Off-Axis Holography

Fill a rigid tray about 400 mm square to a depth of 100 mm with clean dry sand. Stand it on a heavy table, on top of a slab of foam or a partly inflated inner tube. Sand damps high frequencies; the soft layer isolates the low ones. Bury the laser module and the plate holder in the sand up to their necks so nothing is standing on a stalk. Everything in the beam path must sit on the SAME mass, or a temperature change will move one part relative to another. Test it before you use a plate: set up a Michelson-like pair of reflections off two objects in the sand, project the interference onto a card, and watch. If the fringes are still moving after five minutes of settling, find the source — a fan, a fridge compressor, footsteps, a draught — and stop it. A fringe is about 0.3 micrometres wide.

Mount the laser module 1 to 1.5 m from the plate holder, aimed so its diverging beam covers the whole plate area evenly. Remove or defocus the collimating lens so the beam spreads; a spatial filter is better but is not required. Put the object directly BEHIND the plate holder, as close as it will go without touching — a millimetre or two. In this geometry the reference beam passes through the plate, hits the object, and reflects straight back into the emulsion. The two beams are almost exactly counter-propagating. Choose the object carefully. Bright, matte and pale wins: a coin, a small unglazed ceramic figure, a pale shell. Chrome and glass fail, and the notebook shows why in numbers — the reference-to-object intensity ratio goes past 50:1 and the fringe contrast collapses. Put a card in front of the laser as a shutter. You will open it by moving the card, never by switching the laser, because switching changes the wavelength as the diode cools.

Work under the amber safelight only, and only after checking the plates are not panchromatic — red-sensitive emulsions such as PFG-03M need near-total darkness or a very dim green safelight, not amber. Check the box. Load the plate emulsion-side towards the laser. Let everything settle for five minutes with the card shutter closed. Expose by lifting the card without touching the tray, for the time from the notebook. Replace the card. Develop for the time on the packet, stop, fix, and wash. At this point you have a dark amplitude hologram that will play back faintly. Then BLEACH it. The bleach converts the developed silver into a transparent salt, turning an absorbing pattern into a pattern of refractive index. The hologram goes from dark and dim to clear and bright, because it now diffracts instead of absorbing. Wash thoroughly, add a drop of wetting agent, and dry vertically in still air.

VIEW IT PROPERLY. Hold the dried plate under a single small bright source — a bare LED bulb, a halogen spot, or direct sunlight — from the same side and the same angle the laser came from. A diffuse room light will show nothing, because a diffuse source is a superposition of every angle at once and washes the reconstruction out. WHAT GABOR'S PROBLEM ACTUALLY WAS. A recorded interference pattern is a real quantity, so on playback it necessarily produces TWO diffracted orders: the wanted virtual image and a conjugate real image at the mirror angle. With an in-line reference the two lie on the same axis and cannot be told apart. Leith and Upatnieks' off-axis reference imposes a spatial carrier, and the two orders separate in angle. It is the same trick as putting a signal on a carrier in radio, which is exactly where they got it — from synthetic-aperture radar. WHY DENISYUK'S VERSION PLAYS BACK IN WHITE LIGHT. Counter-propagating beams write fringes as planes lying parallel to the plate surface, half a wavelength apart INSIDE the emulsion. That is a volume Bragg stack, and a Bragg stack reflects only the wavelength it was tuned to. Shine white light on it and it selects its own colour. A transmission hologram has no such selectivity and needs a laser or a very narrow filter. WHY HALF A HOLOGRAM IS STILL WHOLE. Every point on the plate received light from every point on the object, so every region records the entire scene — just from its own viewing angle. Break the plate and each fragment still shows the whole object, with a narrower window to look through and a coarser resolution. This is not true of a photograph, where each point on the film maps to one point on the subject, and it is the sharpest possible statement of what the difference between the two actually is. WHEN IT FAILS. Blank plate: coherence length too short, or the object too far behind the plate, or something moved. Dim but present: underexposed, or you skipped the bleach. Milky: incomplete fixing or drying too fast. Fringes visible but no image: the object moved relative to the plate, which is nearly always the plate holder flexing rather than the object.