Skip to main content
Version: 3.4.20 (unreleased)

Charts, Histograms and Filtering

The Charts panel plots whatever attribute is currently active, and it is the quickest way to answer three questions about a model you have just built or imported: what is the range of this attribute, where does most of the data sit, and which parts of the model fall inside the range I care about.

It is not a passive display. Dragging on a histogram filters the attribute, and the 3D view follows immediately.

Opening the panel​

On the Home ribbon tab, in the Windows group, click Charts Tab. The panel docks like any other; drag its title bar to move or float it.

The chart always follows the active attribute on the active object. Make an attribute active in the Data Tree and the chart redraws for it — there is no separate data picker to keep in step. Select two or three attribute rows and it plots them against each other; see Cross-plots.

Choosing a chart​

A toolbar sits across the top of the panel with the common chart types on it; the right-click menu carries the rest, plus every display option below.

ChartBest for
HistogramA continuous attribute — intensity, roughness, dip, height, reprojection error. The default, and the one that filters.
Pie chartA discrete attribute — facies codes, classification results, anything with a small set of values. Shows the proportion each value takes.
Bar chartComparing values across categories.
Scatter (first two selected attributes)How two attributes vary together, one point per element. See Cross-plots.
Scatter coloured by third attributeThe same plot with a third attribute as the colour of each point.
Triangular diagram (first three attributes)The proportions of three components.
StereonetOrientation data — poles, planes, contours and a rose diagram, mirroring the Stereonet window so you can keep it beside the model.
Mohr diagramHow the planes on the stereonet sit under a stress tensor: normal against shear stress, with the failure envelope. From the stereonet's right-click menu; see Mohr diagram. Geomechanics licence feature.
VariogramThe directional semivariograms of a Variogram 3D. See Variogram chart.
The stereonet in this panel is the stereonet

It reads the project's stereonet display options, shared with the full Stereonet window: the contour method, the poles, the cardinal and degree labels and the regional orientation. A change to one of those appears in both. What each view shows is its own: planes, minor lines, whether contours and the rose are on, the rose shape, and whether the regional correction is applied. The Stereonet User Guide covers all of it; this page covers the attribute charts.

Reading and shaping the histogram​

Right-click the chart for the options. The ones that change what you conclude:

OptionWhat it does
BinsA fixed bin count (16 to 256), or Auto (Sturges) / Auto (Freedman-Diaconis). Sturges assumes a roughly normal distribution and gives few, wide bins; Freedman–Diaconis works from the interquartile range and copes far better with the long tails and outliers that outcrop attributes usually have. If a distribution looks suspiciously smooth, change the bin rule before believing it.
Y AxisCount, Log Count, Percent of Total or Density. Log count is the one that reveals a small but real population hiding beside a large one — the handful of high-error points among a million good ones.
CumulativePlots the running total instead of the per-bin count. Read a percentile straight off it.
Density Curve (KDE)Overlays a smooth kernel density estimate, which is independent of the bin choice — a useful check that a peak is real and not an artefact of the bins.
Statistics ▸ Show Statistics, Mean Line, Median LineThe summary figures, and vertical markers on the plot. Mean far from median means a skewed attribute, which is worth knowing before you set a threshold on it.
Include No DataWhether the bin holding the attribute's no-data value is drawn and counted. On by default, and greyed unless that value falls inside the attribute's range, which it usually does not.
Include No Data SliceOn the pie chart, whether the elements with no value get a slice of their own and count towards the percentages. On by default. It has no effect on the histogram, whose menu also lists it.
Gridlines, Titles & Axis Labels, SolidPresentation, for when the chart is going into a report.

Reset Chart Options puts all of this back to the defaults, and Reset View (Zoom & Pan) undoes navigation without touching the settings.

Filtering from the histogram​

This is the part worth knowing about.

The histogram carries two handles, one at each end of the range. Drag one and you set the minimum or maximum of the attribute's filter, and the 3D view redraws as you go. What "filtered" looks like depends on the object and on the attribute's Use Filter property:

  • On a mesh with Use Filter on, the attribute colours are confined to the window and the rest of the surface shows its photo texture (or its plain mesh colour), so you see the kept range in context on the outcrop. With Use Filter off the whole surface stays coloured.
  • On a point cloud with Use Filter on, the points outside the window are hidden.
  • With Scale Colours to Filter on, which is the default, the ramp is stretched across the window and values beyond it clamp to the end colours. Turn it off to keep the ramp over the full range while you filter, so a colour means the same value however the window moves.

