• Please use real names.

    Greetings to all who have registered to OPF and those guests taking a look around. Please use real names. Registrations with fictitious names will not be processed. REAL NAMES ONLY will be processed

    Firstname Lastname


    We are a courteous and supportive community. No need to hide behind an alia. If you have a genuine need for privacy/secrecy then let me know!
  • Welcome to the new site. Here's a thread about the update where you can post your feedback, ask questions or spot those nasty bugs!

Canon EOS C300 - the sensor scheme

Doug Kerr

Well-known member
Canon has recently released three White Papers discussing the sensor system of the new EOS C300 cinema video camera.

One, "New 35mm CMOS Image Sensor for Digital Cine Motion Imaging", covers the basic concept of the sensor and some of its photometric and sensitometric properties. It is available here:


The second, "Sensitometric Characteristics of the EOS C300 Digital Cinema Camera", gives further insight into the photometric and sensitometric properties of the system and image processing, including information about the "Canon-log" tonal response function:


The third, "RGB Resolution Considerations in a New CMOS Sensor for Cine Motion Imaging" has other valuable information about the imaging scheme:


A handy summary of the papers is found here:


The camera basically develops a 1920 px x 1080 px image from a 3820 photodetector x 2160 photodetector sensor, with a Bayer CFA arrangement. No interpolation is involved in developing the image; the system is thought of as a "true RGB image sensor". (Note that R, G, and B in this case do not refer to the R, G, and B coordinates of the various RGB color spaces, but rather are the usual somewhat fanciful names of the three kinds of photodetectors in the usual Bayer CFA sensor.)

I had initially assumed that the arrangement of the microlenses or whatever would result in one image pixel's worth of the optical image being delivered to a fixed group of four photodetectors (which group would in effect constitute a "true RGB fat pixel photodetector").

In fact, this is apparently not so. Evidently, each sensel is equipped with a distinct microlens, just as in the familiar use of a CFA sensor to be used in conjunction with demonstrating to develop an output image whose pixel dimensions correspond to the sensel dimensions of the sensor array.

In any case, for the R and B "aspects", the image is sampled on a 1920 x 1080 grid, with the sampling "apertures" having a width about half the sample pitch.

But the R and B sampling grids are offset, both vertically and horizontally, by half the sample pitch. Thus, we do not really have an true "RGB pickup" of the color of the image at any sampling points.

I have not yet read the papers so I don't know yet whether anything is done in processing to overcome any undesirable effects of this.

The first papers shows the presence of an optical low pass filter in the chain (presumably what we often call an "antialising filter"), and its frequency-domain response is shown in a figure in the third paper mentioned above. Its presence should somewhat mitigate the problem of which I speak. (I haven't read that paper at all yet!)

My guess is that Canon has avoided the use of the term "antialising filter" because the "true RGB" image pickup of the C300 (such as it is) is (supposedly) free from "color aliasing" and they want to emphasize that. Of course the purpose of an antialising filter is not just to mitigate "color aliasing" but also to mitigate the more fundamental type of aliasing found in any sampled system (even a monochrome system).​

Well, that's what I know so far. I'll probably know more after I actually read the papers.

Best regards,