archive/blender-archive
Archived
1
1
Fork 0
Code Pull Requests Packages Activity
This repository has been archived on 2023-10-09.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
2e3bc99590d13dc5a1ae2f9f574371f99b0677d6
blender-archive/source/blender/compositor/intern/COM_NodeOperationBuilder.cpp

693 lines
21 KiB
C++
Raw Normal View History

Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/*
* 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.
Cleanup: conform headers to have license first Also remove doxy comments for licenses and add missing GPL header.
2019-02-18 07:21:50 +11:00
*
* Copyright 2013, Blender Foundation.
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
*/
extern "C" {
#include "BLI_utildefines.h"
}
#include "COM_NodeConverter.h"
#include "COM_Converter.h"
#include "COM_Debug.h"
#include "COM_ExecutionSystem.h"
#include "COM_Node.h"
#include "COM_SocketProxyNode.h"
#include "COM_NodeOperation.h"
#include "COM_PreviewOperation.h"
#include "COM_SetValueOperation.h"
#include "COM_SetVectorOperation.h"
#include "COM_SetColorOperation.h"
#include "COM_SocketProxyOperation.h"
#include "COM_ReadBufferOperation.h"
#include "COM_WriteBufferOperation.h"
Fix T40986: crash on using the viewer node inside of group nodes. Viewers were activated both inside the active group as well as the top level tree (the latter being a quick fix for getting a fallback viewer). This caused a race condition on the shared viewer image. Now the active viewer is defined at node conversion time in the converter so that only one can be active at a time without each node having to follow complicated rules for exclusion.
2014-07-08 12:48:41 +02:00
#include "COM_ViewerOperation.h"
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
#include "COM_NodeOperationBuilder.h" /* own include */
NodeOperationBuilder::NodeOperationBuilder(const CompositorContext *context, bNodeTree *b_nodetree) :
m_context(context),
Fix T40986: crash on using the viewer node inside of group nodes. Viewers were activated both inside the active group as well as the top level tree (the latter being a quick fix for getting a fallback viewer). This caused a race condition on the shared viewer image. Now the active viewer is defined at node conversion time in the converter so that only one can be active at a time without each node having to follow complicated rules for exclusion.
2014-07-08 12:48:41 +02:00
m_current_node(NULL),
m_active_viewer(NULL)
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
{
m_graph.from_bNodeTree(*context, b_nodetree);
}
NodeOperationBuilder::~NodeOperationBuilder()
{
}
void NodeOperationBuilder::convertToOperations(ExecutionSystem *system)
{
/* interface handle for nodes */
NodeConverter converter(this);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (int index = 0; index < m_graph.nodes().size(); index++) {
Node *node = (Node *)m_graph.nodes()[index];
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
m_current_node = node;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
DebugInfo::node_to_operations(node);
node->convertToOperations(converter, *m_context);
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
m_current_node = NULL;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* The input map constructed by nodes maps operation inputs to node inputs.
* Inverting yields a map of node inputs to all connected operation inputs,
* so multiple operations can use the same node input.
*/
OpInputInverseMap inverse_input_map;
for (InputSocketMap::const_iterator it = m_input_map.begin(); it != m_input_map.end(); ++it)
inverse_input_map[it->second].push_back(it->first);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (NodeGraph::Links::const_iterator it = m_graph.links().begin(); it != m_graph.links().end(); ++it) {
const NodeGraph::Link &link = *it;
NodeOutput *from = link.getFromSocket();
NodeInput *to = link.getToSocket();
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
NodeOperationOutput *op_from = find_operation_output(m_output_map, from);
const OpInputs &op_to_list = find_operation_inputs(inverse_input_map, to);
if (!op_from || op_to_list.empty()) {
/* XXX allow this? error/debug message? */
Fix T39799: Backdrop (compositor) ignores alpha. This issue is because of a somewhat "special" behavior in old code, which got lost during rB09874df: There was a variant of the `relinkConnections` function which would leave the socket completely unconnected. This is not a valid state really (given that each unconnected input must otherwise connected to a constant `Set` type node), but was used as a way to distinguish connected alpha/depth sockets in composite and viewer output nodes. https://developer.blender.org/diffusion/B/browse/master/source/blender/compositor/intern/COM_InputSocket.cpp;28a829893c702918afc5ac1945a06eaefa611594$69 After the large cleanup patch ({D309}) every socket is now automatically connected to a constant, such that `getInputSocketReader` will never return a NULL pointer. This breaks the previous test method, which needs to be replaced by more explicit flags. Luckily this was done only for very few output nodes (Composite, Viewer, Output-File). These now use the regular SetValueOperation default in case "use alpha" is disabled, but set this to an explicit 1.0 value instead of mapping to the node socket.
