What can limit frames per second on a mirrorless camera?

Question

I'm looking at the specs of the recently announced Sony A7 against the specs of the NEX-7, and I am confused about one detail: the amount of shots it can take per second.

The NEX-7 can capture 10 shots per second with its 24MP sensor. The a7 can capture half of that, 5 shots per second with its 24MP sensor. As far as I can tell, the main differences are that the a7 houses a full-frame sensor against the APS-C sensor in the NEX-7 and that the a7 can record 14-bit RAW images (I believe the NEX-7 can record 12-bit only).

I don't think it's the sensor size that makes the a7 slower, as the end result is still a 24MP image so is it the difference between 12-bit and 14-bit? But this confuses me somewhat as I own a D7000 that I can set to record 14-bit or 12-bit RAW's and the camera can keep the same frame rate regardless of being set to 12-bit or 14-bit. I can also make more sense of a camera with a mirror and a mechanical shutter having a lower frame rate because they have actual moving mechanical parts, but then comes along something like a D4 or 1DX that can exceed 11 shots per second!

So what is the real limit for this statistic? The memory used for the frame buffer? If that is the case, does it mean that the a7 has a less capable frame buffer than the NEX-7 even if it's a more modern, more expensive camera? And would it be possible to implement some really quick, ample memory to increase such a buffer? I recall some Nikon cameras increasing their burst-rate when a vertical grip is installed. Is that what the grip is adding on them? More memory?

Answer 1

Several things can limit the maximum potential frame rate. While sensor read on a mirrorless camera can be very fast, theoretically allowing frame rates above 12, the whole read and image processing pipeline has to support the desired frame rate.

In addition to the sensor, you also have the ADC (which may be on or off die, and parallel or not), the image processing unit itself (i.e. its data processing rate), the data rate of the buffer memory, the maximum write speed of your memory card controller, and the write speed of the memory card itself. Any one of these components can limit the maximum frame rate. While image processors used in digital cameras are highly specialized and very fast at what they do, they still have their limitations.

The 1D X, for example, which is capable of 14fps at its absolute maximum, uses a pair of DIGIC 5+ processing units. The DIGIC chips each house eight ADC units and image processing logic, and are capable of processing approximately 250mb/s worth of data. As a pair, the total data throughput rate is 500mb/s. Given that one raw sensor frame from the 1D X sensor is 19.1mp (yes, EVERY pixel in the sensor is read, so while the megapixel count of the final image is 18mp, the sensor has a total of around 19.1mp, including the masked border pixels used for calibration), one single frame at 14bit is 33.425mb in size. Multiply that by 14, and your total raw sensor read data size is 467.950mb. To get 15fps, you would require 501.375mb of data throughput, which is above the capabilities of the dual DIGIC 5+ setup the 1D X has (not to mention that there is going to be at least some overhead above and beyond just the raw pixels, meaning the 1D X is probably pretty close to its 500mb/s limit...a clear sign of fairly careful engineering of the 1D X to keep costs as low as possible while still achieving goals.)

Every camera, including mirrorless cameras, are going to have similar limitations. The 10fps of the NEX-7 is probably limited by data throughput of some component somewhere. Assuming it is 12bit data, then at 24.7mp (the total pixel count) 10fps would require at least 370.5mb/s. The image processor in the NEX-7 is then probably capable of 400mb/s data throughput in order to handle any overhead.

While a mechanical shutter is obviously a key factor that limits frame rate, eliminating it does not necessarily mean you are immediately able to achieve a significantly higher rate...every component in the pipeline must be capable of handling the higher data rate. That means making the whole entire camera more expensive...possibly considerably more expensive. It's all a balancing act, between making a product appealing to customers without breaking their bank in order to elicit the maximum number of sales without losing money. (Which, ironically, is NOT what Sony does...their electronics division has been hemorrhaging money for over a decade, to the tune of over 10 billion dollars...go figure. :P)

Answer 2

What you are comparing are the limits manufacturers chose to implement on thos models. It is not a reflection of what is possible but of what they choose to build in order to make it into a viable product.

Speed is limited by the slowest component in a chain. So, in order to make faster continuous shooting possible, the camera must have correspondingly fast components: shutter, sensor, processor, I/O controller, etc.

Mechanical shutters are presently at the limit of what is constructable for a reasonable price. This is why you see at most 14 FPS with a mechanical shutter but this has improved in recent years too. Faster mechanical shutters do exist but they are built into ultra expensive cameras for cinematography.

Going the electronic shutter route removes that one barrier which is why there are cameras which shoot at up to 60 FPS like Nikon's 1 series of mirrorless cameras. At that point another limit shows up which I would guess is the read-out speed of the sensor. If you speed up the sensor, you need to speed up the processor that processes its data and so forth.

Even if one can technologically bypass a limit, say by producing a faster processor there are consequences beyond cost. For example, a faster processor may consume more power and generate more heat. It took a long time for large sensors to be usable for live-view and then video because they use to overhead quickly.