Bayer-Filter Sensors and Beyond

This presentation was submitted first to the 2022 FIAF Congress Symposium in Budapest, Hungary, on 24–29 April 2022, but was rejected. The same proposal has been submitted a second time for “Archives of the Future: Open, Sustainable and Equitable”, the IASA 2022 Annual Conference in Mexico on 26–29 September 2022, and was accepted.


This presentation delves into several aspects of the physics and mathematics of mosaic colour filters and so-called raw data, which have a huge importance in both sensor design and video codec design. These aspects should also be considered in digital preservation and archiving. Additionally this would allow us to better document the digitisation process and to increase the reversibility that restorers require.

The most important aspect is the support of any type of video channel. Firstly this includes the raw data generated by any Bayer-type sensor prior to the de-mosaicking process. The majority of the film scanners that film archives and service providers are using today generate their digital data by the use of a Bayer sensor.

Secondly this includes the possibility to restore films as if they were videos. When working on early photographic colour processes, such as the additive ones used during the first half century of cinema, it is often more effective to handle independently the luma signal Y′ (the black and white photographic image) and the chroma components C1 and C2 (the colour filters). Working with CO and CG is easier than working with the common CB and CR, because the de-correlation of the colour planes is better (and the transcoding from and to R′G′B′ is much faster, therefore one can continually check the results on modern monitors). For example, Dufaycolor can be restored more easily, in a better way and more quickly.

The support of 1D and 3D LUTs allows us to document in a transparent manner non-standard digitisations, which are often necessary when colour-degraded film reels have to be digitised. This makes obsolete part of the following colour restoration process.


  • The presentation’s slides (PDF, 5.0 MB, MD5 = 025d6f1b17f7b7e22524d477e67924e7).


Having graduated in both mathematics and computer science, Reto Kromer became involved in audio-visual conservation and restoration thirty-six years ago. He has been running his own preservation company, AV Preservation by, and also lecturing at the Bern University of Applied Sciences, the Academy of Fine Arts Vienna and the Elías Querejeta Zine Eskola in Donostia (San Sebastián). His current research includes colour spaces, look-up tables and codec programming and emulation. Previously he was head of preservation at the Cinémathèque suisse (the Swiss National Film Archive) as well as a lecturer at the University of Lausanne. He served as an AMIA board member for four years. His work has seen him honoured with an inaugural Joint Technical Symposium Award.