Work with scalar fields

A scalar field is one number per point. Some arrive with the file — intensity, reflectance, GPS time, a return index — and others are written by Phytograph's own tools: curvature and verticality from Compute normals, ground_class from Segment ground, height_above_ground from Generate a DEM, wood_class, tree_instance, noise_class.

The Scalar Fields tool (Tools → Scalar Fields, or ++cmd+k++ → "Scalar Fields") lets you inspect those numbers, derive new fields from them with a formula, and tidy up the ones you no longer need.

A field you create here is an ordinary scalar field. It appears in the Color by dropdown, in the Filter panel's field list, and in the export column picker, exactly like one that came from the file.

Work on several clouds at once

The panel has its own cloud list at the top. It starts out matching whatever you have selected in the scene, but from then on it is the tool's own input: checking and unchecking there changes what the tool measures and writes to, without disturbing your selection in the viewer.

With more than one cloud checked:

• Stats pools them into one distribution — the clouds' points are measured together, not side by side.
• Compute runs your formula on each checked cloud in turn, so each ends up with its own copy of the new field.
• Rename, duplicate and delete likewise apply to every checked cloud.

The field list shows only the fields every checked cloud carries. If one cloud has a field the others lack, it is not offered, and a note under the list says how many fields were hidden and why — acting on a field only some clouds have would half-apply and leave them out of step.

Coordinates are measured one cloud at a time

x, y and z cannot be pooled. Each cloud stores its coordinates in its own frame — a large global offset is subtracted when the cloud is imported, so two clouds imported at different offsets hold coordinate numbers that do not share an origin. A pooled mean or percentile over them would be arithmetic across two different frames, and the result looks like a perfectly ordinary number, so Phytograph declines it rather than warning about it. The rows stay visible, marked per cloud only. Check a single cloud to measure its coordinates.

If something fails on one cloud partway through, the others still finish, and the panel names the clouds that failed rather than reporting one opaque error.

Inspect a field

Check one or more clouds, open the tool, and go to the Stats tab. Pick a field and Phytograph reports its distribution:

Points	How many points the statistics cover
Mean, Std dev	Average and spread
Min, Max	The true extremes, outliers included
Median, percentiles	5th, 25th, 75th and 95th

Above the table is a histogram. Hover a bar to read its value range and point count.

What the numbers are measured over

Statistics cover the points that are visible and are real returns — hidden points and sky/miss points are excluded. That is the same population the color scale uses, so the numbers always agree with the legend beside them. A sky/miss point is a laser pulse that hit nothing, recorded about a kilometer out along the beam; including those would make a mean or a histogram meaningless.

The panel says so under the table whenever the two counts differ.

With several clouds checked, each cloud's own hidden and sky/miss points are excluded before the clouds are pooled, and the caption says how many clouds the distribution covers.

Class columns pool differently from measurements

A pooled histogram of a class column (wood_class, tree_instance) is useful — it shows the class mix across the plot. A pooled mean or percentile over one is not: class ids are labels, and tree_instance numbers objects per cloud, so the same id means a different tree on each. The panel flags this beside the numbers; read the histogram instead.

The histogram is binned over the 1st–99th percentile rather than the full range, so a single spike cannot squash every real value into the first bar. Points outside that span are still counted, and the caption under the chart says how many fall above or below.

Derive a new field

Go to the Compute tab, type a formula, give the result a name, and click Compute Field. Reuse the name of a field you derived earlier and the button reads Recompute Field — editing a formula and running it again replaces the field in place, so you do not have to delete it first. An imported or tool-written column can never be replaced this way; those are measurements, and overwriting one would invalidate anything already computed from it.

intensity * 2
sqrt(x**2 + y**2)
z - height_above_ground
degrees(acos(nz))

A formula may also be a bare constant (0, pi * 2), which fills every point with the same value — useful for seeding a column you then edit by hand.

Refer to any field by its name. Show available names lists everything the checked clouds offer, including x, y and z.

With several clouds checked the formula runs on each in turn, and the names you may use are the ones all of them carry — otherwise the formula would work on the first cloud and fail partway down the list. Coordinates stay available here even when they are unavailable in Stats: each cloud is computed on its own, so z - height_above_ground never mixes frames.

Operators and functions

Arithmetic	+ - * / // % **
Comparison	> < >= <= == !=, and and / or / not
Maths	sqrt abs exp log log10 log2 floor ceil round sign
Trigonometry	sin cos tan asin acos atan atan2 degrees radians
Pick a value	min(a, b) max(a, b) clamp(v, lo, hi) ifelse(test, a, b)
Constants	pi e nan inf

A comparison produces 1 where it holds and 0 where it does not, so ifelse(curvature > 0.1, 1, 2) builds a two-class column you can then color by or filter on.

Whole-field values

mean, std, median, sum, count and percentile take a whole field and give back one number, which you can use inside a larger formula:

(intensity - mean(intensity)) / std(intensity)
intensity / percentile(intensity, 95)

The first is a z-score — the field rescaled to mean 0 and standard deviation 1, which makes two scans with different gain settings directly comparable.

These are measured over the same visible, real-return points the Stats tab reports, so mean(intensity) is the mean you can see there.

When a value cannot be computed

Dividing by zero or taking the log of a negative gives infinity or NaN rather than failing the whole field. Phytograph counts them, says so in the toast, and excludes them from the statistics. They are also excluded from the color scale, so those points render at the end of the ramp.

Rename, duplicate and delete

The Fields tab lists every field on the cloud. Hover a row for its menu:

• Rename — the field keeps its values, and the viewer keeps coloring by it. Type the name you want to see; it appears as typed in this list, the Color by picker and the Scans panel. Formulas need a plain identifier, so a name like Height [m] is used in formulas as Height_m, and the box shows that before you confirm.
• Duplicate — an independent copy, useful before an edit you may want to undo.
• Delete — removes the column from the cloud.

Some fields are marked locked and have no menu. Other tools read those by name — is_miss is what leaf area density uses to know which pulses returned nothing, and the class columns are what the segmentation tools and the color palettes look for — so renaming or removing one would quietly break them. x, y, z and intensity are marked built-in: you can use them in a formula, but they are not columns to be renamed.

Rename and Duplicate open a small name box in the row; Delete asks for confirmation there too.

Deleting a field cannot be undone

Point-cloud edits are not kept in the undo history (they are far too large), so a deleted column is gone. Duplicate it first if you are unsure.

Export a derived field

Derived fields appear in the export column picker like any other. See Import & export.

• LAS/LAZ — written as an extra dimension under its own name.
• CSV/TXT/XYZ/ASC — written as a named column.
• PLY — written as property float <name>.

Field names are limited to letters, digits and underscores, and must start with a letter or underscore, because the name has to survive as a LAS dimension name and an ASCII column header. Phytograph also refuses a name that import would read back as a different column — time, elevation, row, target_index and similar, in any capitalization — since a field called Time would come back as the GPS time column. reflectance is the exception: naming a column that says what it is, and a round trip reads it back as the cloud's reflectance.

Large clouds

On a cloud over about 5 million points the new column is written immediately but the display catches up in the background: the field is usable for export, for another formula and for every other tool right away, while the recoloring finishes shortly after.

Rename and delete are instant at any size: they change only the field's name in the display data, not the points. Duplicate adds a column, so it catches up in the background like a new field. The exception is a cloud that already has a display refresh pending (for example after an erase you haven't applied yet); there, rename and delete wait for that refresh too.