Forgotten in its old age

by Aditya

submit your photo


Hall of Fame
View past winners from this year

Please participate in Meta
and help us grow.

Take the 2-minute tour ×
Photography Stack Exchange is a question and answer site for professional, enthusiast and amateur photographers. It's 100% free, no registration required.

Why do high resolution cameras shoot stills at high resolutions, but only typically shoot video at 1080p which is about 2 megapixels?

For example I have a Sony-NEX 6 which can shoot stills at 16 megapixels but only 1080p by 1920p video.

Also how do digital cameras separate the frames in video if the lens shutter never closes?

And if you can shoot video at 30 fps or 60 fps then why can't cameras have a continuous drives at 30 to 60 fps?

share|improve this question
    
See how long time some camera models need to process and store one image into memory card (in RAW format though): Shot-to-shot cycle times. –  Esa Paulasto Oct 18 '13 at 6:33

3 Answers 3

up vote 5 down vote accepted

Why do high resolution cameras shoot stills at high resolutions, but only typically shoot video at 1080p which is about 2 megapixels?

For example I have a Sony-NEX 6 which can shoot stills at 16 megapixels but only 1080p by 1920p video.

There are several reasons why most high resolution still cameras that also shoot video don't produce video at the same resolution, and why they can't maintain the same frame rates when shooting high resolution stills that they use when shooting video.

  • By far the biggest issue is the amount of data produced for a given amount of time. Since a 20MP JPEG saved with a low compression rate is typically 10 times the size of a single 1080p HD video frame, and a 20MP RAW file will be 2-3 times the size of the JPEG, that means that 20-30 HD video frames are roughly the same size as a single 20MP RAW file. To put it another way, you can store an entire second's worth of HD video in the same space as ONE 20MP RAW file.
  • The data coming off the sensor must be interpreted by the camera's processor. Like all other digital processors, each camera is limited by how fast the CPU can crunch the numbers. Increasing the data rate of the processor 20 fold would be prohibitively expensive, and create the need for cooling systems too large to fit in a DSLR sized package.
  • Current write speeds of even the fastest UHS-1 SD cards and UDMA-8 CF cards could not keep up with a data rate high enough to record 30 fps of 20MP RAW files.
  • When shooting still images, most cameras pause to recalculate things like focus and exposure between each frame. With a DSLR, this also requires the reflex-mirror to cycle down and back up between each frame, and the lens to open to maximum aperture between each shot. The highest end FF DSLRs can do all of this at frame rates of 10-12 fps! Lower cost DSLRs typically shoot at 3-5 fps. But they can only maintain that pace for a few seconds before the rate of transferring all of those bits to a memory card forces them to wait for space in the buffer memory to clear.

Also how do digital cameras separate the frames in video if the lens shutter never closes?

There is no mechanical shutter on the lenses of your Sony NEX-6. There is a mechanical two curtain focal plane shutter in front of the sensor assembly inside the camera's body. Depending on the settings you have selected, it normally operates as the same type of shutter does on a DSLR or even old film camera when you are shooting still images. The pixel sites on your sensor are energized just before the first curtain opens until shortly after the second curtain closes. The amount of time the sensor is energized and collecting light can be anywhere from a tad longer than the flash sync speed of 1/160 second to the maximum 30 second exposure time, or even longer when using Bulb mode.

When you are shooting video, the shutter stays open and the amount of light collected by each pixel site are read out to the processor at specific intervals, the 'counter' for each pixel is reset to zero and the pixel starts counting the number of photons that strike it again. CMOS sensors recording in 1080P (the p is for progressive) read each pixel site from the top to bottom of the frame in sequence, then start over at the top for the next frame. Older Standard Definition and 720i cameras read the odd numbered lines from top to bottom, then go back and read the even numbered lines (The i in 720i stands for interlaced). Higher end CCD sensors read the entire sensor at the same instant, dump that data to an on sensor buffer, and begin collecting more light while the data from the previous frame is sent to the camera's processor.