2014-04-25 12:00:35 +02:00
//BLI_assert(false);
/* XXX note: this can happen with certain nodes (e.g. OutputFile)
* which only generate operations in certain circumstances (rendering)
* just let this pass silently for now ...
*/
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
continue;
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (OpInputs::const_iterator it = op_to_list.begin(); it != op_to_list.end(); ++it) {
NodeOperationInput *op_to = *it;
addLink(op_from, op_to);
}
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Fix T44149: Compositing : Node Groups do not work correctly Input constants are to be connected before removing proxies, otherwise node groups might give totally different result. This is a regression and to be put into final release.
2015-03-27 17:25:12 +05:00
add_operation_input_constants();
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Fix T44040: Blender crashes when nodes are muted It was actually an old issue with wrong conversion happening for muted nodes, which wasn't visible before memory optimization commit. This is to be backported to the final release.
2015-03-19 16:04:49 +05:00
resolve_proxies();
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
add_datatype_conversions();
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
determineResolutions();
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* surround complex ops with read/write buffer */
add_complex_operation_buffers();
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* links not available from here on */
/* XXX make m_links a local variable to avoid confusion! */
m_links.clear();
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
prune_operations();
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* ensure topological (link-based) order of nodes */
/*sort_operations();*/ /* not needed yet */
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* create execution groups */
group_operations();
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* transfer resulting operations to the system */
system->set_operations(m_operations, m_groups);
}
void NodeOperationBuilder::addOperation(NodeOperation *operation)
{
m_operations.push_back(operation);
}
void NodeOperationBuilder::mapInputSocket(NodeInput *node_socket, NodeOperationInput *operation_socket)
{
BLI_assert(m_current_node);
BLI_assert(node_socket->getNode() == m_current_node);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* note: this maps operation sockets to node sockets.
* for resolving links the map will be inverted first in convertToOperations,
* to get a list of links for each node input socket.
*/
m_input_map[operation_socket] = node_socket;
}
void NodeOperationBuilder::mapOutputSocket(NodeOutput *node_socket, NodeOperationOutput *operation_socket)
{
BLI_assert(m_current_node);
BLI_assert(node_socket->getNode() == m_current_node);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
m_output_map[node_socket] = operation_socket;
}
void NodeOperationBuilder::addLink(NodeOperationOutput *from, NodeOperationInput *to)
{
if (to->isConnected())
return;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
m_links.push_back(Link(from, to));
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* register with the input */
to->setLink(from);
}
void NodeOperationBuilder::removeInputLink(NodeOperationInput *to)
{
for (Links::iterator it = m_links.begin(); it != m_links.end(); ++it) {
Link &link = *it;
if (link.to() == to) {
/* unregister with the input */
to->setLink(NULL);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
m_links.erase(it);
return;
}
}
}
NodeInput *NodeOperationBuilder::find_node_input(const InputSocketMap &map, NodeOperationInput *op_input)
{
InputSocketMap::const_iterator it = map.find(op_input);
return (it != map.end() ? it->second : NULL);
}
const NodeOperationBuilder::OpInputs &NodeOperationBuilder::find_operation_inputs(const OpInputInverseMap &map, NodeInput *node_input)
{
static const OpInputs empty_list;
OpInputInverseMap::const_iterator it = map.find(node_input);
return (it != map.end() ? it->second : empty_list);
}
NodeOperationOutput *NodeOperationBuilder::find_operation_output(const OutputSocketMap &map, NodeOutput *node_output)
{
OutputSocketMap::const_iterator it = map.find(node_output);
return (it != map.end() ? it->second : NULL);
}