Shade Filtered Range marks the kept range on the chart so you can see what you are keeping while you drag.

It is the fastest way to answer "where in the model is this?" — set the range on the distribution, look at the outcrop, adjust. A few examples:

  • Isolate the steep faces by dragging the low end of a dip histogram up.
  • Find the poorly resolved parts of a photo-textured mesh by keeping only the coarse end of Texture GSD.
  • Strip noise out of a comparison by keeping the middle of an M3C2 distance distribution and seeing what survives.

Reset Filter on the toolbar clears it and brings the whole attribute back. The filter belongs to the attribute's colour map, so it is the same filter you can set in the object's properties — the chart is simply a much faster way to drive it.

A filter changes what you see, not what you have

Filtering hides data or takes the colour off it; it does not delete it, and it does not alter the attribute. Clear the filter before you judge a result or export anything, or reach for Reset Filter if a model looks emptier than it should.

Cross-plots​

Select more than one attribute row and the panel plots them against each other. Click the first row, then hold Ctrl and click the others, in the Data tree (the layers of a mesh or point cloud, or the attributes under Samples) or the Interpretation tree (the attributes under Orientation or Polylines).

Rows selectedThe panel opens onAlso on the right-click menu
OneThe histogram—
TwoScatter (first two selected attributes)Histogram (first selected attribute)
Three or moreScatter coloured by third attributeScatter (first two selected attributes), Triangular diagram (first three attributes), Histogram (first selected attribute)

The order you click sets the axes. The first row is X, the second Y and the third the colour. A row you deselect and select again goes to the end of the order, and rows added together with Shift take their order from the tree. With more rows selected than the plot can use, it takes the first two or three and its note says so.

What can be plotted together. Values are paired element by element, so the attributes must be numeric, belong to the same mesh, point cloud or Attributes node, be all per vertex or all per triangle, and have the same number of values. Otherwise the panel shows Choose numeric attributes on the same owner and element domain, with equal lengths in place of the plot. Vector and tensor layers and string attributes cannot be plotted.

Missing values. An element with a missing or non-finite value in any of the plotted attributes is left out whole. The note under the plot counts them, as in 1200 plotted; 34 omitted.

Point sampling. A cross-plot draws every element until you tell it otherwise. For a large mesh or cloud, right-click ▸ Point sampling and choose Every 2nd point, Every 5th point and so on up to Every 1000th point; All points goes back. The sample starts at the first element and takes the same elements from every attribute, so missing values do not shift it. The axis and colour ranges come from the sample and the note states the stride. The setting carries over when you change the selection and applies to the SVG export, so choose All points for a final figure.

Scatter plots​

The axes span the values plotted. On Scatter coloured by third attribute each point runs from blue at the third attribute's lowest value to red at its highest, with a labelled scale beside the plot. This is the chart's own ramp, not the attribute's colour map.

Log X Axis and Log Y Axis on the right-click menu make either axis base-10 logarithmic, which is how Length against Height fracture plots are read. A log axis runs over whole decades around the data and its label says (log scale). Minor ticks mark 2 to 9 in each decade on an axis of up to six decades; the decade labels thin out on a wider one. A value at or below zero cannot be shown, so that element is omitted and the note counts it separately. If nothing is above zero, the panel says so in place of the plot.

Show Statistics, on the same menu and on by default, lists above the plot the number of points, the correlation coefficient r, r², the covariance, and each attribute's mean and standard deviation. On a log axis they are of the log10 values and the text says so, because that is what the plot shows: on a log-log plot, r is how straight the cloud is. They describe the points plotted, so point sampling changes them.

Best Fit Line, on the same menu, draws a least-squares line through the points and gives its equation above the plot:

LineEquationFitted where it is straight, inLeaves out
Lineary = a + b xx and y—
Power (Straight on Log-Log Axes)y = a x^blog10 x and log10 yx or y at or below zero
Exponentialy = a e^(b x)x and log10 yy at or below zero
Logarithmicy = a + b ln xln x and yx at or below zero

R² and the scatter are measured in the same space as the fit, so a Power or Exponential fit's scatter is the standard deviation of log10 y about the line, in log10 units. The fit does not change with the axes: a Power fit is straight on log-log axes and a curve on linear ones. The line stops at the first and last point it used, and any part outside the plot is cut off. Points it left out are counted beside the equation.

A Power fit of Length against Height gives the three numbers a DFN set takes to draw each fracture's height from its length; see Fracture shape.

