21
\$\begingroup\$

I was recently watching a documentary on nature photographers and noticed that they frequently fired off dozens of frames in very rapid succession when taking their shots.

They then indicated that they would go through their shots to find the best.

As far as I understand it, there is no difference between the digital light sensor used in still photography and digital video, so what would be the advantage of using a still camera set to "machine gun" mode, rather than taking a short video and going through it frame by frame?

\$\endgroup\$
2

5 Answers 5

54
\$\begingroup\$

There are at least four reasons.

  • Resolution First, although the same sensor may be used, video generally uses only a portion of it (either by skipping lines or some more sophisticated method). 4K UHD video is the equivalent of about 8 megapixels. Most DSLRs these have three or four times the pixel count, and many have even more. That means it's possible to have higher resolution images (and possibly crop more tightly).
  • Image Compression Second, video is usually compressed frame-to-frame, and the image quality requirements for any given frame much lower.
  • Post Processing Flexibility Third, you almost certainly want to be capturing RAW data — "uncooked" data right from the sensor. While some high-end video cameras can capture a raw video stream, most don't — and, even then, it'd be a huge amount of data.
  • Shutter Time And, fourth: video typically doesn't give good control over shutter speed — and is limited to electronic shutter.
\$\endgroup\$
3
  • \$\begingroup\$ Dropping lines does not sound like a good idea. I hope they at least combine pixels or lines to reduce noise. \$\endgroup\$ Apr 26, 2017 at 13:30
  • \$\begingroup\$ @Trilarion Many do, though. It's faster to read the sensor, and results in less data to process through the pipeline. Some use pixel-binning at the sensor level (combining pixels), which can be worse for artifacts than more sophisticated demosaicing. Others do "full sensor readout" and do process on the fly (like the Sony A9). But in any case, it's downsampled from the full sensor resolution. \$\endgroup\$
    – mattdm
    Apr 26, 2017 at 14:25
  • \$\begingroup\$ at 18MP, that's about 50MB per frame in RAW, and assuming 24 FPS, that's 24*50=1200, or 1.2 GB every second. \$\endgroup\$ Apr 26, 2017 at 14:41
10
\$\begingroup\$

I would add two items to Matts answer

  • equipment cost: nature photographers want to have a spare camera, and expect to break things when traveling in rough places. It adds up.
  • lighting: not always an issue in nature photography, but it is easier to set up flash than continuous light.

Having said that I expect 4K video to overcome stills as technique of choice in some areas, such as news, sports and wildlife.

Resolution, the main advantage of stills camera, is less of an issue when shooting for web or print. The camera manufacturers are investing heavily in video and it improves each product cycle.

Edit: an interesting update on costs involved in switching systems. It is from the sports business, but wildlife is not that much different.

\$\endgroup\$
1
  • \$\begingroup\$ I believe I already heard that last argument when first FullHD video cameras were advertised. \$\endgroup\$ Apr 25, 2017 at 11:56
5
\$\begingroup\$

It basically comes down to the fact that affordable and portable cameras can't deliver top speed and top quality at the same time. The problem is not the sensor.

Firstly mechanical shutter gives better results than electronic but the mechanical shutter on cameras while fast is not designed to run continuously at high framerates. A high end DSLR might have a rated shutter life of 300000 cycles which sounds like a lot, but at 60FPS it translates to only about 1.4 hours.

Secondly and more importantly the data just becomes unmanagable. Lets say you have 20 megapixels and assume 12 bits per pixel raw. That works out to about 1.8 gigabytes per second at 60fps. Storing that kind of data gets into the world of big raid arrays.

It would be possible to build a camera that could capture 20 megapixels raw at 60fps and with a mechanical shutter system designed to survive such use but such a camera would be extremely bulky, heavy and expensive.

So DSLR vendors offer seperate "video" and "burst" modes with different tradeoffs.

"video" modes are designed to offer sustained high framerates but at a heavy price in image quality. Electronic shutter is used, the image is typically rendered from raw to RGB or YUV, then cropped or downscaled, then heavy compression is applied. This reduces the data rate to something managable for storing on a fast SD card.

"burst" modes can give full quality but the framerate is lower and there is a limited length of burst before the photographer must stop and wait for the camera to write out the collected data.

\$\endgroup\$
4
  • \$\begingroup\$ +1, although note that on higher-end cameras with fast cards, burst can be effectively unlimited. \$\endgroup\$
    – mattdm
    Apr 25, 2017 at 14:13
  • \$\begingroup\$ @mattdm are there DSLRs that can do unlimited burst in Raw mode? JPEG yes, but I don't think I've seen that feature in raw, admittedly I'm not across the full feature set of the very top end cameras \$\endgroup\$ Apr 26, 2017 at 10:34
  • \$\begingroup\$ @JosephRogers Not a DSLR, but Sony A9. \$\endgroup\$
    – mattdm
    Apr 26, 2017 at 11:36
  • \$\begingroup\$ @mattdm interesting, Sony certainly are producing some very interesting products recently \$\endgroup\$ Apr 26, 2017 at 12:00
2
\$\begingroup\$

Matt answered the question pretty well. I will add the problem with using the electronic shutter for high speed subjects is the possibility of distortion. Think about moving the camera back and forth quickly, you'll get a jello effect. Also, video is usually shot at low shutter speeds (often 1/30 or 1/60), meaning smooth motion, but blurry frames. You can shoot at higher shutter speeds, but that means you still won't have great stills for all the other reasons mentioned, and you will also have lousy video.

Also, some cameras shoot video with a crop, so you get a more narrow field of view.

Let's take a simple mirrorless camera as an example.

Shooting stills at high speed, you can take 11 shots per second at 24mp each, at a good shutter speed. Let's say the image is saved as a jpg, compressed to 10mb or so. If you shoot video and extract a frame from it, you'll have a 4mp or 8mp image which will probably be blurry and highly compressed. The highest bitrate is 50mbps, and you shot at 60fps, meaning 5mb per second of video. divided by 60 frames, that's less than 100kb per frame. So you have a 4 megapixal, 100kb image compared to a 24 megapixel, 10mb jpg or raw file.

\$\endgroup\$
1
\$\begingroup\$

There are so many extra advantages to using a digital still photo in comparison to a video for any type of photograph.

The first advantage is resolution - most video cameras film at 1080p, some can get up to 4K - 1080p is only about 2 megapixels and 4k is still only about 8 megapixels. When you consider that my Canon DSLR does a picture at 16mp and my phone camera even goes up to 21mp that's a massive difference in resolution.

Resolution can be very important when taking nature photos as at times you might need to crop the image or zoom in on something small in macro mode, that extra resolution allows you to do so as it provides so much more clarity than video can.

Then you've got shutter speed - most video cameras film at 50fps (some go up to 100fps) - now if you want to catch the beating of a swans wings and freeze that moment - 50fps your not going to do that. My Canon DSLR can take pictures up to 1/4000 of a second which is just a little bit faster than 1/50. In addition to this shutter speed can work the other way round. Say if you wanted to take a picture of a waterfall but you want to make the water appear smooth or you wanted to get the path of a bee as it flies around and create trails - I suppose in video mode you could create this effect by overlaying numerous frames together but why not just change the shutter speed on a still camera and get that smooth flowing waterfall without having to separate the video and then merge it all together.

Then in addition to this you've also got potential damage to the sensor - if you use a DSLR for constant video filming, especially over a long period there is a risk you can overheat the sensor - erm yes not something I'd really like to do with such an expensive camera!

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.