA_access.c.
*/
struct SocketPropertyType {
/* Create the actual propery used to store the data for the modifier. */
IDProperty *(*create_prop)(const bNodeSocket &socket, const char *name);
/* Reused to build the "soft_min" property too. */
IDProperty *(*create_min_ui_prop)(const bNodeSocket &socket, const char *name);
/* Reused to build the "soft_max" property too. */
IDProperty *(*create_max_ui_prop)(const bNodeSocket &socket, const char *name);
/* This uses the same values as #create_prop, but sometimes the type is different, so it can't
* be the same function. */
IDProperty *(*create_default_ui_prop)(const bNodeSocket &socket, const char *name);
PropertyType (*rna_subtype_get)(const bNodeSocket &socket);
bool (*is_correct_type)(const IDProperty &property);
void (*init_cpp_value)(const IDProperty &property, void *r_value);
};
static IDProperty *socket_add_property(IDProperty *settings_prop_group,
IDProperty *ui_container,
const SocketPropertyType &property_type,
const bNodeSocket &socket)
{
const char *new_prop_name = socket.identifier;
/* Add the property actually storing the data to the modifier's group. */
IDProperty *prop = property_type.create_prop(socket, new_prop_name);
IDP_AddToGroup(settings_prop_group, prop);
prop->flag |= IDP_FLAG_OVERRIDABLE_LIBRARY;
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
/* Make the group in the ui container group to hold the property's UI settings. */
IDProperty *prop_ui_group;
{
IDPropertyTemplate idprop = {0};
prop_ui_group = IDP_New(IDP_GROUP, &idprop, new_prop_name);
IDP_AddToGroup(ui_container, prop_ui_group);
}
/* Create the properties for the socket's UI settings. */
if (property_type.create_min_ui_prop != nullptr) {
IDP_AddToGroup(prop_ui_group, property_type.create_min_ui_prop(socket, "min"));
IDP_AddToGroup(prop_ui_group, property_type.create_min_ui_prop(socket, "soft_min"));
}
if (property_type.create_max_ui_prop != nullptr) {
IDP_AddToGroup(prop_ui_group, property_type.create_max_ui_prop(socket, "max"));
IDP_AddToGroup(prop_ui_group, property_type.create_max_ui_prop(socket, "soft_max"));
}
if (property_type.create_default_ui_prop != nullptr) {
IDP_AddToGroup(prop_ui_group, property_type.create_default_ui_prop(socket, "default"));
}
if (property_type.rna_subtype_get != nullptr) {
const char *subtype_identifier = nullptr;
RNA_enum_identifier(rna_enum_property_subtype_items,
property_type.rna_subtype_get(socket),
&subtype_identifier);
if (subtype_identifier != nullptr) {
IDPropertyTemplate idprop = {0};
idprop.string.str = subtype_identifier;
idprop.string.len = BLI_strnlen(subtype_identifier, MAX_NAME) + 1;
IDP_AddToGroup(prop_ui_group, IDP_New(IDP_STRING, &idprop, "subtype"));
}
}
return prop;
}
static const SocketPropertyType *get_socket_property_type(const bNodeSocket &bsocket)
{
switch (bsocket.type) {
case SOCK_FLOAT: {
static const SocketPropertyType float_type = {
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueFloat *value = (bNodeSocketValueFloat *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.f = value->value;
return IDP_New(IDP_FLOAT, &idprop, name);
},
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueFloat *value = (bNodeSocketValueFloat *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.d = value->min;
return IDP_New(IDP_DOUBLE, &idprop, name);
},
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueFloat *value = (bNodeSocketValueFloat *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.d = value->max;
return IDP_New(IDP_DOUBLE, &idprop, name);
},
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueFloat *value = (bNodeSocketValueFloat *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.d = value->value;
return IDP_New(IDP_DOUBLE, &idprop, name);
},
[](const bNodeSocket &socket) {
return (PropertyType)((bNodeSocketValueFloat *)socket.default_value)->subtype;
},
[](const IDProperty &property) { return property.type == IDP_FLOAT; },
[](const IDProperty &property, void *r_value) {
*(float *)r_value = IDP_Float(&property);
},
};
return &float_type;
}
case SOCK_INT: {
static const SocketPropertyType int_type = {
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueInt *value = (bNodeSocketValueInt *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.i = value->value;
return IDP_New(IDP_INT, &idprop, name);
},