And if you can shoot video at 30 fps or 60 fps then why can't cameras have a continuous drives at 30 to 60 fps?

Some do, but any priced at consumer levels are limited when shooting such high frame rates to lower resolutions similar to the resolution of HD video (30 fps) or even lower resolution (60fps), and they are also limited to shooting with the shutter remaining open such as occurs when shooting video or stills in Live View. And it should be quite obvious if you are shooting 60 fps you are limited to shutter speeds faster than 1/60 second. In effect, those cameras are doing nothing more than taking video and saving each frame as a separate file.

share|improve this answer
    
This is not a compelling answer. Why can the iPhone shoot 120 FPS video at 720p while the NEX-5R can't? Qualcomm's new mobile SoCs can encode 4k video, so why can't the NEX? –  Kartick Vaddadi Dec 12 '13 at 16:45
    
Probably because Sony decided not to spend the money to put a processor such as the iPhone and SoCs have into a dedicated camera that doesn't need that much processing power to do what it is primarily designed for: taking stills. In exchange for all of that processing power in the smart phones you settle for a fairly crappy lens. If you want both the higher grade processor and pro quality lenses then buy a dedicated video cam. Sony sells those too. –  Michael Clark Dec 12 '13 at 17:52
    
Reading out information off of a 20MP sensor and down-converting it to 720p or even 4K video resolution also takes a fair amount of processing power. The smart phones generally don't need to do that kind of down conversion because their sensors don't have that kind of resolution to begin with. The ones that do have higher resolution sensors generally just skip lines in the sensor instead of reading out the entire chip and then downsizing it. –  Michael Clark Dec 12 '13 at 18:02
    
Thanks, Michael. Yes, I'd read about skipping lines in the sensor while recording video, and that the RX10 is one of the few cameras that does't do that. BTW, the SoC in the iPhone 5S is estimated to cost only $19: idownloadblog.com/2013/09/24/… When I paid $1200 for my NEX with two lenses, $20 is a trivial amount more to pay if it results in better capabilities, as the OP wanted, too. Basically, a better than 1080p resolution at 60 fps, or better than 60fps at 1080p. Buying and carrying separate photo and video cameras is impractical. –  Kartick Vaddadi Dec 12 '13 at 18:14
    
But the point is, to get 720p, 1080p, or 4K video out of the sensor in your NEX requires even more processing power than getting it out of the lower resolution sensors. And the expectation of quality in terms of AWB, low light/low noise performance, etc. are considerably higher from the average user of a camera such as the NEX than the average user of a smart phone. All of that requires more in-camera processing power. Like many technology choices it comes down to a decision between a device that does several things well or a device that does one thing very well. –  Michael Clark Dec 12 '13 at 18:35

Data rates, data rates, data rates. A high resolution photo from a 24 megapixel sensor for example is 20 to 30MB in raw or 10MB in high quality jpeg. If you were to store that many frames for video, that would be 240MB to 720MB per second. No SD card can write that fast and the amount of information that would have to be processed by the image processing circuitry is even more ridiculous.

To give an idea of the raw data rates, a typical DSLR has either a 12 or 14 bit sensor, so at 14 bits, a 24 megapixel camera is going to produce 36 megabytes of data per frame, that means that for 24fps video, the image processing circuitry has to be able to keep up with processing 864 megabytes per second. That's almost 52 gigabytes per minute or 3.2 terabytes per hour. Comparatively, for 1080p video, it only has to deal with 1/12 that information. 75MB a second is far easier to manage and with compression, that size can get down to something manageable (and keep in mind, the bigger the raw data gets, the harder it becomes to do compression in real time too).

It simply isn't possible to keep up with or store the necessary amount of video. This is also why burst mode on even high end DSLRs can only do 8 to 11 fps and why they run out of buffer space relatively quickly (typically within 20 shots or so when shooting RAW).

share|improve this answer

There's no good reason why cameras can't do a better job at video.

