Global attributes now appear in the MMB information panel for TOP Network nodes in any context. The global attributes also use the same color scheme—pink for Strings, green for Integers, orange for Floats, and so on—as regular work item-level attributes.

The Total Generate Time, Total Cook Time, and Average Cook Time Per Work Item for a given node now appears in the MMB information panel for all TOP nodes.

Job script command line information and compatible services for a given node now appear in the MMB information panel for all Processor TOP nodes.

Select which output files to remove from disk for any TOP node with the Delete This Node’s Output Files from Disk RMB-menu option’s new Make Selection window.

Filter your bundle nodes by PDG node type (like by Schedulers, Processors, or Partitioners) with the new TOP filters located in the Bundle List pane > Filter drop-down list.

Add custom zones for your PDG path maps and view which schedulers created each mapping with the improved PDG Path Map pane.

Specify a default work item label for all the tasks in a TOP network that have the default label setting with the new Default Item Label parameter on the TOP Network node across all contexts.

New HOUDINI_DELAYSYNC_HDA environment variable that enables the delayed syncing of asset contents until the assets cook. This is available for all network types (including TOPs), and it is useful when you want to save on load/initialization time.

Save the global attributes that are stored on a TOP graph to its .hip file with the new Save Graph Attributes to .hip parameter on the TOP Network SOP node. This makes it so that when you next open the .hip file, all the global attributes and their settings will still be available on the graph.

Control how geometry from work items are grouped in a SOP node’s output with the new Group Work Items Using and Piece Attribute Name parameters on the TOP Geometry SOP node. For example, you can now group geometry by Point Groups, Primitive Groups, Point Attribute, Primitive Attribute, or not group the output at all.

Set up and manage PDG services directly in your PDG graphs with the new Service TOP nodes:

A new Python Script Service type is now available in the PDG Services window.

You can now configure a PDG service at any time during registration or after registration, but not when the service is running.

You can now register multiple PDG services of the same type (like multiple HDA Processor Services) and give them different names.

Control which attributes and relationships are imported from a target LOP node or .usd file with the new Import Attributes and Import Relationships parameters on the USD Import TOP node.

Write the lists of imported attribute names and relationship names to your work item attributes with the new Attribute Names and Relationship Names parameters on the USD Import TOP node.

Trim tokens from USD attributes (starting from the left-hand or right-hand sides of the names) when converting them to work item attributes with the new Namespace Cutoff Index parameter on the USD Import TOP node.

Apply a custom file tag to all the outputs produced by the USD Render TOP node’s work items with the new Output File Tag parameter.

When retrying failed URL requests, you can now specify the time interval between retries and which HTTP status codes to retry with the new Retry Backoff and Retry Status Codes parameters on the URL Request TOP node.

Failed attribute evaluations like out-of-bounds indices or invalid file tags in TOP node expressions now produce explicit warning messages that describe in detail the cause of the failed evaluations.

All the PDG methods for adding, removing, and accessing attributes have been moved to the new pdg.AttributeOwner base class from pdg.WorkItem. This change does not affect existing code that accesses attributes since pdg.WorkItem now inherits from this new base class and all the methods are still accessible from pdg.WorkItem instances.

The pdg.ValuePattern class now supports a more condensed syntax for specifying a range of values by using a string prefix for each value. The range is defined by square brackets after the prefix. For example, prefix[1-4] generates the following sequence of values: prefix1, prefix2, prefix3. The range can also include an optional step size. For example, prefix[1-10:2].

New pdg.AttributeUtils.ignoreWarnings() context block lets you temporarily suppress @attrib expression warning messages caused by the lack of valid work item data.

New prepareDirty() method for the pdg.Graph, pdg.Node, and pdg.WorkItem classes allows you to create a pdg.WorkItemDirty object. This object holds the description of a dirty operation that can be either discarded or performed at a later time.

New evaluateBool() method for pdg.Port lets you evaluate an integer port as a boolean.

The pdg.WorkItem.cookDuration() property now returns useful values at any stage of its work item’s cook. For example, pdg.WorkItem.cookDuration() will now return 0 if the work item has not started to cook and current_time - start_time if the work item is in the process of cooking.