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueInt *value = (bNodeSocketValueInt *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.i = value->min;
return IDP_New(IDP_INT, &idprop, name);
},
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueInt *value = (bNodeSocketValueInt *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.i = value->max;
return IDP_New(IDP_INT, &idprop, name);
},
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueInt *value = (bNodeSocketValueInt *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.i = value->value;
return IDP_New(IDP_INT, &idprop, name);
},
[](const bNodeSocket &socket) {
return (PropertyType)((bNodeSocketValueInt *)socket.default_value)->subtype;
},
[](const IDProperty &property) { return property.type == IDP_INT; },
[](const IDProperty &property, void *r_value) { *(int *)r_value = IDP_Int(&property); },
};
return &int_type;
}
case SOCK_VECTOR: {
static const SocketPropertyType vector_type = {
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueVector *value = (bNodeSocketValueVector *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.array.len = 3;
idprop.array.type = IDP_FLOAT;
IDProperty *property = IDP_New(IDP_ARRAY, &idprop, name);
copy_v3_v3((float *)IDP_Array(property), value->value);
return property;
},
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueVector *value = (bNodeSocketValueVector *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.d = value->min;
return IDP_New(IDP_DOUBLE, &idprop, name);
},
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueVector *value = (bNodeSocketValueVector *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.d = value->max;
return IDP_New(IDP_DOUBLE, &idprop, name);
},
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueVector *value = (bNodeSocketValueVector *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.array.len = 3;
idprop.array.type = IDP_FLOAT;
IDProperty *property = IDP_New(IDP_ARRAY, &idprop, name);
copy_v3_v3((float *)IDP_Array(property), value->value);
return property;
},
[](const bNodeSocket &socket) {
return (PropertyType)((bNodeSocketValueVector *)socket.default_value)->subtype;
},
[](const IDProperty &property) {
return property.type == IDP_ARRAY && property.subtype == IDP_FLOAT &&
property.len == 3;
},
[](const IDProperty &property, void *r_value) {
copy_v3_v3((float *)r_value, (const float *)IDP_Array(&property));
},
};
return &vector_type;
}
case SOCK_BOOLEAN: {
static const SocketPropertyType boolean_type = {
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueBoolean *value = (bNodeSocketValueBoolean *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.i = value->value != 0;
return IDP_New(IDP_INT, &idprop, name);
},
[](const bNodeSocket &UNUSED(socket), const char *name) {
IDPropertyTemplate idprop = {0};
idprop.i = 0;
return IDP_New(IDP_INT, &idprop, name);
},
[](const bNodeSocket &UNUSED(socket), const char *name) {
IDPropertyTemplate idprop = {0};
idprop.i = 1;
return IDP_New(IDP_INT, &idprop, name);
},
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueBoolean *value = (bNodeSocketValueBoolean *)socket.default_value;
IDPropertyTemplate idprop = {0};
idprop.i = value->value != 0;
return IDP_New(IDP_INT, &idprop, name);
},
nullptr,
[](const IDProperty &property) { return property.type == IDP_INT; },
[](const IDProperty &property, void *r_value) {
*(bool *)r_value = IDP_Int(&property) != 0;
},
};
return &boolean_type;
}
case SOCK_STRING: {
static const SocketPropertyType string_type = {
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueString *value = (bNodeSocketValueString *)socket.default_value;
return IDP_NewString(
value->value, name, BLI_strnlen(value->value, sizeof(value->value)) + 1);
},
nullptr,
nullptr,
[](const bNodeSocket &socket, const char *name) {
bNodeSocketValueString *value = (bNodeSocketValueString *)socket.default_value;
return IDP_NewString(
value->value, name, BLI_strnlen(value->value, sizeof(value->value)) + 1);
},
nullptr,
[](const IDProperty &property) { return property.type == IDP_STRING; },
[](const IDProperty &property, void *r_value) {
new (r_value) std::string(IDP_String(&property));
},
};
return &string_type;
}
default: {
return nullptr;
}
}
}
/**
* Rebuild the list of properties based on the sockets exposed as the modifier's node group
* inputs. If any properties correspond to the old properties by name and type, carry over
* the values.