PreviewOperation *NodeOperationBuilder::make_preview_operation() const
{
BLI_assert(m_current_node);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
if (!(m_current_node->getbNode()->flag & NODE_PREVIEW))
return NULL;
/* previews only in the active group */
if (!m_current_node->isInActiveGroup())
return NULL;
/* do not calculate previews of hidden nodes */
if (m_current_node->getbNode()->flag & NODE_HIDDEN)
return NULL;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
bNodeInstanceHash *previews = m_context->getPreviewHash();
if (previews) {
PreviewOperation *operation = new PreviewOperation(m_context->getViewSettings(), m_context->getDisplaySettings());
operation->setbNodeTree(m_context->getbNodeTree());
operation->verifyPreview(previews, m_current_node->getInstanceKey());
return operation;
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
return NULL;
}
void NodeOperationBuilder::addPreview(NodeOperationOutput *output)
{
PreviewOperation *operation = make_preview_operation();
if (operation) {
addOperation(operation);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
addLink(output, operation->getInputSocket(0));
}
}
void NodeOperationBuilder::addNodeInputPreview(NodeInput *input)
{
PreviewOperation *operation = make_preview_operation();
if (operation) {
addOperation(operation);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
mapInputSocket(input, operation->getInputSocket(0));
}
}
Fix T40986: crash on using the viewer node inside of group nodes. Viewers were activated both inside the active group as well as the top level tree (the latter being a quick fix for getting a fallback viewer). This caused a race condition on the shared viewer image. Now the active viewer is defined at node conversion time in the converter so that only one can be active at a time without each node having to follow complicated rules for exclusion.
2014-07-08 12:48:41 +02:00
void NodeOperationBuilder::registerViewer(ViewerOperation *viewer)
{
if (m_active_viewer) {
if (m_current_node->isInActiveGroup()) {
/* deactivate previous viewer */
m_active_viewer->setActive(false);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Fix T40986: crash on using the viewer node inside of group nodes. Viewers were activated both inside the active group as well as the top level tree (the latter being a quick fix for getting a fallback viewer). This caused a race condition on the shared viewer image. Now the active viewer is defined at node conversion time in the converter so that only one can be active at a time without each node having to follow complicated rules for exclusion.
2014-07-08 12:48:41 +02:00
m_active_viewer = viewer;
viewer->setActive(true);
}
}
else {
if (m_current_node->getbNodeTree() == m_context->getbNodeTree()) {
m_active_viewer = viewer;
viewer->setActive(true);
}
}
}
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/****************************
**** Optimization Steps ****
****************************/
void NodeOperationBuilder::add_datatype_conversions()
{
Links convert_links;
for (Links::const_iterator it = m_links.begin(); it != m_links.end(); ++it) {
const Link &link = *it;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Fix T41067: Muted nodes still do data type conversions. Proxy operations from muted nodes would still create conversion operations where the datatypes don't match, which creates unexpected behavior. Arguably datatype conversion could still happen even when the main operation is muted, but this would be a design change and so is disabled now.
2014-07-15 10:55:49 +02:00
/* proxy operations can skip data type conversion */
NodeOperation *from_op = &link.from()->getOperation();
NodeOperation *to_op = &link.to()->getOperation();
Fix T41573: Compositing reroute converts floats to red color At compo level, only skip data type conversion if both ends of a link do not require it. Thanks to Lukas for advices and review.
2014-09-08 12:27:00 +02:00
if (!(from_op->useDatatypeConversion() || to_op->useDatatypeConversion()))
Fix T41067: Muted nodes still do data type conversions. Proxy operations from muted nodes would still create conversion operations where the datatypes don't match, which creates unexpected behavior. Arguably datatype conversion could still happen even when the main operation is muted, but this would be a design change and so is disabled now.
