X5 Sensor crop example

[FishTales]

I just shot this example that illustrates the different crops you get depending on wether you are shooting still or video. Usually, I just grab a still frame from video to post on my blog, but if I really need a still, obviously it's going to be better to shoot a raw still frame and work from that. So while doing my fourth flight after upgrading to the X5, I decided to shoot a couple stills of a scene that I had been filming. I have my Inspire setup to shoot 4:3 aspect ratio for stills, so I was figuring on a bit more image top to bottom. My initial surprise was how much extra image I also got side to side. But when I though about it for a second, not so surprising. I'm setup to shoot 3840 30P video, so naturally it isn't going to use the whole 4608 pixel width of the sensor. The unfortunate effect of this is that your angle of coverage goes way down for video compared to a still photo. It's clearly obvious to me that I'm going to have to get an Oly 12mm, and wish there was something even wider for more dramatic shots. Here is the sample still frame, with a red outline showing what I got on the corresponding video scene:

 

[InterMurph]

This image demonstrates the crop modes of the Panasonic GH4, which uses a very similar sensor. We have no details on DJI's crop modes, but it makes sense that they would be similar.

The entire area, including the black bars at the top and bottom, is used for still photos. The magenta rectangle is used for 1080p; note that it uses the entire width, and that data is "pixel binned" down to 1920x1080.

The Blue rectangle is used for 4K video. Note that this graphic isn't 100% accurate; it uses the blue area for both 3840x2160 (2.3x crop) and 4096x2160 (2.2x crop).

The MFT sensor has a 2.0x crop factor when used in its entirety. That means that the Olympus 12mm lens has the equivalent field of view of a 24mm full-frame lens (12mm x 2.0 = 24mm).

So when that 12mm lens is used for 3840x2160 video, its field of view is equivalent to a 27.6mm full-frame lens.

Again, this is for the GH4; we don't really know for sure what DJI is doing.

 

[InterMurph]

It's also important to note why the GH4 does this for video: to increase the video quality.

When you do 3840x2160 video on the GH4, the sensor uses an area that measures 3840 pixels wide and 2160 pixels high. So one pixel on the sensor produces one pixel in the resulting video.

If it instead used the entire sensor, and pixel-binned, then it would have to sample 4608 pixels, which is 20% more than 3840. So each pixel on the resulting video would be produced from 1.2 pixels on the sensor. This sub-dividing of pixels, called "pixel binning" doesn't work well when the pixels counts are close together. It works much better for HD video, when the 4608 pixels are binned into 1920 pixels; tha'ts 2.4 pixels on the sensor for each pixel in the resulting image.

 

[Simon Mallin]

Thanks for the explanation InterMurph. Do you by any chance know what happens in terms of cropping with the X3 sony 1.2/3" sensor?

 

[InterMurph]

I'm not sure. My X3 still images are 4384 x 3288 pixels, which is pretty close to the X5's resolution. So it wouldn't surprise me if DJI did the same sort of cropping on that camera as well.