*/
void MOD_nodes_update_interface(Object *object, NodesModifierData *nmd)
{
if (nmd->node_group == nullptr) {
return;
}
IDProperty *old_properties = nmd->settings.properties;
{
IDPropertyTemplate idprop = {0};
nmd->settings.properties = IDP_New(IDP_GROUP, &idprop, "Nodes Modifier Settings");
}
IDProperty *ui_container_group;
{
IDPropertyTemplate idprop = {0};
ui_container_group = IDP_New(IDP_GROUP, &idprop, "_RNA_UI");
IDP_AddToGroup(nmd->settings.properties, ui_container_group);
}
LISTBASE_FOREACH (bNodeSocket *, socket, &nmd->node_group->inputs) {
const SocketPropertyType *property_type = get_socket_property_type(*socket);
if (property_type == nullptr) {
continue;
}
IDProperty *new_prop = socket_add_property(
nmd->settings.properties, ui_container_group, *property_type, *socket);
if (old_properties != nullptr) {
IDProperty *old_prop = IDP_GetPropertyFromGroup(old_properties, socket->identifier);
if (old_prop != nullptr && property_type->is_correct_type(*old_prop)) {
IDP_CopyPropertyContent(new_prop, old_prop);
}
}
}
if (old_properties != nullptr) {
IDP_FreeProperty(old_properties);
}
DEG_id_tag_update(&object->id, ID_RECALC_GEOMETRY);
}
void MOD_nodes_init(Main *bmain, NodesModifierData *nmd)
{
bNodeTree *ntree = ntreeAddTree(bmain, "Geometry Nodes", ntreeType_Geometry->idname);
nmd->node_group = ntree;
ntreeAddSocketInterface(ntree, SOCK_IN, "NodeSocketGeometry", "Geometry");
ntreeAddSocketInterface(ntree, SOCK_OUT, "NodeSocketGeometry", "Geometry");
bNode *group_input_node = nodeAddStaticNode(nullptr, ntree, NODE_GROUP_INPUT);
bNode *group_output_node = nodeAddStaticNode(nullptr, ntree, NODE_GROUP_OUTPUT);
nodeSetSelected(group_input_node, false);
nodeSetSelected(group_output_node, false);
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
group_input_node->locx = -200 - group_input_node->width;
group_output_node->locx = 200;
group_output_node->flag |= NODE_DO_OUTPUT;
nodeAddLink(ntree,
group_output_node,
(bNodeSocket *)group_output_node->inputs.first,
group_input_node,
(bNodeSocket *)group_input_node->outputs.first);
ntreeUpdateTree(bmain, ntree);
}
static void initialize_group_input(NodesModifierData &nmd,
const bNodeSocket &socket,
const CPPType &cpp_type,
void *r_value)
{
const SocketPropertyType *property_type = get_socket_property_type(socket);
if (property_type == nullptr) {
cpp_type.copy_to_uninitialized(cpp_type.default_value(), r_value);
return;
}
if (nmd.settings.properties == nullptr) {
blender::nodes::socket_cpp_value_get(socket, r_value);
return;
}
const IDProperty *property = IDP_GetPropertyFromGroup(nmd.settings.properties,
socket.identifier);
if (property == nullptr) {
blender::nodes::socket_cpp_value_get(socket, r_value);
return;
}
if (!property_type->is_correct_type(*property)) {
blender::nodes::socket_cpp_value_get(socket, r_value);
}
property_type->init_cpp_value(*property, r_value);
}
static void fill_data_handle_map(const DerivedNodeTree &tree,
blender::bke::PersistentDataHandleMap &handle_map)
{
Set<ID *> used_ids;
findUsedIds(*tree.btree(), used_ids);
int current_handle = 0;
for (ID *id : used_ids) {
handle_map.add(current_handle, *id);
current_handle++;
}
}
/**
* Evaluate a node group to compute the output geometry.