2014-07-15 10:55:49 +02:00
continue;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
if (link.from()->getDataType() != link.to()->getDataType())
convert_links.push_back(link);
}
for (Links::const_iterator it = convert_links.begin(); it != convert_links.end(); ++it) {
const Link &link = *it;
NodeOperation *converter = Converter::convertDataType(link.from(), link.to());
if (converter) {
addOperation(converter);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
removeInputLink(link.to());
addLink(link.from(), converter->getInputSocket(0));
addLink(converter->getOutputSocket(0), link.to());
}
}
}
void NodeOperationBuilder::add_operation_input_constants()
{
/* Note: unconnected inputs cached first to avoid modifying
* m_operations while iterating over it
*/
typedef std::vector<NodeOperationInput*> Inputs;
Inputs pending_inputs;
for (Operations::const_iterator it = m_operations.begin(); it != m_operations.end(); ++it) {
NodeOperation *op = *it;
for (int k = 0; k < op->getNumberOfInputSockets(); ++k) {
NodeOperationInput *input = op->getInputSocket(k);
if (!input->isConnected())
pending_inputs.push_back(input);
}
}
for (Inputs::const_iterator it = pending_inputs.begin(); it != pending_inputs.end(); ++it) {
NodeOperationInput *input = *it;
add_input_constant_value(input, find_node_input(m_input_map, input));
}
}
void NodeOperationBuilder::add_input_constant_value(NodeOperationInput *input, NodeInput *node_input)
{
switch (input->getDataType()) {
case COM_DT_VALUE: {
float value;
if (node_input && node_input->getbNodeSocket())
value = node_input->getEditorValueFloat();
else
value = 0.0f;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
SetValueOperation *op = new SetValueOperation();
op->setValue(value);
addOperation(op);
addLink(op->getOutputSocket(), input);
break;
}
case COM_DT_COLOR: {
float value[4];
if (node_input && node_input->getbNodeSocket())
node_input->getEditorValueColor(value);
else
zero_v4(value);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
SetColorOperation *op = new SetColorOperation();
op->setChannels(value);
addOperation(op);
addLink(op->getOutputSocket(), input);
break;
}
case COM_DT_VECTOR: {
float value[3];
if (node_input && node_input->getbNodeSocket())
node_input->getEditorValueVector(value);
else
zero_v3(value);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
SetVectorOperation *op = new SetVectorOperation();
op->setVector(value);
addOperation(op);
addLink(op->getOutputSocket(), input);
break;
}
}
}
void NodeOperationBuilder::resolve_proxies()
{
Links proxy_links;
for (Links::const_iterator it = m_links.begin(); it != m_links.end(); ++it) {
const Link &link = *it;
/* don't replace links from proxy to proxy, since we may need them for replacing others! */
if (link.from()->getOperation().isProxyOperation() &&
Style cleanup
2014-04-17 20:00:49 +10:00
!link.to()->getOperation().isProxyOperation())
{
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
proxy_links.push_back(link);
}
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (Links::const_iterator it = proxy_links.begin(); it != proxy_links.end(); ++it) {
const Link &link = *it;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
NodeOperationInput *to = link.to();
NodeOperationOutput *from = link.from();
do {
/* walk upstream bypassing the proxy operation */
from = from->getOperation().getInputSocket(0)->getLink();
} while (from && from->getOperation().isProxyOperation());
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
removeInputLink(to);
/* we may not have a final proxy input link,
* in that case it just gets dropped
*/
if (from)
addLink(from, to);
}
}
void NodeOperationBuilder::determineResolutions()
{
/* determine all resolutions of the operations (Width/Height) */
for (Operations::const_iterator it = m_operations.begin(); it != m_operations.end(); ++it) {
NodeOperation *op = *it;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
if (op->isOutputOperation(m_context->isRendering()) && !op->isPreviewOperation()) {
unsigned int resolution[2] = {0, 0};
unsigned int preferredResolution[2] = {0, 0};
op->determineResolution(resolution, preferredResolution);
op->setResolution(resolution);
}
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (Operations::const_iterator it = m_operations.begin(); it != m_operations.end(); ++it) {
NodeOperation *op = *it;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
if (op->isOutputOperation(m_context->isRendering()) && op->isPreviewOperation()) {
unsigned int resolution[2] = {0, 0};
unsigned int preferredResolution[2] = {0, 0};
op->determineResolution(resolution, preferredResolution);
op->setResolution(resolution);
}
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* add convert resolution operations when needed */
{
Links convert_links;
for (Links::const_iterator it = m_links.begin(); it != m_links.end(); ++it) {
const Link &link = *it;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
if (link.to()->getResizeMode() != COM_SC_NO_RESIZE) {
NodeOperation &from_op = link.from()->getOperation();
NodeOperation &to_op = link.to()->getOperation();
if (from_op.getWidth() != to_op.getWidth() || from_op.getHeight() != to_op.getHeight())
convert_links.push_back(link);
}
}
for (Links::const_iterator it = convert_links.begin(); it != convert_links.end(); ++it) {
const Link &link = *it;
Converter::convertResolution(*this, link.from(), link.to());
}
}
}
NodeOperationBuilder::OpInputs NodeOperationBuilder::cache_output_links(NodeOperationOutput *output) const
