Usage guide

From install to a cropped roll in five minutes — plus the tuning reference for tricky scans.

Install

Download and unzip. You get a folder named Cut Corners.lrplugin.
Move it somewhere permanent (e.g. Documents/Lightroom Plugins), then in Lightroom Classic open File → Plug-in Manager → Add and select it.
The standard download includes the detection engine — nothing else to install. If you chose the lightweight build instead, install Python 3 with pip install opencv-python numpy and check the Python executable path in the Plug-in Manager's Cut Corners section.

Not sure the engine is working? Plug-in Manager → Cut Corners: Support → Run Diagnostics checks everything and tells you exactly what's missing.

Crop a roll

Select one photo or a whole roll in the Library.
Run Library → Plug-in Extras → Cut Corners: Auto-crop film frames.
Pick the film type:
- Color negative — orange-masked base
- B&W negative — bright clear base, dark frame
- Slide / positive — dark rebate, bright frame
- Already inverted — scans converted to positive before cropping
- Auto-detect — per-image characterization, for mixed rolls
Pick the film format (35mm full frame, half frame, XPan, 6×4.5, 6×6, 6×7, 6×9, 4×5, or custom) — it sets the aspect ratio detection expects.
Optionally enable Dry run to see what would happen without changing anything.
Review the summary. Skipped frames carry a color label (red by default), so you can filter to them in the grid and crop those few by hand.

Frames that already have a crop are skipped unless you enable Re-crop photos that already have a crop, which resets the crop first so detection sees the full scan. Everything is non-destructive: a single Undo (or the Crop tool in Develop) reverts any of it.

Settings & Preview

For a tricky roll, click Settings & Preview… in the run dialog before committing. It shows the detection overlay on the first few selected photos and re-runs live as you adjust anything:

red — candidate mask
blue — detected film area
green — detected frame
magenta — the crop Lightroom will apply

Use these settings saves the values for the batch run; Cancel discards them.

Tuning reference

Setting	Default	Meaning
Edge band	0.10	Corners must lie within this fraction of each film-area edge
Crop inset	0.005	Shrinks the crop slightly to keep the rebate edge out
Aspect target	1.5	Long/short ratio; set by the film format preset, or custom
Aspect tolerance	0.08	How far off-aspect a frame may be before it's flagged
Min frame area	0.64	Smallest accepted frame as a fraction of the film area
Max frame area	0.99	Largest accepted frame as a fraction of the whole scan
Manual threshold	0 (adaptive)	Fixed brightness cutoff for rolls the adaptive threshold misreads
Mask dark film holder edges	off	Excludes dark scan surround (holders/carriers) from detection
Max preview size	1500 px	Detection preview resolution; larger is more precise but slower

Troubleshooting detection

Start with Settings & Preview… — it shows the exact rule that rejected a frame, updating live as you adjust.

Frame found but crop slightly off? Raise the max preview size, or nudge the crop inset.
Scanner bed or film holder visible? Blown white or black background is excluded automatically; for dark-gray holder edges enable Mask dark film holder edges.
Underexposed or unevenly lit rolls? Detection falls back through a threshold sweep and a backlight-flattening pass automatically; if it still misses, try the manual threshold.
Whole roll skipped with an aspect reason? Check the film format preset matches the roll (half frame vs full frame is the usual culprit).

For a full roll, enable Save diagnostic images in the run dialog: the temp folder (path shown in the summary) keeps, per frame, the exported preview, the detection verdict, and an overlay image using the same colors as the preview dialog.

Limitations

Mirrored photo orientations (flipped in Lightroom) are skipped.
Rotation is limited to Lightroom's ±45° straighten range.
Detection assumes one full frame per scan with film border visible around it.