* Currently, this uses a fairly basic and inefficient algorithm that might compute things more
* often than necessary. It's going to be replaced soon.
*/
static GeometrySet compute_geometry(const DerivedNodeTree &tree,
Span<const DOutputSocket *> group_input_sockets,
const DInputSocket &socket_to_compute,
GeometrySet input_geometry_set,
NodesModifierData *nmd,
const ModifierEvalContext *ctx)
{
blender::ResourceCollector resources;
blender::LinearAllocator<> &allocator = resources.linear_allocator();
blender::nodes::MultiFunctionByNode mf_by_node = get_multi_function_per_node(tree, resources);
Map<const DOutputSocket *, GMutablePointer> group_inputs;
if (group_input_sockets.size() > 0) {
Span<const DOutputSocket *> remaining_input_sockets = group_input_sockets;
/* If the group expects a geometry as first input, use the geometry that has been passed to
* modifier. */
const DOutputSocket *first_input_socket = group_input_sockets[0];
if (first_input_socket->bsocket()->type == SOCK_GEOMETRY) {
GeometrySet *geometry_set_in = allocator.construct<GeometrySet>(
std::move(input_geometry_set));
group_inputs.add_new(first_input_socket, geometry_set_in);
remaining_input_sockets = remaining_input_sockets.drop_front(1);
}
/* Initialize remaining group inputs. */
for (const DOutputSocket *socket : remaining_input_sockets) {
const CPPType &cpp_type = *blender::nodes::socket_cpp_type_get(*socket->typeinfo());
void *value_in = allocator.allocate(cpp_type.size(), cpp_type.alignment());
initialize_group_input(*nmd, *socket->bsocket(), cpp_type, value_in);
group_inputs.add_new(socket, {cpp_type, value_in});
}
}
Vector<const DInputSocket *> group_outputs;
group_outputs.append(&socket_to_compute);
blender::bke::PersistentDataHandleMap handle_map;
fill_data_handle_map(tree, handle_map);
GeometryNodesEvaluator evaluator{
group_inputs, group_outputs, mf_by_node, handle_map, ctx->object};
Vector<GMutablePointer> results = evaluator.execute();
BLI_assert(results.size() == 1);
GMutablePointer result = results[0];
GeometrySet output_geometry = std::move(*(GeometrySet *)result.get());
return output_geometry;
}
/**
* \note This could be done in #initialize_group_input, though that would require adding the
* the object as a parameter, so it's likely better to this check as a separate step.