{
OpInputs inputs;
for (Links::const_iterator it = m_links.begin(); it != m_links.end(); ++it) {
const Link &link = *it;
if (link.from() == output)
inputs.push_back(link.to());
}
return inputs;
}
WriteBufferOperation *NodeOperationBuilder::find_attached_write_buffer_operation(NodeOperationOutput *output) const
{
for (Links::const_iterator it = m_links.begin(); it != m_links.end(); ++it) {
const Link &link = *it;
if (link.from() == output) {
NodeOperation &op = link.to()->getOperation();
if (op.isWriteBufferOperation())
return (WriteBufferOperation *)(&op);
}
}
return NULL;
}
Compositor: Code cleanup, prepare for strict C++ flags
2015-03-27 15:49:07 +05:00
void NodeOperationBuilder::add_input_buffers(NodeOperation * /*operation*/,
NodeOperationInput *input)
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
{
if (!input->isConnected())
return;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
NodeOperationOutput *output = input->getLink();
if (output->getOperation().isReadBufferOperation()) {
/* input is already buffered, no need to add another */
return;
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* this link will be replaced below */
removeInputLink(input);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* check of other end already has write operation, otherwise add a new one */
WriteBufferOperation *writeoperation = find_attached_write_buffer_operation(output);
if (!writeoperation) {
D627: Memory usage optimization for the compositor. The compostor used a fixed size of 4 floats to hold pixel data. this patch will select size of a pixel based on its type. It uses 1 float for Value, 3 float for vector and 4 floats for color data types. When benchmarking on shots (opening shot of caminandes) we get a reduction of memory of 30% and a tiny speedup as less data transformations needs to take place (but these are negligable. More information of the patch can be found on https://developer.blender.org/D627 and http://wiki.blender.org/index.php/Dev:Ref/Proposals/Compositor2014_p1.1_TD Developers: jbakker & mdewanchand Thanks for Sergey for his indept review.
2015-01-19 18:13:26 +01:00
writeoperation = new WriteBufferOperation(output->getDataType());
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
writeoperation->setbNodeTree(m_context->getbNodeTree());
addOperation(writeoperation);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
addLink(output, writeoperation->getInputSocket(0));
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
writeoperation->readResolutionFromInputSocket();
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* add readbuffer op for the input */
D627: Memory usage optimization for the compositor. The compostor used a fixed size of 4 floats to hold pixel data. this patch will select size of a pixel based on its type. It uses 1 float for Value, 3 float for vector and 4 floats for color data types. When benchmarking on shots (opening shot of caminandes) we get a reduction of memory of 30% and a tiny speedup as less data transformations needs to take place (but these are negligable. More information of the patch can be found on https://developer.blender.org/D627 and http://wiki.blender.org/index.php/Dev:Ref/Proposals/Compositor2014_p1.1_TD Developers: jbakker & mdewanchand Thanks for Sergey for his indept review.
2015-01-19 18:13:26 +01:00
ReadBufferOperation *readoperation = new ReadBufferOperation(output->getDataType());
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
readoperation->setMemoryProxy(writeoperation->getMemoryProxy());
this->addOperation(readoperation);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
addLink(readoperation->getOutputSocket(), input);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
readoperation->readResolutionFromWriteBuffer();
}
void NodeOperationBuilder::add_output_buffers(NodeOperation *operation, NodeOperationOutput *output)
{
/* cache connected sockets, so we can safely remove links first before replacing them */
OpInputs targets = cache_output_links(output);
if (targets.empty())
return;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
WriteBufferOperation *writeOperation = NULL;
for (OpInputs::const_iterator it = targets.begin(); it != targets.end(); ++it) {
NodeOperationInput *target = *it;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* try to find existing write buffer operation */
if (target->getOperation().isWriteBufferOperation()) {
BLI_assert(writeOperation == NULL); /* there should only be one write op connected */
writeOperation = (WriteBufferOperation *)(&target->getOperation());
}
else {
/* remove all links to other nodes */
removeInputLink(target);
}
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* if no write buffer operation exists yet, create a new one */
if (!writeOperation) {
D627: Memory usage optimization for the compositor. The compostor used a fixed size of 4 floats to hold pixel data. this patch will select size of a pixel based on its type. It uses 1 float for Value, 3 float for vector and 4 floats for color data types. When benchmarking on shots (opening shot of caminandes) we get a reduction of memory of 30% and a tiny speedup as less data transformations needs to take place (but these are negligable. More information of the patch can be found on https://developer.blender.org/D627 and http://wiki.blender.org/index.php/Dev:Ref/Proposals/Compositor2014_p1.1_TD Developers: jbakker & mdewanchand Thanks for Sergey for his indept review.