Most pdg.Attribute-type classes (for example, like pdg.AttributeFile and so on) now have a new isArray property which is set to True for data types (int, float, string, and file) that support arrays of values.

PDG Python and Job API result data methods are now deprecated and have been replaced by new methods with names that reference input file and output file. For example, the new pdg.WorkItem.inputFiles method replaces pdg.WorkItem.inputResultData and the new pdg.WorkItem.outputFiles method replaces pdg.WorkItem.resultData. However, all old PDG API ResultData methods will continue to work.

The pdg.utils.addItemFromSerialized(..) utility method is now deprecated. Please use the pdg.WorkItemHolder.addWorkItemFromJSONFile method instead.

The code used to produce wrapped ROP TOP assets is now available in the pdg.hda.ropwrapper module.

Query information about a node’s job script with the new pdg.Node.scriptInfo property.

Evaluate a PDG node parameter without applying back tick or $variable expansion with the new pdg.Port.evaluateRawString method.

Python scripts can now write the pdg.File.size property. It is no longer read-only.

The from, to, and scheduler pdg.PathMapEntry methods have been renamed to mapFrom, mapTo, and createdBy respectively.

Specify the name of the expanded value attribute that is added to expanded file list or partition work items with the new Value Attribute parameter on the Work Item Expand TOP node.

Import work items from .json files on disk or from other TOP nodes in your scene file (including any attribute data or output files listed on your source work items) with the new Work Item Import TOP node.

Boost the priority of a work item on a Remote Graph TOP node with the new RMB menu Boost Priority option for the work item.

Make attributes accessible to all the nodes in your TOP graph (/topnet) by storing their data globally with the new global attributes. You can create global attributes directly on your graph or bind them to specific work items, and you can access global attributes with either the @attribute syntax in a scene or by using the PDG Python API. For more information, see Global vs. Work Item Attributes.

Convert between output files, intrinsic fields on work items, and attributes with the new Attribute Promote TOP node. You can also use this node to set global graph attributes from work item attributes, or set work item attributes from global graph attributes.

Reduce array attributes per work item down to a single value with the new Attribute Reduce TOP node.

Set connection, timeout, and retry count options for Scheduler RPC endpoints with the new RPC Server parameters on all Scheduler nodes.

Specify the path to the node you want to cook when submitting a job with the new Output Node parameter on the HQueue Scheduler TOP, Deadline Scheduler TOP, and Tractor Scheduler TOP farm scheduler nodes.

Specify how long to wait for the Scheduler’s task files before timing out with the new Task File Timeout parameter on the Deadline Scheduler TOP node.

Prevent Houdini packages from being loaded into child processes created by the Local Scheduler with the new Skip Loading Packages parameter on the Local Scheduler TOP node. You can also add this parameter to individual nodes that use the Local Scheduler as a Schedulers tab ▸ TOP Scheduler Override.

HQueue Scheduler TOP node now automatically sets HOUDINI_MAXTHREADS so that it matches the current CPUs per Job parameter value unless the Max Threads parameter is explicitly set. However, if CPUs per Job is not set, then HOUDINI_MAXTHREADS will default to 0 and each farm jobs will be allowed to use all CPU cores.

HQueue Scheduler TOP node now automatically applies the single Job Parms > Tags to any jobs that do not have an explicitly set CPUs per Job parameter value.

Automatically retry child tasks when their parent tasks are retried manually with the new Auto retry downstream tasks parameter on the Tractor Scheduler node.

All Processor nodes that schedule work items now have a new Schedule When parameter. This parameter allows you to specify whether or not a given work item is scheduled, or use an expression to conditionally turn on/off the scheduling of a given work item.

Choose whether to cook your Processor work items in-process or out-of-process with the parameters listed below.

Cook your Processor work items using a service with the new Service cook type parameter option and Service Name parameter on the following nodes:

Use a custom Python script to filter render outputs and report them as work item output files with the new Output Filter parameter on the Render IFD TOP node.

