How-to 01 · Dry Session 1 and 2

The frame check in Fiji (ImageJ)

Four numbers per frame — the apparent width of each chart, in pixels — and one ratio. It takes a couple of minutes a frame once you have done it twice.

What you need

Fiji. Free, public domain, runs on Windows, macOS and Linux with no installation beyond unzipping. Download it from imagej.net/software/fiji/downloads. Fiji is ImageJ with the common plugins bundled; plain ImageJ works too.
The camera's files for the photo point — every frame in the capture set, including the ones you suspect are unusable. Raw files open through Fiji's Bio-Formats importer; JPEGs open directly. Work on a copy if you like, but you are only measuring, never saving over the record.
Slate B and Section 3 of the Site Entry, where the widths and the ratio go.

First, the usable-frame test

Before measuring anything, open each frame and answer the only two questions that can make a frame unusable:

Is any chart missing from the frame or unreadable? Cut off at the edge, buried in sediment, flipped. If so, the frame is unusable — log it with the reason and move on.
Is the white patch clipped? Hover the cursor over the white panel of each chart and read the pixel value in the Fiji status bar (it shows value= as you move). For an 8-bit JPEG, values at or near 255 across the panel mean it clipped; the histogram (Analyze → Histogram) will show a spike at the right-hand edge. For a raw file the ceiling is the sensor's maximum, and the importer reports the bit depth. Clipped means unusable — log it and move on.

Everything else — a chart on a bump, a visible tilt, a slightly soft focus — is reported through the ratio, not rejected. Those are the course's two faults with an unambiguous answer, and the only two.

Measuring a chart width

Open the frame. File → Open, or drag the file onto the Fiji toolbar.
Make sure Fiji is measuring in pixels. Analyze → Set Scale…, and if a scale has been set from a previous image, click Click to Remove Scale. The unit shown should be pixel. A leftover scale from another frame would silently convert your widths to millimeters and the ratio would still be right — but the numbers in the Entry would not be what the form asks for.
Select the straight-line tool on the toolbar (the diagonal-line icon; if it shows a segmented or freehand line, right-click it and choose Straight Line).
Zoom in on chart A until it fills a good part of the window (+ and −, or the magnifier tool; hold the space bar to pan).
Draw a line across the chart's width — the same physical dimension on every chart, every frame. The course uses the outside width of the L's horizontal arm, edge to edge, along the printed scale. Pick one convention and write it in the Entry; what matters is that all four charts, and both visits, use the same one. Click at one edge, drag to the other, release.
Measure. Analyze → Measure (or Ctrl/Cmd+M). A Results window opens with a Length column — that is the width in pixels. If Length is not shown, go to Analyze → Set Measurements… and tick it.
Repeat for B, C and D. Four rows in the Results window, one per chart, in order.
Write down the four widths on Slate B or straight into Section 3, then divide the largest by the smallest. That is the ratio — say 1.06, or 1.19. Record it to two decimal places.

Which chart read largest? Section 3 asks. If one chart reads clearly wider than its three neighbours, that chart sat higher than they did — it was on a bump, and the substrate between the charts was not one plane. If the two far charts read narrower than the two near ones, the camera was tilted. The ratio does not tell the two apart on its own; the pattern of the four widths does, and the form has a line for it.

Which frame to score

Do this for every usable frame at the photo point — three, usually. Then apply the rule, which is a rule and not a judgment: the usable frame with the smallest ratio is the one you score. Where two tie, the earlier frame. Anyone with the same three frames makes the same pick. Record which frame it was and its ratio in Section 3; the ratio travels with the cover figure for a later reader to weigh.

There is no pass mark. A frame with a poor ratio is still scored — its cover figure just carries a ratio that says how square the capture was. A limit set here would describe some other reef, some other camera, some other substrate; and it would fail the governing test, because a loose limit lets worse geometry make a claim easier to support.

Keeping the measurements

Save the Results table (File → Save As… in the Results window) alongside the frame in the archive, named so it can be matched to the file — IMG_4471_framecheck.csv, that sort of thing. It is not required by the Entry, but it is a minute's work and it lets someone re-check your ratio without re-measuring.

Record the Fiji version (Help → About ImageJ… shows it) in Section 12 of the Entry, where the manifest asks for the image viewer and its version.

Last checked against Fiji 2.x (ImageJ 1.54) in September 2026. Menu names and locations are as of that check; the method — four widths, one ratio, smallest ratio scored — comes from the Course Guide and does not change with the software. If a step here no longer matches what you see, say so.