Digital audio got this good. Why does film still look different?

PhotographyNo. 06By Stuart SingletonUpdated 7 min read

Stuart asks

Digital is getting so good with audio. Why isn't it there with photos yet?

The short answer

Because light is a much bigger signal than sound. At a microphone, sound is one value changing over time, and sampling theory says it can be captured exactly. Light at every point in an image is a whole spectrum of wavelengths, across millions of points. A digital sensor reduces each point to three numbers and estimates two of them. Film reduces light to three layers too, but its look is made after capture, in chemistry, in the print, and in how light comes back off the paper. Audio only had to capture a wave. Photography has to capture a spectrum, and then get the rest of the process right.

Why was sound easier to digitize?

A microphone measures one thing: air pressure over time. Sample it at more than twice the highest frequency and the wave can be rebuilt exactly, which CD audio already does for the range of human hearing. And when you play it back, a speaker turns it back into air pressure, the same kind of thing it started as.

What makes light harder?

  • It's a spectrum, not a single valueLight at each point is a mix of many wavelengths at different strengths. Sound at a microphone is one pressure reading at a time.
  • It's spread across millions of pointsA photo is that spectrum at every point across a two-dimensional image.
  • Capture throws information awayMost digital cameras put a grid of red, green, and blue filters over the sensor, the Bayer pattern invented at Kodak in 1976: half green, a quarter red, a quarter blue. Each pixel measures one of the three, and the other two are estimated from its neighbors (demosaicing).
  • Different light can look identicalBecause each spectrum is reduced to three numbers, two different spectra can produce the same values. That's called metamerism, and human eyes work the same way.

Audio only had to capture a wave. Light is a spectrum.

Doesn't film reduce light to three colors too?

Yes. Color film has three stacked layers sensitive to blue, green, and red light. The difference is what happens next. Each layer forms dye during development, the layers affect each other as they develop, a print adds a second exposure through the negative onto paper, and you see the print by light that passes through the dye, reflects off the paper, and comes back out. Much of film's look is made in those steps, after the picture is taken.

A digital file usually skips them. It's captured once and shown on a screen that gives off its own light.

Why is film emulation still behind audio emulation?

Audio plug-ins model the gear itself: tape machines and circuits, built from how the hardware behaves. Most film looks are made with a lookup table, or LUT, which maps input colors to output colors sampled from a film stock. A LUT records what a film stock produced under certain conditions. It doesn't model the chemistry, so it can't predict what the stock would do with different light, exposure, development, or paper.

Stuart's take

Most people chase film's color. Color is the wrong dragon to chase. The depth is what you actually feel, the way analog warmth is what you feel in a mix.

Which brings it down to the real question: what physically makes film look different?

Frequently asked questions

Why is light harder to digitize than sound?

Sound at a microphone is one value over time. Light at each point in an image is a full spectrum, and a sensor reduces it to three color values, estimating two of them at each pixel.

What is a Bayer filter?

A grid of red, green, and blue filters over a camera sensor, invented by Bryce Bayer at Kodak in 1976. It is half green, a quarter red, and a quarter blue.

Why do most film presets fall short?

Most are lookup tables that record what a film stock produced under certain conditions. They do not model the chemistry, so they cannot predict how the film would respond to different light or processing.

About

Stuart Singleton is a Nashville producer, songwriter, and photographer, and the founder of One Three Seven Music.

Conversations is his series of questions about music, photography, and making things. Each one is researched and drafted with Claude, Anthropic's AI model, then checked and edited by Stuart. Numbers are sourced or marked as estimates. This is general information, not legal advice.

Sources

  1. Bayer filter, Wikipedia (US Patent 3,971,065)
  2. Bayer Color Filter Array, 1976, IEEE Milestone proposal
  3. C-41 process and color negative layers, Wikipedia

More conversations