For work items cooked in-process, you can now reuse existing .hda instances instead of creating/destroying fresh instances for each cook with the new Delete HDA Instance after Cooking parameter on the HDA Processor TOP node.

Use a specific operator type or the first definition found in a target .hda with the new Specify Operator Using > First Definition in Asset and Custom String parameter options and Operator Type parameter on the HDA Processor TOP node.

Rename upstream output files with the new Upstream Output File option for the File Source parameter on the File Rename TOP node.

Control how your archive files are compressed with the new compression controls on the File Compress TOP node.

Customize the labels for your Output TOP node work items with the new Work Item Label parameter and options.

OP Notify TOP node can now trigger a Reload button on its target TOP node when work items on that node cook.

Store matching files to a work item attribute instead of as work item output files with the new Output Attribute parameters on the File Pattern TOP node.

When other columns have a numeric suffix, the column headers produced by the CSV Output TOP node now include a numeric suffix for the 0th value of an attribute. Additionally, columns with a numeric suffix now also use the same format as the @attrib accessor. For example, writing out an attribute named scale with three components would produce the CSV column names: scale.0, scale.1 and scale.2.

Specify columns using an attribute pattern with the new Attribute Pattern parameter on the CSV Output TOP node.

Python Script TOP node now runs its script code during work item generation by default. You also still have the option to run the script code during the node’s in-process cook.

Specify whether or not a script parameter is evaluated with variable expansion or as a raw string with the new Expand Variables in Script parameter on the Python Script TOP node.

Select a cache mode for your scripts and specify where their output files will come from with the new Caching and Output Files parameters on the Python Script TOP node. These parameters allow you to cook work items from cache when their outputs are already found on disk.

When a Python Processor TOP node needs to read the contents of input files to be able to generate its work items, you can now turn on the new Requires Input Data parameter on the node to indicate that it requires access to input file data.

Control how attribute name conflicts should be resolved when loading work item .json data with the new On Attribute Collision parameter on the JSON Input TOP node.

Write partition contents to JSON as an array of work items with the new Write Partition Contents parameter on the JSON Output TOP node.

Preserve the array size of incoming attributes when they are merged by Partitioners with the new Preserve Arrays parameter on most Partitioner nodes.

The Merge Operations multiparm on Partitioner nodes is now produced by the pdg.mergeOperationMenu() function instead of manually specifying its settings in each Partitioner node’s parameter interface.

Sort nodes by name or input index with the new Sort Nodes By parameter on the Partition by Node TOP node.

Choose whether to cook your work items in-process, out-of-process, or using a service with the new Cook Type parameter options on all the ROP TOP nodes. You can use this for cooking both single frame and batch work items, but not for cooking ROPs in external .hip files or distributed sims.

Specify which service you want to use to cook your ROP TOP nodes with the new Service Name parameter.

Choose to ignore errors when work items load their scene files with the new Ignore Scene Load Errors parameter on all the ROP TOP nodes.

Start your batches only once a specific number or percentage of frames have cooked with the new Fixed Number of Frames are Ready and Percentage of Frames are Ready options for the Cook Batch When parameter on all the ROP TOP nodes.

Control whether or not work items should look for output log messages and specify a format string for those message with the new Output Log Parsing and Custom Log Format parameters all the ROP TOP nodes.

Generate work items that cook an FBX ROP and output .fbx files with the new ROP FBX Output TOP node.

Set-up your ROP Karma TOP nodes with greater ease with the new simplified Karma parameters. For more information, see the Solaris v19.5 What’s New.

Customize the labels for your ROP Karma TOP node work items with the new Work Item Label parameter and options.

Specify the path to a parameter that should be pressed before each frame cook with the new Reload Parm Path parameter on the the ROP Fetch TOP and ROP Geometry TOP nodes. This is helpful when cooking a ROP that references nodes that need to manually reload external resources with a Reload button.

ROP Fetch TOP node now sets its pre and post frame hooks on its target ROP using the hou.RopNode.addRenderEventCallback HOM API method instead of writing to the pre and post frame script parameters on the ROP node.