FESEM Viewer

Turn a folder of SEM/FESEM TIFF exports into a single interactive HTML page for side-by-side sample comparison — no manual importing or arranging.

One command scans the session folder, converts every .tif to browser-friendly JPG, reads the instrument metadata, and generates _viewer/index.html:

• Filters — view type (Surface / Cross), magnification chips, sample multi-select with text search
• Grid layout — e.g. all samples at x500 side by side
• Matrix layout — table of samples (rows) × magnifications (columns); missing shots show as gaps at a glance
• Lightbox — click any image for fullscreen; arrow keys walk the filtered set; click again for 100 % pixel zoom
• Pin & compare — pin 2–4 images, open them side by side
• Each image is captioned with magnification, accelerating voltage (kV), working distance (WD), and detector signal from the instrument's .txt file

Why the conversion step: browsers cannot display TIFF. The originals are never touched — JPGs are written to a separate _viewer/img/ folder.

Requirements

• Python 3.9+
• Pillow

pip install -r requirements.txt

Usage

python fesem_viewer.py "C:/path/to/SEM_session_folder"

Then double-click SEM_session_folder/_viewer/index.html. It opens in any browser directly from disk — no server, no internet needed.

Options:

Option	Effect
--force	Re-convert every TIFF (default skips already-converted, so re-running after adding new images is fast)
--out DIR	Write the viewer somewhere other than <root>/_viewer

Re-run the same command any time you add images to the folder — conversion is incremental.

How to structure the session folder

The viewer expects the layout most JEOL FESEMs export naturally:

SEM_session_folder/                ← pass this path to the script
├── SampleA/                       ← one folder per sample
│   ├── Surface_x250.tif
│   ├── Surface_x250.txt           ← paired instrument metadata (optional)
│   ├── Surface_x500.tif
│   ├── Surface_x500.txt
│   ├── Cross_x500.tif
│   └── Cross_x500.txt
├── SampleB/
│   └── ...
└── Group1/                        ← nesting is fine
    └── SubsampleC/                ← shown in the viewer as "Group1/SubsampleC"
        └── ...

Naming rules

Sample = folder. The folder name is the sample label shown in the viewer, so name folders after your samples (05BA4, 20GP4W3, …). Nested subfolders work; the label becomes the relative path (DEW/BP 1).

Image file = <View>_x<mag>.tif

Part	Meaning	Examples
<View>	Observation method. Surface or Cross get their own filter chips; anything else is grouped under "Other"	Surface_x500.tif, Cross_x250.tif
x<mag>	Magnification	x250, x500, x1000, x2500
_2 suffix	Second shot of the same view/mag (shown with a #2 badge)	Cross_x500_2.tif

Bare-number filenames (1300.tif) also work if the paired .txt exists — the magnification is taken from the metadata.

Metadata file = same name, .txt extension. JEOL writes this automatically next to each image. The viewer reads:

Key	Used for
$CM_MAG	Magnification — overrides the filename
$CM_ACCEL_VOLT	kV in the caption
$$SM_WD	Working distance in the caption
$CM_SIGNAL_NAME	Detector (SEI, …) in the caption

Without a .txt, the magnification falls back to the filename and the caption has no kV/WD.

Why metadata beats filenames

Filenames are typed by the operator and can be wrong. In a real 154-image session, two files disagreed with the instrument record (one file named x250 was actually captured at x500). The viewer always displays the instrument's value. Nearby magnifications are grouped for filtering (x251 files land in the x250 chip), so a typo doesn't create a stray filter button.

Output

SEM_session_folder/
└── _viewer/
    ├── index.html    ← open this (self-contained, ~portable with the img folder)
    └── img/          ← converted JPGs, named <sample>__<original>.jpg

Delete _viewer/ at any time and re-run to rebuild from scratch — the original .tif/.txt files are read-only inputs.

Tips

• To share a session with someone, zip the _viewer/ folder — it contains everything the page needs.
• Big sessions: images are lazy-loaded, so the page opens fast even with hundreds of micrographs.
• The x prefix match is case-insensitive; Surface/surface/SURFACE all work.