*/
static void check_property_socket_sync(const Object *ob, ModifierData *md)
{
NodesModifierData *nmd = reinterpret_cast<NodesModifierData *>(md);
int i = 0;
LISTBASE_FOREACH_INDEX (const bNodeSocket *, socket, &nmd->node_group->inputs, i) {
/* The first socket is the special geometry socket for the modifier object. */
if (i == 0 && socket->type == SOCK_GEOMETRY) {
continue;
}
IDProperty *property = IDP_GetPropertyFromGroup(nmd->settings.properties, socket->identifier);
if (property == nullptr) {
if (socket->type == SOCK_STRING) {
BKE_modifier_set_error(ob, md, "String socket can not be exposed in the modifier");
}
else if (socket->type == SOCK_OBJECT) {
BKE_modifier_set_error(ob, md, "Object socket can not be exposed in the modifier");
}
else if (socket->type == SOCK_GEOMETRY) {
BKE_modifier_set_error(ob, md, "Node group can only have one geometry input");
}
else if (socket->type == SOCK_COLLECTION) {
BKE_modifier_set_error(ob, md, "Collection socket can not be exposed in the modifier");
}
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
else {
BKE_modifier_set_error(ob, md, "Missing property for input socket \"%s\"", socket->name);
}
continue;
}
const SocketPropertyType *property_type = get_socket_property_type(*socket);
if (!property_type->is_correct_type(*property)) {
BKE_modifier_set_error(
ob, md, "Property type does not match input socket \"(%s)\"", socket->name);
continue;
}
}
bool has_geometry_output = false;
LISTBASE_FOREACH (const bNodeSocket *, socket, &nmd->node_group->outputs) {
if (socket->type == SOCK_GEOMETRY) {
has_geometry_output = true;
}
}
if (!has_geometry_output) {
BKE_modifier_set_error(ob, md, "Node group must have a geometry output");
}
}
static void modifyGeometry(ModifierData *md,
const ModifierEvalContext *ctx,
GeometrySet &geometry_set)
{
NodesModifierData *nmd = reinterpret_cast<NodesModifierData *>(md);
if (nmd->node_group == nullptr) {
return;
}
check_property_socket_sync(ctx->object, md);
blender::nodes::NodeTreeRefMap tree_refs;
DerivedNodeTree tree{nmd->node_group, tree_refs};
if (tree.has_link_cycles()) {
BKE_modifier_set_error(ctx->object, md, "Node group has cycles");
return;
}
Span<const DNode *> input_nodes = tree.nodes_by_type("NodeGroupInput");
Span<const DNode *> output_nodes = tree.nodes_by_type("NodeGroupOutput");
if (input_nodes.size() > 1) {
return;
}
if (output_nodes.size() != 1) {
return;
}
Span<const DOutputSocket *> group_inputs = (input_nodes.size() == 1) ?
input_nodes[0]->outputs().drop_back(1) :
Span<const DOutputSocket *>{};
Span<const DInputSocket *> group_outputs = output_nodes[0]->inputs().drop_back(1);
if (group_outputs.size() == 0) {
return;
}
const DInputSocket *group_output = group_outputs[0];
if (group_output->idname() != "NodeSocketGeometry") {
return;
}
geometry_set = compute_geometry(
tree, group_inputs, *group_outputs[0], std::move(geometry_set), nmd, ctx);
}
static Mesh *modifyMesh(ModifierData *md, const ModifierEvalContext *ctx, Mesh *mesh)
{
GeometrySet geometry_set = GeometrySet::create_with_mesh(mesh, GeometryOwnershipType::Editable);
geometry_set.get_component_for_write<MeshComponent>().copy_vertex_group_names_from_object(
*ctx->object);
modifyGeometry(md, ctx, geometry_set);
Mesh *new_mesh = geometry_set.get_component_for_write<MeshComponent>().release();
if (new_mesh == nullptr) {
return BKE_mesh_new_nomain(0, 0, 0, 0, 0);
}
return new_mesh;
}
static void modifyGeometrySet(ModifierData *md,
const ModifierEvalContext *ctx,
GeometrySet *geometry_set)
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
{
modifyGeometry(md, ctx, *geometry_set);
}
/* Drawing the properties manually with #uiItemR instead of #uiDefAutoButsRNA allows using
* the node socket identifier for the property names, since they are unique, but also having
* the correct label displayed in the UI. */
static void draw_property_for_socket(uiLayout *layout,
PointerRNA *settings_ptr,
const IDProperty *modifier_props,
const bNodeSocket &socket)
{
const SocketPropertyType *property_type = get_socket_property_type(socket);
if (property_type == nullptr) {
return;
}
/* The property should be created in #MOD_nodes_update_interface with the correct type. */
IDProperty *property = IDP_GetPropertyFromGroup(modifier_props, socket.identifier);
/* IDProperties can be removed with python, so there could be a situation where
