Noise performance of modern cinema cameras

2025‑11‑23

A 30‑minute talk
at Oslo Digital Cinema Conference (ODCC)

by Charles Poynton

Video camera noise has traditionally been measured with the lens capped. When the camera lens is capped, the sensor obviously senses no light. But what if there was a noise source inherent in light itself? Video camera sensitivity was historically characterized by the lens aperture required to achieve reference video output level with a fixed scene luminance. However, sensitivity and noise are two sides of the same coin. How can it be that the sensitivity stated on a camera spec sheet fails to mention noise?

Light is a quantum phenomenon, so even before electronic noise is introduced in the sensor readout, there is noise inherent in the light itself. Photons arrive in discrete quanta (photons) that are converted to discrete electrical quanta that are collected (electrons). The uncertainty in photon arrival represents a very important noise source. This photon (“shot”) noise is not merely significant, it is dominant across most of the exposure range of a modern HD or digital cinema sensor. However, because the noise specification of a video camera historically involved a capped lens, this important source of noise has been absent from commonly quoted specifications.

In this talk, I will describe the physical origin of photon noise. I will explain the regions of the exposure range where photon noise matters, and I will describe its effect on images. I will provide some suggestions on how to predict or measure photon noise, and I will propose how photon noise could be captured in a specification. Photon noise has major implications in characterizing sensors and cameras intended to originate high dynamic range imagery; I will discuss how photon noise relates to good HDR performance.

Charles Poynton is an independent researcher and image/colour scientist based in Toronto. He wrote the book Digital Video and HD Algorithms and Interfaces, now in its second edition. Thirty years ago, he chose 1080 image rows for HD standards, by which “square pixels” were established for HD and digital cinema. He earned his PhD in 2018 from Simon Fraser University with a dissertation entitled Colour Appearance Issues in Digital Video, HD/UHD, and D‑cinema.