2015-01-19 18:13:26 +01:00
writeOperation = new WriteBufferOperation(operation->getOutputSocket()->getDataType());
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
writeOperation->setbNodeTree(m_context->getbNodeTree());
addOperation(writeOperation);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
addLink(output, writeOperation->getInputSocket(0));
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
writeOperation->readResolutionFromInputSocket();
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* add readbuffer op for every former connected input */
for (OpInputs::const_iterator it = targets.begin(); it != targets.end(); ++it) {
NodeOperationInput *target = *it;
if (&target->getOperation() == writeOperation)
continue; /* skip existing write op links */
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
D627: Memory usage optimization for the compositor. The compostor used a fixed size of 4 floats to hold pixel data. this patch will select size of a pixel based on its type. It uses 1 float for Value, 3 float for vector and 4 floats for color data types. When benchmarking on shots (opening shot of caminandes) we get a reduction of memory of 30% and a tiny speedup as less data transformations needs to take place (but these are negligable. More information of the patch can be found on https://developer.blender.org/D627 and http://wiki.blender.org/index.php/Dev:Ref/Proposals/Compositor2014_p1.1_TD Developers: jbakker & mdewanchand Thanks for Sergey for his indept review.
2015-01-19 18:13:26 +01:00
ReadBufferOperation *readoperation = new ReadBufferOperation(operation->getOutputSocket()->getDataType());
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
readoperation->setMemoryProxy(writeOperation->getMemoryProxy());
addOperation(readoperation);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
addLink(readoperation->getOutputSocket(), target);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
readoperation->readResolutionFromWriteBuffer();
}
}
void NodeOperationBuilder::add_complex_operation_buffers()
{
/* note: complex ops and get cached here first, since adding operations
* will invalidate iterators over the main m_operations
*/
Operations complex_ops;
for (Operations::const_iterator it = m_operations.begin(); it != m_operations.end(); ++it)
if ((*it)->isComplex())
complex_ops.push_back(*it);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (Operations::const_iterator it = complex_ops.begin(); it != complex_ops.end(); ++it) {
NodeOperation *op = *it;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
DebugInfo::operation_read_write_buffer(op);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (int index = 0; index < op->getNumberOfInputSockets(); index++)
add_input_buffers(op, op->getInputSocket(index));
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (int index = 0; index < op->getNumberOfOutputSockets(); index++)
add_output_buffers(op, op->getOutputSocket(index));
}
}
typedef std::set<NodeOperation*> Tags;
static void find_reachable_operations_recursive(Tags &reachable, NodeOperation *op)
{
if (reachable.find(op) != reachable.end())
return;
reachable.insert(op);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (int i = 0; i < op->getNumberOfInputSockets(); ++i) {
NodeOperationInput *input = op->getInputSocket(i);
if (input->isConnected())
find_reachable_operations_recursive(reachable, &input->getLink()->getOperation());
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* associated write-buffer operations are executed as well */
if (op->isReadBufferOperation()) {
ReadBufferOperation *read_op = (ReadBufferOperation *)op;
MemoryProxy *memproxy = read_op->getMemoryProxy();
find_reachable_operations_recursive(reachable, memproxy->getWriteBufferOperation());
}
}
void NodeOperationBuilder::prune_operations()
{
Tags reachable;
for (Operations::const_iterator it = m_operations.begin(); it != m_operations.end(); ++it) {
NodeOperation *op = *it;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* output operations are primary executed operations */
if (op->isOutputOperation(m_context->isRendering()))
find_reachable_operations_recursive(reachable, op);
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* delete unreachable operations */
Operations reachable_ops;
for (Operations::const_iterator it = m_operations.begin(); it != m_operations.end(); ++it) {
NodeOperation *op = *it;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
if (reachable.find(op) != reachable.end())
reachable_ops.push_back(op);
else
delete op;
}
/* finally replace the operations list with the pruned list */
m_operations = reachable_ops;
}
/* topological (depth-first) sorting of operations */
static void sort_operations_recursive(NodeOperationBuilder::Operations &sorted, Tags &visited, NodeOperation *op)
{
if (visited.find(op) != visited.end())
return;