But first, a qualification: I'm not talking about shooting full-res 20MB 16 megapixel frames at 60 FPS, which the other answers point out is hard / impossible for now.

But I see no reason why cameras shouldn't be able to do a BETTER job than they do now, which is shoot 60 FPS at anything better than 1080p, like 1440p (2560 x 1440). Or video at a frame rate greater than 60 fps, and any decent resolution. Heck, the iPhone shoots 120 FPS at 720p.

And if we compress that down to 30mbps (< 4 MB /s), SD cards should be able to handle that comfortably. Phones are starting to support 4k video, and Qualcomm's new mobile SoC handles 4k video encoding, so it's not a hard thing to do. And standalone cameras usually have bigger batteries than ultra-thin phones, so power is not the constraint, either.

I was told that iPhones have more powerful processors than standalone cameras, but I see no reason why that can't be improved. The SoC in the iPhone 5S is estimated to cost only $19. When people pay more than a thousand dollars for an NEX with lenses, $19 is a trivial amount more to pay if it results in better capabilities.

Like you, I own an NEX, which cost me a substantial amount of money. Don't get me wrong, I like my NEX, and I'm not saying, of course, that the iPhone is better overall than the NEX for photography; just that it's better at SOME things. And I see no intrinsic reason why the NEX can't do everything or most of what the iPhone 5s can do:

  • Take 720p video at 120 FPS.

  • Take video at a higher resolution than 1080p.

  • Take a burst of photos and merge them in software to reduce noise. This is called image stacking.

  • Have a panorama mode that works. The NEX's panorama mode, at least on my NEX-5R, is useless. It says I'm moving my camera too fast, or too slow, or not in a perfect line, or something else.

  • Tap to set exposure for a certain part of the scene. Maybe the NEX-5R supports this, but the touch-screen is so bad it's not worth using.

Again, my point is not that the NEX sucks (I wouldn't have bought it if that were the case) but that there are SOME things the iPhone does better, and I don't see these as intrinsic tradeoffs.

If you're talking about putting an APS-C sensor in a pocketable camera, that's probably an inherent tradeoff for now. If you're talking about making a 50x zoom lens for an APS-C sensor camera at a reasonable price, size and weight, that's an inherent tradeoff, again. But I don't why the things above are inherent tradeoffs. They can be fixed with software, and a more powerful processor. We're not talking about any fundamental limitations of optical engineering.

So, my answer to your question is not something you'd perhaps like to hear: there's no reason (standalone) cameras can do a better job at some of the things you're asking about.

Footnotes:

1 Yes, it takes more power to process the output of a 16MP sensor (like the one used in the NEX) than it does to process the output of a 8MP sensor (like the one used in the iPhone), but we can easily get around that by reading only some lines from the sensor.

2 Yes, NEX users expect better quality in their videos, but the least I'd expect is that my NEX at least does what my iPhone can: 120 fps at 720p, with comparable quality. If it can do better, great, but it should at least be as good as my iPhone. Saying "we can't do a better job than the iPhone, so let's not even try doing as good a job as the iPhone" is not excusable.

(Credit to Michael for raising these two questions.)

share|improve this answer
    
No one is saying they can't. But apparently no one sees the market demanding such a device for what it would cost to bring it to market, either. –  Michael Clark Dec 13 '13 at 2:35
    
Your NEX-5R can do image stacking. It is called Hand-held Twilight mode. –  Michael Clark Dec 13 '13 at 2:37
    
Thanks, Michael. I stand corrected. I'll try out the Hand-held Twilight mode. (I didn't notice it so far because I shoot in RAW and it doesn't seem to show up in the menus in RAW mode.) –  Kartick Vaddadi Dec 13 '13 at 4:25
    
I don't think any camera will do in-camera stacking with RAW images. Not even your iPhone. :-) –  Michael Clark Dec 13 '13 at 4:59
    
:) Just curious -- why can't the camera save all the RAWs and then stack them together and save as JPEG? Or is it something just not implemented? –  Kartick Vaddadi Dec 13 '13 at 5:37

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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