Silver crystals and pixels
A film camera captures light by a chemical reaction. In film, light striking tiny crystals of silver halide changes them so that the developer can turn them into visible metallic silver. In a digital camera, light striking a silicon sensor frees electrons, and the camera counts them. One is an analog, chemical record; the other is a set of numbers. This article follows that change in how the image is recorded and stored, setting aside the social side of the story.
What a sensor does
An image sensor is a silicon chip covered in a grid of tiny light-sensitive sites, called photosites. During an exposure, each site collects electrical charge in proportion to the light that hits it. The camera then converts each site’s charge into a number. A large grid of numbers is a digital image, and each entry is a pixel. A sensor with 4000 photosites across and 3000 down has 12 million, which is how 12-megapixel cameras are described.
Two designs: CCD and CMOS
CCD. The charge-coupled device, developed at Bell Labs in 1969, moves charge from site to site across the chip to a single output point, where it is measured. It produces clean images and dominated early digital cameras, but it uses a lot of power and is relatively costly to make.
CMOS. In a CMOS sensor, each photosite has its own small amplifier, so the charge can be read directly. A key advance was the active-pixel design, developed in the early 1990s at NASA’s Jet Propulsion Laboratory by a team that included Eric Fossum. CMOS sensors can be made on the same kinds of production lines as other silicon chips and integrated circuits, and they integrate with the processing circuits on the same piece of silicon. Early CMOS images were noisier, but improvements over the 2000s made them competitive, and they are now the dominant type in smartphones and most cameras. Related digital detectors also replaced film in medical X-ray machines, though those use sensors designed for that purpose.
Seeing in color
A photosite only measures brightness. To capture color, a mosaic of tiny red, green and blue filters covers the sensor. The most common pattern was devised by Bryce Bayer at Kodak and patented in the mid-1970s. It uses twice as many green filters as red or blue, which roughly matches the eye’s greater sensitivity to green. The camera then calculates a full-color value for every pixel by interpolating from its neighbors, a process called demosaicing. Alternatives exist, such as sensors that stack layers sensitive to different colors, but the filter mosaic is by far the most common approach.
Processing inside the camera
Film had a fixed chemistry, but a digital camera has a processor. After reading the sensor, it adjusts white balance, reduces noise, sharpens edges and applies a tone curve. It then saves either a processed file, usually in JPEG format, or a raw file that keeps most of the sensor data for later editing. JPEG compression, standardized in the early 1990s, discards some information, but at a level most viewers do not notice, and it made it practical to fit many pictures on a small card.
Storage: cassettes to memory cards
The first digital camera prototype recorded to an audio cassette. Early consumer cameras used built-in memory that held a handful of pictures. Removable cards then made storage flexible: CompactFlash, introduced by SanDisk in 1994, was followed by SmartMedia and then SD cards at the end of the 1990s. Capacities rose while prices fell, so that a single card could hold more pictures than anyone would have shot on film in a year. Many cameras also moved from an optical viewfinder to a rear screen, and later to electronic viewfinders in mirrorless models.
Cameras built around sensors
Because the sensor sits where the film used to be, early digital single-lens reflex cameras were often modified film bodies. Kodak’s professional DCS cameras of the early 1990s paired a digital back with a Nikon body. The Nikon D1 of 1999 was a landmark professional digital SLR, and Canon’s D30 of 2000 used a CMOS sensor at a lower price. Sensor sizes were often smaller than the 35mm frame of film, so lenses behaved as if they were zoomed in. “Full-frame” sensors that match a 36 by 24 millimeter frame arrived in the early 2000s at high prices.
The end of the film era
Digital sales overtook film sales in the first years of the 2000s in many markets, and film camera production dwindled. Kodak, which invented much of the technology and held patents, filed for bankruptcy protection in 2012. Explanations for the decline vary, and the role of a single decision or technology is easy to exaggerate.
What the change preserved and altered
Optics, shutters, apertures and the basic rules of exposure carried over from film. What changed was the recording medium and everything that depends on it: immediate review, copying without loss, and editing on a computer. Whether a digital file will last as long as a well-kept film negative is an open question, since formats and storage devices change more quickly than a strip of film.