The Olympus Pen turned a single roll of 35mm film into 72 exposures by cutting the frame in half, and D Half in RfCamera revives that math for anyone shooting on a phone. This is not a nostalgia filter — it is a working half-frame camera with its own exposure logic, its own crop, and its own case for slowing down.
Where the 72-Frame Format Came From
Yoshihisa Maitani designed the original Olympus Pen in 1959 to solve a specific problem: 35mm film was expensive in postwar Japan, and a standard 36-exposure roll cost real money relative to average wages. Maitani's answer was to turn the camera on its side inside the frame. Instead of exposing the full 24×36mm area horizontally, the Pen exposed an 18×24mm vertical slice, then advanced the film by half a standard frame pitch for the next shot. A 36-exposure roll suddenly yielded 72 photographs. A 24-exposure roll yielded 48. The Pen sold well enough that Olympus followed it with an entire half-frame lineup: the Pen EE series with a fixed-focus zone-focus lens for point-and-shoot use, and in 1963 the Pen F, a genuine half-frame SLR with a rotary shutter and interchangeable lenses instead of a mirror-slap focal-plane shutter. Konica built its own half-frame SLR, the Auto-Reflex, aimed at the same market. Ricoh's Auto Half and later the Yashica Samurai pushed the format toward compact, motorized bodies through the 1970s and 1980s. Half-frame cameras never disappeared because they were bad — they disappeared because color print film got cheaper through the 1970s and 1980s, and the cost argument that justified doubling your frame count stopped mattering as much. The format re-emerged commercially in 2022 with the Kodak Ektar H35 and again in 2024 with the Pentax 17, both aimed at photographers who wanted the doubled frame count back, this time for the compositional discipline it forces rather than for saving money on film stock.
The Geometry Behind an 18×24mm Frame
A half-frame exposure measures roughly 18×24mm, sitting inside the standard 24×36mm 35mm gate rotated ninety degrees. The aspect ratio is still the familiar 2:3 of full-frame 35mm — you are not changing proportions, you are cutting the usable area in half and orienting it vertically by default. That single geometric fact drives everything else about the format. Halving the frame area means the same length of film gives you twice the exposures: a 36-exposure cartridge becomes 72 shots, a 24-exposure cartridge becomes 48. It also means you are asking half as much silver halide to cover the same print size. To make an 8×10 print from a half-frame negative, the enlarger has to magnify the image roughly 1.4 times more linearly than it would from a full 24×36mm frame, because area scales with the square of linear enlargement — half the area means about 41% more linear magnification to hit the same output size. That extra magnification is why half-frame shooters historically leaned toward slower, finer-grained emulsions: Kodachrome 25, Panatomic-X, and later Kodak Gold 100 all show up disproportionately often in half-frame galleries, because coarse grain that looks fine at full-frame enlargement starts to read as texture rather than detail once you push it 41% further. Lens performance matters more too — the same optical flaws, like field curvature or corner softness, occupy a larger fraction of a smaller negative, so they show up sooner as you enlarge.
Building a Diptych That Actually Reads as One Image
Pairing two verticals into one diptych works when the two frames are doing different jobs, not when they are simply two photos of the same thing. A detail-to-context pair puts a close, specific object on one side — a hand wrapped around a ceramic cup, the worn edge of a doorframe — and the wider space that contains it on the other, so the viewer's eye supplies the connection the camera left out. A before-and-after pair separated by a few seconds or minutes turns two static frames into implied motion: a cyclist entering the edge of the frame on the left, gone from the right. A light-and-shadow pair repeats the same subject or vantage point under two different qualities of light — direct sun on one half, the same scene stepped into shade or backlit into silhouette on the other — and lets the contrast in exposure do the storytelling instead of a caption. The strongest diptychs share one deliberate anchor between the two halves: a color that repeats, a line that continues across the gutter between frames, a shape that rhymes. Without that anchor, two half-frame exposures just read as two small photos stacked vertically, which is the most common way the format fails. Because you are committing to a pairing at the moment you press the shutter — the two halves land on the same physical strip in sequence — half-frame shooting rewards previsualizing the second frame before you have even taken the first, rather than hunting for a match afterward.
