Plot measurement results

Turn object measurements into distributions, individual-point plots or scatter plots without leaving VIPP.

New in 0.16.0a1

Plot Results provides descriptive plots, not statistical tests. Opening a plot does not change an existing summary recipe.

Start with one image

1. Label the structures, then calculate Measure Objects or Measure Objects + Intensity.
2. Connect the resulting table to Plot Results, under Measurements → Tables.
3. Choose a Plot type and Measurement. For a scatter plot, also choose X measurement.
4. Leave Group by unset and Each point represents set to Objects when each table row describes one object in this image.
5. Calculate the node, then choose Open plot… for a larger editable window.

Your workflow stays open behind the plot window. The inspector and plot window edit the same saved settings; changing plot settings does not rerun the source image or its measurements. Save the workflow to keep your choices. Recalculate if the measurement table itself becomes stale.

For controls beside tables, summaries and figures in one window, choose Open Results Workspace…. Its Plots view edits the same Plot Results node. The top selection bar (Workflow → Data source → Statistics node → Plot) lets you activate any open workflow and browse its connected nodes without changing graph connections; choose None — use input data for plots directly connected to the data source. Tabs only change the view. Change plot input… opens controls to deliberately reconnect the selected plot; see Results Workspace before interpreting the plotted values.

Plot settings groups the measurement, grouping and observation choices. Appearance contains the title, axis and styling controls. Bold text marks these section headings; individual field labels use regular text. Its outlined chevron points right when collapsed and down when expanded, and stays readable in light and dark themes across the inspector, plot window and Results Workspace.

The node's optional Name is for navigating the workflow and Results Workspace. It does not replace the figure's Title under Appearance. Without a custom name, VIPP uses a meaningful figure title or derives a label from the selected plot settings. See Name a node.

In Results Workspace, the figure fits the available preview area while the settings scroll independently. Use the title-bar Maximize button, resize the window or drag the divider for a larger preview. This changes only the on-screen view, not the data, saved plot recipe or exported figure size.

While an update is queued, preparing measurements or calculating, the inspector and plot window show an activity indicator and the current stage, not an estimated percentage. Any previous drawing is out of date and cannot be exported until the updated plot is ready. The activity row stays reserved when its indicator is hidden, so starting and finishing an update does not shift the plot or controls.

Plot type	Use it to	Useful choices
Compare groups	See individual measurements, with or without groups	Mean, median or no summary line; each visible point remains an observation.
Distribution	See which values are common or how many fall below a value	Histogram with shared bins, or Cumulative; histogram count or percentage.
Scatter	Compare two measurements, such as area and mean intensity	X and Y measurement; optional group colours. No fitted line or significance test is added.

Axis titles use readable measurement names and show the declared unit once. Long numeric tick labels are shortened for display without changing their values. Category labels wrap, rotate or spread out to fit the window or export; a note appears if some labels must be omitted. All groups remain plotted. Click a point or choose View plotted data… to inspect exact group values.

Histogram Count axes and object measurements explicitly recorded in count units, such as branch counts, use whole-number ticks. Percentages, cumulative proportions and other measurements retain decimals where useful; Mean per image can be fractional even when the original measurements are counts.

Set grid and axis-label spacing

Under Appearance, leave X axis label interval and Y axis label interval on Auto to let VIPP choose readable spacing. Uncheck Auto and enter a positive interval for evenly spaced numeric labels and matching grid lines: for example, 100 for area ticks at 100-unit steps or 2 for count ticks at two-count steps.

This changes the axis markings, not histogram bins, measurements or summary values. The same saved interval is used in the inspector, plot window, Results Workspace and exported figures.

• Count axes require whole-number intervals of at least 1.
• The Compare groups horizontal axis is categorical, so numeric X intervals do not apply. Category label spacing remains automatic.
• Logarithmic axes use Auto; switch the corresponding interval to Auto before using a log scale. A saved incompatible interval is reported rather than silently replaced.
• If a custom interval would make the axis too dense, increase it or use Auto. VIPP does not silently substitute a different custom interval.

Missing, nonnumeric and non-finite measurements are not treated as zero; review the included/excluded counts. Logarithmic axes cannot show zero or negative values, and report those exclusions. They do not change the original measurements.

