unreal.PCGClusterSettings

class unreal.PCGClusterSettings(outer: Object | None = None, name: Name | str = 'None')

Bases: PCGSettings

Given a desired number of clusters (categories), find the best fit cluster for each point by distance, using one of various clustering algorithms.

C++ Source:

  • Plugin: PCG

  • Module: PCG

  • File: PCGClusterElement.h

Editor Properties: (see get_editor_property/set_editor_property)

  • algorithm (PCGClusterAlgorithm): [Read-Write] Mathematical algorithm for selecting clusters.

  • break_debugger (bool): [Read-Write] If a debugger is attached, a breakpoint will be triggered in the execution code to enable debugging. Only applies when inspecting a debug object. Transient and Editor-only.

  • category (Text): [Read-Write]

  • cluster_attribute (PCGAttributePropertyOutputNoSourceSelector): [Read-Write] Cluster IDs will be written to this attribute on the main output.

  • debug (bool): [Read-Write]

  • debug_buffer_size (int32): [Read-Write] Size (in number of floats) of the shader debug print buffer.

  • debug_settings (PCGDebugVisualizationSettings): [Read-Write]

  • description (Text): [Read-Write]

  • determinism_settings (PCGDeterminismSettings): [Read-Write]

  • dump_cooked_hlsl (bool): [Read-Write] Dump the cooked HLSL into the log after it is generated.

  • dump_data_descriptions (bool): [Read-Write] Dump the data descriptions of input/output pins to the log.

  • enabled (bool): [Read-Write]

  • execute_on_gpu (bool): [Read-Write] Whether this node should be executed on the GPU.

  • execution_dependency_required (bool): [Read-Write] If enabled, the execution dependency pin will require to be connected for execution (as any other required pin) and will be used for culling.

  • expose_to_library (bool): [Read-Write]

  • final_centroid_element_count_attribute (PCGAttributePropertyOutputNoSourceSelector): [Read-Write] Final centroid element count will be written to this attribute on the Final Centroid output.

  • max_iterations (int32): [Read-Write] A limit on the number of iterations to run on each algorithm, if it doesn’t otherwise converge. A higher number can offer more accuracy at the cost of processing time.

  • num_clusters (int32): [Read-Write] Number of clusters (segments) to group the points into. Each point will be assigned a cluster at the end.

  • output_final_centroid_element_count (bool): [Read-Write] Output the element count for each of the final centroids.

  • output_final_centroids (bool): [Read-Write] Output the final location of the centroids or gaussians.

  • print_shader_debug_values (bool): [Read-Write] Enable use of ‘WriteDebugValue(uint Index, float Value)’ function in your kernel. Allows you to write float values to a buffer for logging on the CPU.

  • profile_kernel_index (int32): [Read-Write] Index of kernel emitted by this node to repeatedly dispatch every frame to enable profiling. Set to -1 to disable profiling. PCG_GPU_KERNEL_PROFILING must be defined in PCG.Build.cs.

  • seed (int32): [Read-Write]

  • tolerance (double): [Read-Write] For EM, the maximum allowed difference between the last two iterations’ “Log Likelihood”–which converges from positive infinity to 0 in relation to point-to-cluster probabilities. It is exponential, so raising this number can offer faster iteration if exact precision isn’t needed.

  • trigger_render_capture (bool): [Read-Write] Will trigger a render capture when this node executes and a debug object is selected in the graph editor. Transient and Editor-only. Render captures must be enabled (e.g. -AttachRenderDoc or -AttachPIX).

  • use_seed (bool): [Read-Write] deprecated: Implement the PCGSettings virtual UseSeed() override.

property algorithm: PCGClusterAlgorithm

[Read-Write] Mathematical algorithm for selecting clusters.

Type:

(PCGClusterAlgorithm)

property cluster_attribute: PCGAttributePropertyOutputNoSourceSelector

[Read-Write] Cluster IDs will be written to this attribute on the main output.

Type:

(PCGAttributePropertyOutputNoSourceSelector)

property final_centroid_element_count_attribute: PCGAttributePropertyOutputNoSourceSelector

[Read-Write] Final centroid element count will be written to this attribute on the Final Centroid output.

Type:

(PCGAttributePropertyOutputNoSourceSelector)

property max_iterations: int

[Read-Write] A limit on the number of iterations to run on each algorithm, if it doesn’t otherwise converge. A higher number can offer more accuracy at the cost of processing time.

Type:

(int32)

property num_clusters: int

[Read-Write] Number of clusters (segments) to group the points into. Each point will be assigned a cluster at the end.

Type:

(int32)

property output_final_centroid_element_count: bool

[Read-Write] Output the element count for each of the final centroids.

Type:

(bool)

property output_final_centroids: bool

[Read-Write] Output the final location of the centroids or gaussians.

Type:

(bool)

property tolerance: float

[Read-Write] For EM, the maximum allowed difference between the last two iterations’ “Log Likelihood”–which converges from positive infinity to 0 in relation to point-to-cluster probabilities. It is exponential, so raising this number can offer faster iteration if exact precision isn’t needed.

Type:

(double)