visited.insert(op);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (int i = 0; i < op->getNumberOfInputSockets(); ++i) {
NodeOperationInput *input = op->getInputSocket(i);
if (input->isConnected())
sort_operations_recursive(sorted, visited, &input->getLink()->getOperation());
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
sorted.push_back(op);
}
void NodeOperationBuilder::sort_operations()
{
Operations sorted;
sorted.reserve(m_operations.size());
Tags visited;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
for (Operations::const_iterator it = m_operations.begin(); it != m_operations.end(); ++it)
sort_operations_recursive(sorted, visited, *it);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
m_operations = sorted;
}
static void add_group_operations_recursive(Tags &visited, NodeOperation *op, ExecutionGroup *group)
{
if (visited.find(op) != visited.end())
return;
visited.insert(op);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
if (!group->addOperation(op))
return;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* add all eligible input ops to the group */
for (int i = 0; i < op->getNumberOfInputSockets(); ++i) {
NodeOperationInput *input = op->getInputSocket(i);
if (input->isConnected())
add_group_operations_recursive(visited, &input->getLink()->getOperation(), group);
}
}
ExecutionGroup *NodeOperationBuilder::make_group(NodeOperation *op)
{
ExecutionGroup *group = new ExecutionGroup();
m_groups.push_back(group);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
Tags visited;
add_group_operations_recursive(visited, op, group);
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
return group;
}
void NodeOperationBuilder::group_operations()
{
for (Operations::const_iterator it = m_operations.begin(); it != m_operations.end(); ++it) {
NodeOperation *op = *it;
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
if (op->isOutputOperation(m_context->isRendering())) {
ExecutionGroup *group = make_group(op);
group->setOutputExecutionGroup(true);
}
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
/* add new groups for associated memory proxies where needed */
if (op->isReadBufferOperation()) {
ReadBufferOperation *read_op = (ReadBufferOperation *)op;
MemoryProxy *memproxy = read_op->getMemoryProxy();
Cleanup: trailing space for compositor
2018-06-17 17:05:29 +02:00
Structural cleanup and improvements for the compositor. Many parts of the compositor are unnecessarily complicated. This patch aims at reducing the complexity of writing nodes and making the code more transparent. == Separating Nodes and Operations == Currently these are both mixed in the same graph, even though they have very different purposes and are used at distinct stages in the compositing process. The patch introduces dedicated graph classes for nodes and for operations. This removes the need for a lot of special case checks (isOperation etc.) and explicit type casts. It simplifies the code since it becomes clear at every stage what type of node we are dealing with. The compiler can use static typing to avoid common bugs from mixing up these types and fewer runtime sanity checks are needed. == Simplified Node Conversion == Converting nodes to operations was previously based on "relinking", i.e. nodes would start with by mirroring links in the Blender DNA node trees, then add operations and redirect these links to them. This was very hard to follow in many cases and required a lot of attention to avoid invalid states. Now there is a helper class called the NodeConverter, which is passed to nodes and implements a much simpler API for this process. Nodes can add operations and explicit connections as before, but defining "external" links to the inputs/outputs of the original node now uses mapping instead of directly modifying link data. Input data (node graph) and result (operations graph) are cleanly separated. == Removed Redundant Data Structures == A few redundant data structures have been removed, notably the SocketConnection. These are only needed temporarily during graph construction. For executing the compositor operations it is perfectly sufficient to store only the direct input link pointers. A common pointer indirection is avoided this way (which might also give a little performance improvement). == Avoid virtual recursive functions == Recursive virtual functions are evil. They are very hard to follow during debugging. At least in the parts this patch is concerned with these functions have been replaced by a non-virtual recursive core function (which might then call virtual non-recursive functions if needed). See for example NodeOperationBuilder::group_operations.
2014-04-15 16:06:12 +02:00
if (memproxy->getExecutor() == NULL) {
ExecutionGroup *group = make_group(memproxy->getWriteBufferOperation());
memproxy->setExecutor(group);
}
}
}
}
Copy Permalink