New in 0.16.0a2: logarithmic axis labels

Power-of-ten labels on logarithmic axes render with superscripts, such as 10², in the inspector, plot window, Results Workspace and exported figures.

Choose categories or a numeric axis

Compare groups treats every distinct Group by value as a separate, equally spaced category—even when the values are numbers. VIPP never silently bins or merges those values, or changes your plot type. More than 12 numeric groups triggers a warning to review this choice.

Read the highlighted warning beside the plot before interpreting it. Warning text adapts to light/dark themes and wraps to the inspector width, expanding vertically so the full message remains available.

To compare two numeric measurements, choose Scatter, put the numeric field in X measurement, and leave Group by unset or choose a category such as treatment. This makes horizontal distances represent numerical differences.

Try the guided example

Open Morphology & Intensity Plots in the example chooser. It contains one image with 60 separated synthetic ellipses at 0.5 micrometer per pixel. Thresholding and labelling lead to shape and intensity tables, joined by label ID, followed by four Plot Results nodes:

• Area distribution: inspect the range of object areas and try a different number of histogram bins.
• Area–intensity scatter: larger objects were deliberately made brighter. Change Y to eccentricity to explore shape instead of intensity.
• Elongation points: compare individual major-to-minor axis ratios with the median line. A ratio near 1 describes a more rounded ellipse.
• Cumulative circularity: read the fraction of objects at or below a chosen circularity value, without choosing histogram bins.

Click a point to inspect its source label/image identity where available. View plotted data… opens the prepared values and source-row references. For image means, one point can refer to several source rows.

These are invented demonstration data. The designed size–brightness relation is not a biological finding, and 60 objects in one image are not automatically 60 independent biological samples.

Use several images or conditions

For a completed batch, collect the measurements, add condition/image annotations as needed, save a VIPP collection and open it through Table Source. Connect that table to the same Plot Results node.

Choose Group by to compare conditions. Each point represents changes the observational level:

• Objects: every eligible row contributes. Images containing more objects contribute more points to the pooled distribution and its summary.
• Mean per image: choose an Image identity column. VIPP first averages eligible values within each image, then plots one point per image. Image means receive equal weight in the displayed group summary.

Image means use the finite measurements before checking whether each mean can appear on a logarithmic axis. A negative contributor is not silently dropped to make the mean positive; non-positive image means are excluded from a log view and reported separately.

Use globally distinguishing image IDs: a local object label is not an image identity. Different groups cannot silently share one image ID. With a single image, a mean-per-image plot naturally contains only one point.

An image is not necessarily an independent experimental sample either. The right independent unit depends on how the experiment was sampled or treated. Plot Results does not infer biological replication or perform statistical tests. An empty image contributes no invented zero-valued object or mean; review the collection's image summary separately for missing or empty images.

If image means cannot be calculated

Mean per image needs one group per image. A Group by measurement such as minor axis length can vary between objects in that image, so it cannot define the image's group.

• For individual objects, set Each point represents to Objects.
• For image means, set Group by to None, or choose an image-level category such as treatment that is the same for every object in each image.
• If different images reuse an ID, choose an Image identity column that uniquely identifies each image.

The inspector and plot window show the problem beside the controls. Choose Review plot… to reopen a failed plot; the compact workflow status offers Details… for the full message. Correct the settings and calculate again. VIPP does not silently regroup objects or change the averaging rule, and an outdated plot stays unavailable for export.

Export a figure

Choose Export figure… from a current plot. Set Width and Height in millimeters, independently of the on-screen window size.

• PNG/TIFF: choose the raster resolution in DPI.
• SVG/PDF: save a vector figure for layout/editing software.
• Optionally include a CSV of plotted values and a JSON settings sidecar to describe the selected fields, grouping and preparation.

Exports use a light publication background. Calculating or opening a plot does not save files; export requires an explicit destination. A stale plot cannot be exported as a current result. Keep the original measurement table and workflow alongside any figure you share.

For a reusable descriptive summary table, use Statistics with explicit object/image/sample aggregation. Paired/time-course views and object-count/fraction plots are not included here. Confidence intervals and inferential tests are outside the 0.16 results scope; they are not a promised next phase.