How to troubleshoot a hologram
Task-oriented. Find your symptom, apply the fix.
Inline holography is simple to build but fussy to get right. Almost every failure comes down to one of four things: coherence, stray light, geometry, or reconstruction settings. Find your symptom below.
Symptom: no rings or fringes at all on the live image
The hologram pattern never forms. Work through these in order:
- Too much stray light. This is the #1 cause. Cover the whole setup with a box or dark cloth so only the LED reaches the sensor. Contrast should jump immediately.
- Pinhole too big. A large pinhole kills spatial coherence. Make a fresh, smaller one (fold foil ~8×, punch, pick the tiniest clean hole).
- Sample too far from the sensor. It must be almost touching the sensor. Even a few millimetres of gap can wash out the pattern.
- Sample too dense. A crowded or thick sample scatters everything into mush. Switch to a sparse sample — a little dust, pollen, or a few isolated particles.
- No colour filter. Without it the LED is too broadband (low temporal coherence). Add the gel filter.
Remove the sample entirely and look at the bare illumination. You should see smooth, even light. If that's messy or full of stray glare, fix the light source and shielding before worrying about samples.
Symptom: the reconstruction never comes into focus
You see rings, but dragging dz never produces a sharp image.
- Wavelength is wrong. Set it to match your filter — green ≈ 532 nm, blue ≈ 450 nm, red ≈ 650 nm. A wrong wavelength shifts where focus lands and degrades it.
- Pixel size is wrong. For the Raspberry Pi camera the default is 3.45 µm. If you changed cameras or binning, update it. (See Parts and parameters.)
- You're not scanning
dzfar enough. Sweep slowly across the whole range; the focus distance can be larger than you expect. Then fine-tune around the best spot. - Wrong colour channel. Read the channel that matches your filter (red filter → red channel). The wrong channel can be nearly blank.
Symptom: a ghostly halo around the sample
That's the twin image, and it is normal for simple inline holography — not a fault. It happens because the camera records only brightness and loses the wave's phase, so the real image and its mirror twin both focus at once. (Full explanation.)
For a first hologram, just recognise it and move on. To reduce it you'd need advanced phase-retrieval algorithms or an off-axis setup (the Mach–Zehnder route).
Symptom: blocky, smeared, or "JPEG-looking" reconstruction
- Compression artifacts. Capture at the highest resolution and least compression the camera offers. Heavy JPEG compression bakes in blocks that the reconstruction then amplifies.
- Save stills in a lossless format (PNG) rather than JPEG when reconstructing offline.
Symptom: the pattern shifts, swims, or won't hold still
Mechanical instability.
- Put everything on a solid, heavy surface. No wobbly desks.
- Make sure every cube is locked on puzzle pieces top and bottom.
- Shield from draughts, and don't lean on or bump the table while capturing.
- Let the setup settle for a few seconds after any adjustment before judging it.
Quick reference table
| You see… | Most likely cause | First thing to try |
|---|---|---|
| No fringes at all | Stray light | Cover the setup |
| No fringes at all | Pinhole too big | Make a smaller pinhole |
| Faint / mushy fringes | Sample too dense or too far | Sparser sample, move it onto the sensor |
| Won't focus | Wrong wavelength / pixel size | Match settings to hardware |
| Won't focus | dz range too small | Sweep the full distance range |
| Halo around sample | Twin image (normal) | Accept it, or go off-axis |
| Blocky image | JPEG compression | Capture lossless, max resolution |
| Pattern swimming | Vibration / draught | Stabilise and shield the setup |
Related
- Your first hologram — the full build.
- Parts and parameters — every setting and its default.
- How reconstruction works — why these settings matter.