Skip to main content

Data Panel

The SPLAT DATA panel provides statistical visualization and analysis tools for understanding the distribution of data properties across your Gaussian splats. It displays an interactive histogram that allows you to view and select splats based on their attribute values.

SPLAT DATA Panel

Overview

The panel consists of three main sections:

  1. Data Selector - Choose which splat attribute to visualize
  2. Histogram Display - Interactive visualization of the selected data distribution
  3. Statistics - Total counts for splat states

The panel is located at the bottom of the interface and is collapsed by default.

tip

The panel can be toggled open/closed by pressing the D key or by clicking the panel header.

Data Selector

The dropdown menu allows you to select from various splat properties for visualization:

PropertyCategoryDescriptionRange/Calculation
XPositionPosition coordinate along the X axis-
YPositionPosition coordinate along the Y axis-
ZPositionPosition coordinate along the Z axis-
Scale XScaleSize of the Gaussian along the X axis-
Scale YScaleSize of the Gaussian along the Y axis-
Scale ZScaleSize of the Gaussian along the Z axis-
RedColorRed channel intensity0-1
GreenColorGreen channel intensity0-1
BlueColorBlue channel intensity0-1
HueColorColor hue0-360°
SaturationColorColor saturation0-1
ValueColorColor brightness/value0-1
OpacityColorTransparency of each splat0-1
DistanceDerivedDistance from the origin (0,0,0)√(x² + y² + z²)
VolumeDerivedCalculated volume of each Gaussianscale_x × scale_y × scale_z
Surface AreaDerivedApproximate surface areascale_x² + scale_y² + scale_z²

Histogram Visualization

The histogram displays the distribution of the selected property across all non-deleted splats in the scene, with blue bars representing unselected splats and yellow bars representing selected splats. Locked and deleted splats are not shown in the histogram.

Interactive Features

Hover Information - Hover over any histogram bar to see a tooltip displaying:

  • value - The data value for this bin
  • cnt - Total count of splats in this bin
  • percentage - Percentage of total splats
  • sel - Number of selected splats in this bin

Selection by Range - Click and drag on the histogram to highlight a range of values. A dashed yellow rectangle will appear showing your selection. Release to select all splats within that value range. Note that locked and deleted splats cannot be selected through the histogram.

Use modifier keys to control the selection operation:

Modifier KeyActionDescription
NoneNew SelectionReplaces the current selection with the histogram range
ShiftAdd to SelectionAdds the histogram range to the existing selection
CtrlRemove from SelectionRemoves the histogram range from the existing selection

Log Scale - Enable the Log Scale checkbox to view the histogram with a logarithmic Y-axis. This is useful when data has a wide range of values, when most splats cluster around certain values with few outliers, or when you want to better visualize the distribution of sparse data.

Statistics (Totals)

The bottom section displays real-time statistics about the current splat:

StatisticDescription
SplatsTotal number of non-deleted splats
SelectedNumber of currently selected splats
LockedNumber of locked splats
DeletedNumber of deleted splats

These values update automatically when splats are selected, deselected, locked, unlocked, deleted, or reset.

Use Cases

Finding Outliers - Select a property like "Distance" or "Opacity", look for isolated bars at the extremes, then click and drag to select those ranges for examination or deletion.

Color-Based Selection - Choose "Hue", "Saturation", or "Value" and select a range to isolate splats of similar colors, useful for separating objects by color in the scene.

Size-Based Selection - Select "Volume" or "Surface Area", enable "Log Scale" for better visualization, then select ranges to find problematic splats for cleanup or adjustment.

Quality Control - Review the distribution of properties to check if opacity is reasonable, verify scale values are within expected ranges, and identify areas that may need cleanup.

Tips

  • Keep the panel collapsed when not in use to improve editor performance
  • Check different properties to understand your data better
  • Use histogram selection together with other selection tools for precise control
  • Understanding your splat distribution helps with optimization and quality improvements

Technical Notes

Data Transformations

Some properties are transformed for visualization to ensure the histogram displays human-readable values rather than the internal storage format:

Property TypeTransformation
Scale valuesExponentiated: exp(value)
Color values (RGB)Converted from spherical harmonics: 0.5 + value × 0.28209479177387814
OpacitySigmoid function: 1 / (1 + exp(-value))

Suppressed Properties

The following properties are not available for histogram visualization: state (internal state flag), transform (internal transformation data), and f_rest_0 through f_rest_44 (higher-order spherical harmonic coefficients).