Triangular diagram​

The first attribute is the left corner, the second the right and the third the top, with the gridlines every 20%. Each element is divided by its own sum, so percentages, fractions or any other common scale land in the same place. The components are not scaled by their ranges, so they must be in comparable units. An element with a negative component or a total of zero is omitted. The diagram has no log axes.

Getting a cross-plot out​

Copy to Clipboard and Export SVG Document... are on the right-click menu. The SVG is built from the same plot, with its sampling, log axes, statistics and best-fit line, at a fixed 800 by 500.

A cross-plot has no filter handles, and the wheel and middle button do not zoom or pan it. To filter, go back with Histogram (first selected attribute) or the Histogram toolbar button; the histogram shows the first selected row.

Regional correction on the panel's stereonet​

The stereonet's right-click menu has a Regional Control section:

  • Correct Regional rotates what this view plots by the project's regional orientation, as Remove Regional Dip does in the Stereonet window. It is on or off for this view alone.
  • Set Regional here takes the dip and azimuth under the cursor as the project's regional orientation and turns Correct Regional on. The orientation belongs to the project, so every stereonet uses it.

Variogram chart​

Selecting a Variogram 3D, or a mesh or point cloud one was built on, opens the panel on its chart: semivariance against lag distance for each axis, the fitted model through them, the sill, and a legend giving each range. A lag with too few pairs is drawn hollow. Selecting something with no variogram returns to the chart that was showing before.

CommandWhat it does
Major, Intermediate, Minor, OmnidirectionalShow or hide that direction.
Model CurvesShow or hide the fitted model.
Show HistogramBack to the chart that was showing before. Variogram on the histogram's menu returns here.
Copy to ClipboardThe chart as an image.
Reset View (Zoom & Pan)The whole chart again.

Mohr diagram​

Right-click the panel's stereonet ▸ Mohr Diagram to plot the same planes on a Mohr circle diagram. Every plane the stereonet draws counts — orientation planes, faults, log picks and scan-line fractures; lineations have no plane and are left out.

  • The three circles are those of the principal stresses: σ1–σ3 (bold), σ1–σ2 and σ2–σ3. Each plane's point lies between them.
  • Stresses are effective: the tensor's pore pressure is subtracted from every normal stress, and the x axis says so. With a pore pressure of 0 they are the total stresses. A tensile effective σ3 extends the axis left of zero.
  • The red line is the Coulomb failure envelope, τ = C0 + μσn′, from the tensor's cohesion and friction coefficient.
  • Each point is a plane, in the colour the stereonet draws it (its Colour by colour), ringed red when it is on or above the envelope — critically stressed. The legend counts them. A point's stresses are the same numbers the Stress Attributes command stores on an orientation.
  • Both axes share one scale, so the circles are round however the panel is shaped.

Right-click the diagram:

CommandWhat it does
Stress TensorWhich stress tensor to plot: Active Tensor, which follows Set as Active, or any tensor by name. Shown when the project has more than one.
Show StereonetBack to the stereonet.
SVG Document…The diagram as a vector file, exactly as drawn here: circles, envelope, points, labels and legend.
Copy to ClipboardThe diagram as an image.
Reset View (Zoom & Pan)The whole diagram again.

To put it in an analysis report, use the report's Figure ▸ Mohr Diagram: the same picture as the SVG export, rebuilt from the panel's selection and tensor when you insert it.

The Mohr diagram needs the Geomechanics licence feature; without it Mohr Diagram says so instead.

  • Mouse wheel — zoom.
  • Middle button drag — pan.
  • Reset View (Zoom & Pan) from the right-click menu returns to the whole range.

The stereonet remembers its own zoom and pan when you switch chart types and come back, rather than resetting.

Getting a figure or the numbers out​

Right-click ▸ Export:

ChoiceGives you
Copy to ClipboardThe chart as an image, for pasting straight into a document or an analysis report.
SVG Document…A vector file that stays sharp at any size and can be edited in Illustrator or Inkscape. Use this for anything going into print.
CSV Data…The binned numbers themselves — bin edges and counts — for your own plotting or statistics.

Export is greyed on the pie chart; Copy to Clipboard still works. The stereonet and rose diagram have their own SVG Document entries, under the stereonet part of the menu, and cross-plots have their own export.

Charts belong in reports

An analysis report can take a figure captured live from the Charts panel, so the histogram you used to make a decision can be the one that appears in the write-up.

See also​