* there isn't a property for a socket or it doesn't have the correct type. */
if (property != nullptr && property_type->is_correct_type(*property)) {
char socket_id_esc[sizeof(socket.identifier) * 2];
BLI_str_escape(socket_id_esc, socket.identifier, sizeof(socket_id_esc));
char rna_path[sizeof(socket_id_esc) + 4];
BLI_snprintf(rna_path, ARRAY_SIZE(rna_path), "[\"%s\"]", socket_id_esc);
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
uiItemR(layout, settings_ptr, rna_path, 0, socket.name, ICON_NONE);
}
}
static void panel_draw(const bContext *C, Panel *panel)
{
uiLayout *layout = panel->layout;
PointerRNA *ptr = modifier_panel_get_property_pointers(panel, nullptr);
NodesModifierData *nmd = static_cast<NodesModifierData *>(ptr->data);
uiLayoutSetPropSep(layout, true);
/* This should be removed, but animation currently doesn't work with the IDProperties. */
uiLayoutSetPropDecorate(layout, false);
uiTemplateID(layout,
C,
ptr,
"node_group",
"node.new_geometry_node_group_assign",
nullptr,
nullptr,
UI: Redesigned data-block selectors The previous design is rather old and has a couple of problems: * Scalability: The current solution of adding little icon buttons next to the data-block name field doesn't scale well. It only works if there's a small number of operations. We need to be able to place more items there for better data-block management. Especially with the introduction of library overrides. * Discoverability: It's not obvious what some of the icons do. They appear and disappear, but it's not obvious why some are available at times and others not. * Unclear Status: Currently their library status (linked, indirectly linked, broken link, library override) isn't really clear. * Unusual behavior: Some of the icon buttons allow Shift or Ctrl clicking to invoke alternative behaviors. This is not a usual pattern in Blender. This patch does the following changes: * Adds a menu to the right of the name button to access all kinds of operations (create, delete, unlink, user management, library overrides, etc). * Make good use of the "disabled hint" for tooltips, to explain why buttons are disabled. The UI team wants to establish this as a good practise. * Use superimposed icons for duplicate and unlink, rather than extra buttons (uses less space, looks less distracting and is a nice + consistent design language). * Remove fake user and user count button, they are available from the menu now. * Support tooltips for superimposed icons (committed mouse hover feedback to master already). * Slightly increase size of the name button - it was already a bit small before, and the move from real buttons to superimposed icons reduces usable space for the name itself. * More clearly differentiate between duplicate and creating a new data-block. The latter is only available in the menu. * Display library status icon on the left (linked, missing library, overridden, asset) * Disables "Make Single User" button - in review we weren't sure if there are good use-cases for it, so better to see if we can remove it. Note that I do expect some aspects of this design to change still. I think some changes are problematic, but others disagreed. I will open a feedback thread on devtalk to see what others think. Differential Revision: https://developer.blender.org/D8554 Reviewed by: Bastien Montagne Design discussed and agreed on with the UI team, also see T79959.
2020-12-18 18:12:11 +01:00
nullptr,
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
0,
false,
nullptr);
if (nmd->node_group != nullptr && nmd->settings.properties != nullptr) {
PointerRNA settings_ptr;
RNA_pointer_create(ptr->owner_id, &RNA_NodesModifierSettings, &nmd->settings, &settings_ptr);
LISTBASE_FOREACH (bNodeSocket *, socket, &nmd->node_group->inputs) {
draw_property_for_socket(layout, &settings_ptr, nmd->settings.properties, *socket);
}
}
modifier_panel_end(layout, ptr);
}
static void panelRegister(ARegionType *region_type)
{
modifier_panel_register(region_type, eModifierType_Nodes, panel_draw);
}
static void blendWrite(BlendWriter *writer, const ModifierData *md)
{
const NodesModifierData *nmd = reinterpret_cast<const NodesModifierData *>(md);
if (nmd->settings.properties != nullptr) {
/* Note that the property settings are based on the socket type info
