SMPTE, together with members of its RIS-OSVP (Rapid Industry Solutions, On-Set Virtual Production) initiative, has conducted a camera and lens measurement study to test optical models used in determining the circle of confusion.
The Camera/Lens Metadata Committee – one of three interoperability groups within RIS-OSVP – carried out the study with both colour and monochrome cameras, aiming to discover the correct levels of focus or blur for computer generated images, said the organisation.
The circle of confusion is the blur circle formed when a point source of light, such as a distant star, is out of focus. The phenomenon, which appears as a small circle on the image plane, can be manipulated by filmmakers and used in the creation of visuals in post production.
Snehal Patel, director/producer and committee leader for the Camera and Lens Metadata committee, said: “Our belief is that modern digital cinema cameras are not accurately representing the usable depth of field achieved with various lenses. The calculations for depth-of-field have traditionally assumed a certain value for circle of confusion, which was based on experience with film technology. A more modern approach has to be achieved.”
Findings from the study will be used to validate the optical model for accurately determining the circle of confusion and the near and far focus planes. The results will inform proposals into improving ways of encoding, with the information also being used for virtual production and visual effects.
“Our camera tests examined multiple camera types from two different manufacturers,” added Patel. “We also examined multiple versions of a camera with and without demosaicing and with and without glass filtration in front of the sensor. The cameras were mated with a set of prime lenses and measured in the centre of the frame using a calibrated depth-of-field test chart. The results of the tests are currently being evaluated and the committee will publish them when the analysis is complete.”
The organisation also said the committee is looking at how this test will integrate into virtual twins of cinema lenses and cameras in computer-generated and ICVFX-specific workflows.