Shooting a Roll That Doesn't Run Out
Doubling your frame count changes the rhythm of a roll more than it changes any individual exposure. A 36-frame roll that used to last one outing now covers two or three, which means the two halves of a diptych are rarely shot back to back — they are shot as opportunities present themselves, sometimes minutes apart, sometimes days apart on the same physical roll. That gap is the actual skill half-frame photography teaches: holding a compositional intention open until its match arrives, instead of resolving every frame in isolation. Meter each half on its own terms rather than assuming the first exposure's settings still apply — light changes between frames shot at different times far more than it changes between two frames shot two seconds apart on a full-frame roll. Film choice matters more here than in ordinary 35mm work because of the 41% extra enlargement half-frame negatives need: a slower stock in the ISO 100–200 range keeps grain fine enough to survive that magnification, while anything at box speed 400 or higher starts showing grain as a visible texture even in a modest 6×8 print. If you want that grain, push it deliberately rather than defaulting to whatever is loaded. Carry a small notebook, physical or in your phone, and log which half of which roll you have already committed to a pairing — with 48 or 72 frames per cartridge stretched across several outings, memory alone will not reliably tell you which of last Tuesday's frames is still waiting for its partner three weeks later.
How D Half Renders Each Half of the Pair
RfCamera runs one FilmEffect pipeline for every camera mode it simulates, and D Half is not a separate rendering path — it runs the same pipeline twice, once per half-frame exposure, then composites the pair. What you see in the live viewfinder is the colour matrix transform for the selected stock plus a GLSL fragment shader (film.frag) applying the optical character of the simulated lens: barrel distortion pushing straight lines outward near the frame edges, and chromatic aberration separating red and blue channels slightly at high-contrast boundaries. Both effects run per-pixel on the GPU, so panning the phone shows the distortion shift live rather than as a static overlay. When you press the shutter, RfCamera does not re-render from scratch or apply a lighter approximation for speed — bake.dart replays the identical pipeline, colour matrix, film.frag shader, then the grain, light-leak, dust, scanline, vignette, and date-stamp overlay stages in the same order, inside a compute() isolate, so the full-resolution bake happens off the UI thread and the saved JPEG matches what the viewfinder showed. For D Half specifically, each vertical half is baked independently through this pipeline before the two are joined into the diptych canvas, which is why grain and leak position differ slightly between the two halves even when they share the same grain plate — a real scanned film-grain plate sampled at a different offset for each half, the same way two frames from the same physical roll never grain identically. None of this touches the network: the bake runs entirely on-device, RfCamera declares no INTERNET permission, and the finished diptych is written straight to the app's documents directory.
Picking Film Stock for a Format That Enlarges Harder
Half-frame's extra enlargement factor makes emulsion structure a bigger variable than it is in full-frame 35mm work, because whatever grain pattern is in the negative gets stretched about 41% further per linear dimension to fill the same print. Kodak's T-grain emulsions, used in Ektar, Gold, and Portra, use flat, tabular silver halide crystals that pack more tightly and scatter less light than the older cubic-grain crystals still used in some black-and-white stocks, which is why T-grain color films read as smoother at a given ISO once you enlarge past what a full-frame negative would need. Cubic-grain black-and-white stocks like Ilford HP5 or Kentmere 400 grain up faster under the same enlargement, which is exactly why they read as classic and gritty in half-frame work rather than as a flaw — the format amplifies whatever character the emulsion already had. Pushing film compounds this: rating HP5 at 800 and pushing one stop in development increases contrast and visible grain at the same time, and because half-frame is already asking more of that grain structure, a one-stop push reads more aggressively in a half-frame diptych than the same push would in a full-frame shot from the same roll. If the goal is a cleaner diptych where grain doesn't fight the composition, box-speed color negative in the 100–200 range is the safer choice. If the goal is texture — grain that becomes part of what the two halves are saying to each other — a pushed black-and-white stock gives the format's enlargement penalty somewhere useful to go.