* and don't necessarily need to be written, but we can't just free them. */
IDP_BlendWrite(writer, nmd->settings.properties);
}
}
static void blendRead(BlendDataReader *reader, ModifierData *md)
{
NodesModifierData *nmd = reinterpret_cast<NodesModifierData *>(md);
BLO_read_data_address(reader, &nmd->settings.properties);
IDP_BlendDataRead(reader, &nmd->settings.properties);
}
static void copyData(const ModifierData *md, ModifierData *target, const int flag)
{
const NodesModifierData *nmd = reinterpret_cast<const NodesModifierData *>(md);
NodesModifierData *tnmd = reinterpret_cast<NodesModifierData *>(target);
BKE_modifier_copydata_generic(md, target, flag);
if (nmd->settings.properties != nullptr) {
tnmd->settings.properties = IDP_CopyProperty_ex(nmd->settings.properties, flag);
}
}
static void freeData(ModifierData *md)
{
NodesModifierData *nmd = reinterpret_cast<NodesModifierData *>(md);
if (nmd->settings.properties != nullptr) {
IDP_FreeProperty_ex(nmd->settings.properties, false);
nmd->settings.properties = nullptr;
}
}
static void requiredDataMask(Object *UNUSED(ob),
ModifierData *UNUSED(md),
CustomData_MeshMasks *r_cddata_masks)
{
/* We don't know what the node tree will need. If there are vertex groups, it is likely that the
* node tree wants to access them. */
r_cddata_masks->vmask |= CD_MASK_MDEFORMVERT;
}
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
ModifierTypeInfo modifierType_Nodes = {
/* name */ "GeometryNodes",
/* structName */ "NodesModifierData",
/* structSize */ sizeof(NodesModifierData),
/* srna */ &RNA_NodesModifier,
/* type */ eModifierTypeType_Constructive,
/* flags */
static_cast<ModifierTypeFlag>(
eModifierTypeFlag_AcceptsMesh | eModifierTypeFlag_SupportsEditmode |
eModifierTypeFlag_EnableInEditmode | eModifierTypeFlag_SupportsMapping),
/* icon */ ICON_NODETREE,
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
/* copyData */ copyData,
/* deformVerts */ nullptr,
/* deformMatrices */ nullptr,
/* deformVertsEM */ nullptr,
/* deformMatricesEM */ nullptr,
/* modifyMesh */ modifyMesh,
/* modifyHair */ nullptr,
/* modifyGeometrySet */ modifyGeometrySet,
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
/* modifyVolume */ nullptr,
/* initData */ initData,
/* requiredDataMask */ requiredDataMask,
Geometry Nodes: initial scattering and geometry nodes This is the initial merge from the geometry-nodes branch. Nodes: * Attribute Math * Boolean * Edge Split * Float Compare * Object Info * Point Distribute * Point Instance * Random Attribute * Random Float * Subdivision Surface * Transform * Triangulate It includes the initial evaluation of geometry node groups in the Geometry Nodes modifier. Notes on the Generic attribute access API The API adds an indirection for attribute access. That has the following benefits: * Most code does not have to care about how an attribute is stored internally. This is mainly necessary, because we have to deal with "legacy" attributes such as vertex weights and attributes that are embedded into other structs such as vertex positions. * When reading from an attribute, we generally don't care what domain the attribute is stored on. So we want to abstract away the interpolation that that adapts attributes from one domain to another domain (this is not actually implemented yet). Other possible improvements for later iterations include: * Actually implement interpolation between domains. * Don't use inheritance for the different attribute types. A single class for read access and one for write access might be enough, because we know all the ways in which attributes are stored internally. We don't want more different internal structures in the future. On the contrary, ideally we can consolidate the different storage formats in the future to reduce the need for this indirection. * Remove the need for heap allocations when creating attribute accessors. It includes commits from: * Dalai Felinto * Hans Goudey * Jacques Lucke * Léo Depoix
2020-12-02 13:25:25 +01:00
/* freeData */ freeData,
/* isDisabled */ isDisabled,
/* updateDepsgraph */ updateDepsgraph,
/* dependsOnTime */ nullptr,
/* dependsOnNormals */ nullptr,
/* foreachIDLink */ foreachIDLink,
/* foreachTexLink */ nullptr,
/* freeRuntimeData */ nullptr,
/* panelRegister */ panelRegister,
/* blendWrite */ blendWrite,
/* blendRead */ blendRead,
};