2024-01-15

Shooting APS-C on the Canon R6 Mark II

The Canon R6 Mark II is a very capable camera - both for photo and video. It uses a full frame sensor with 24 megapixels - a resolution which is not outstanding when compared with other competitors but I consider the resulution as a good balance. While 24MP are enough for a lot of situations and also for larger prints, it allows for faster processing compared to higher resolution peers. This makes it possible to shoot 4k video with 60 fps without any cropping or pixel binning. So you´ll get the same quality and focal length, no matter if you shoot 30 fps or 60 fps. 

There is one particular feature that I want to focus on in this article - the APS-C mode. The camera allows you to apply a crop factor of 1.6, to gain a longer focal length, at the cost of resolution of course. You can even use dedicated APS-C lenses and the camera will automatically use only a cropped region of the sensor. To be honest, I don´t see myself doing this a lot for photography work, because the loss of resolution is substantial. In APS-C mode the camera will capture images with a resolution of 3744*2496 pixels, a total of 9.4MP. Also I would rather capture a 24MP image at the longest focal length that the lens offers and then eventually crop in in post. 

However, I think that for videography this feature could be useful. You can extend the focal length without sacrificing the resolution - you still get 4k (almost). That is far better than recording 4k video from the full sensor and then cropping in. Also, you have the possiblity to use your APS-C lenses.

But there´s one thing I don´t quite understand. In APS-C photo mode, the resolution is 3744*2496. We can assume that this is the largest region of the sensor that is covered by the image circle of an APS-C lens. The diameter - for simplicity measured in pixels, rather than millimeters - is 4500. In other words, the image circle is 4500 pixels wide. The largest rectangle with a 3 by 2 aspect ratio, that fits in this circle is 3744*2496.

In APS-C video mode, the camera records 4k video internally. However, this doesn´t mean it reads a 3840*2160 area from the sensor. Instead it reads a smaller area and upscales it to 4k resolution. The camera can also output raw video via HDMI. In this case the exact resolution is 3744*2106. We can assume that the same readout area is also used for internal recording. As can be seen, the horizontal resolution is the same as in photo mode, which appears reasonable at first glance. However, as this is a 16 by 9 aspect ratio, it means that the diameter is smaller. In fact the diameter is 4296 pixels in this case, much less than the 4500 pixels limit defined by the image circle. 

If you do the math, you´ll find that the largest 16 by 9 frame, that fits into the 4500 pixels image circle is 3922*2206 - quite a bit larger than 4k and substantially larger than 3744*2106.

I really find this design decision that Canon has made strange. It means the camera has to upscale the image by about 2% to be recorded in 4k. If Canon had decided to read a native 4k area of the sensor, this upscaling would not be necessary and the camera could use the processing power for something more useful. Also, capturing only a smaller area means capturing less detail. As it would be easily possible to record native 4k, the decision to record less than that and intentionally limit image quality appears awkward. 


The smaller frame also means that the focal length is not what you´d expect. While the image width is the same as in photo mode (which probably is the reason for Canon´s decision) the focal length is not. The focal length refers to the diameter of the image. Cropping in - be it in both directions or only vertically - means you end up with a longer focal length. 

When recording raw video with an external recorder, using the 3922*2206 area could provide some room for adjustments in post.

Well, I don´t assume a major improvement in image quality if the camera used a larger area of the sensor for APS-C video. I just don´t understand why it doesn´t. I have named quite a few benefits of using a larger area but I can´t think of a single benefit of using only a 3744*2106 pixel area.

Let me know in the comments if you agree or disagree.