Camera manufacturers often advertise their image processors version in advertisements, such as DIGIC 6 (Canon), EXPEED 5A (Nikon) and many more. But we never see they advertise the sensor they are using. Is the image processor more important than the sensor? Wouldn't the sensor be the most significant factor in terms of determining the image quality?

IMO, image processors may help:

  1. Reduce the noise when converting to JPEG. But this can be done in photoshop as well, right?
  2. Improve the burst rate. But this might vary depends on the megapixel count, despite using the same image processor, take example 5DsR 5fps Dual DIGIC 6 vs 7D Mark 2 10 fps.

Am I understanding correctly?

  • Do they? The following question assumes the reverse of your question: photo.stackexchange.com/questions/45721/… – Michael C Dec 30 '15 at 22:47
  • 1
    I think different manufacturers emphasize whatever they feel is to their advantage. If they feel their sensors are superior that is what they talk about. If they feel their processors are better, that is what they will market. – Michael C Dec 30 '15 at 23:04

This is what we call false differentiation in advertising. It applies to any marketing, not just for cameras. Think about, we have no evidence than any of this is better than the other expect for the marketing:

  • Digic 6 is that better than EXPEED 4A or Venus II?
  • Windex with Ammonia-D does that clean better than without? Than other cleaners?

Processors are just given names and a usually increasing number to claim improvement. There are no supporting evidence than one is better within a brand and even less across brands. For sensors, they cannot claim much superiority because most are made by Sony, the remainder are mostly by Canon, Fuji, Aptina and Toshiba (recently sold to Sony). Still, when they can claim a difference, they do:

Again, think about it. Is the superiority of any of these over another obvious? Manufacturers of course claim so which is why they name they new technology: To tell buyers that it is different and better.

These sensors all operate on RAW data which has more sample-depth than JPEG, so they can do a better job to apply image processing on JPEGs. Many times though you can get similar (but rarely identical) results by processing from RAW.

| improve this answer | |
  • I think, to an extent, it is useful when comparing models within the same brand. For example, if comparing two very similar models from Canon and one has the DiGiC 5+ and the other has a DiGiC 6 processor we can know the comparative processing power. At least for Canon cameras, the specs of the various processors are available, even if that info is not highly marketed. – Michael C Dec 30 '15 at 23:02

The processor is key to how the image and/or video is captured from the sensor. There's a tremendous amount of work that has to be done by the processor, even for RAW files, before the file is written to internal memory and made ready for reading out of the camera. With every camera there's a constant tradeoff between processor features, speed, and power consumption. Manufacturers are in a constant race to increase the first two will simultaneously decreasing the latter. When the processor is updated between camera model releases you can be assured that not only has the silicon advanced, but so has the firmware, in order to work those three key issues.

Speaking of firmware, one of the key, unsung features of advancing processors is the ability to update the firmware on the cameras after the sale. Fuji is the most visible example in updating their camera firmware, adding new features and enhancing existing ones. This gives Fuji customers a feeling of continuing value in the cameras they've purchased and builds a sense of loyalty. And they're not the only ones. Before Samsung decided to withdraw from the ILC market they updated the NX1's firmware multiple times, fixing bugs and adding new features. Olympus has been (lately) upgrading the E-M1 and E-M5 II firmware to great fanfare. And Nikon and Canon are not to be left out either. In order to provide this key capability the internal processor has to have enough unused storage to allow for future expansions and updates. Unlike a PC, you can't just drop in new storage to expand what originally came with the camera; everything in a camera is on a single chip (SoC).

| improve this answer | |

Most consumer cameras use some sort of CMOS sensor, and the majority of those are made by a handful of companies (though this is changing). The major distinguishing factors are how colour separation information is handled. There are details regarding how that CMOS is configured and powered, but sensors are generally very similar in how they respond to light.

Once upon a time, Nikon was pretty vocal about their sensor tech, as they were one of the few camera companies to get custom sensors made for them in a specific fab. I think all the major camera companies own their own fabs now, though many do multiple sourcing, and contract out or own pieces of fabs instead of owning a fab outright. Anyway, none of this probably matters, as the sensors themselves are not very interesting.

So, manufacturers are going to talk about how they all take somewhat similar sensor tech and the voltages they make and actually turn that into data that can be turned into an image. Those algorithms are more important, in many ways, than how a transistor behaves when a photon smashes into it. At least, from an engineering and firmware point-of-view.

Of course, those cameras that use somewhat unique sensors, like the Foveon X3, talk that up quite a bit.

The rest of the pack distinguish themselves in other ways, because at the end of the day, whether a sensor-array-with-bayer-filter-and-support-silicon is made in Fab A or Fab B is of not much interest to engineers, or most photographers. These parts of the cameras are sort of commodity items with little to distinguish them from each-other. At least, little to distinguish them when dreaming up advertising copy. I suppose if you have a sensor array that uses a lot less power you might talk that up, especially for those power-hungry compact models. But that isn't very sexy compared with whiz-bang image processing algorithms.

| improve this answer | |
  • 1
    Canon: Own sensor. NIkon - using Sony sensors, which is why they are quiet. And no, less and less cameras use their own fabs - fabs get exponentially more expensive. It is rather concentrating, fab wise. – TomTom Dec 30 '15 at 15:00
  • 1
    Nikon make their own sensors now: they just aren't in every model. And it gets trickier because they have never (well, for DSLR models in the last 10 years or so; not sure what was or is done for their PaS models) taken standard Sony silicon: they are all special designs through specific fabs. (Full disclosure: I used to work for a company that did fab automation. I can neither confirm nor deny that I know anything about specific fabs, and what they make, and for whom.) – user31502 Dec 30 '15 at 15:08
  • 1
    Change the answer to reflect this exchange. – user31502 Dec 30 '15 at 15:36

Because it would show how little progress they did (Canon) or how well - they are using third party sensors (Nikon). Sony does talk about the sensor ;)

To your questions:

1: No - you lose A LOT of data in JPG already (8 instead of 14 bit coloring), so doing this in photohop later will not be as good. This is different from RAW or a TIFF, but JPG is very low reserves for manipulation.

2: Yes and no - burst rate is rather memory dependent as you can then queue up sensor data, it is more the throughput. Given that the processors are quite unspecified besides the name this is a meaningless marketing term anyway.

| improve this answer | |
  • I think the first sub-question was based on the assumption of processing the raw data to produce a jpeg in Ps vs. processing the raw data in-camera (and not of processing the in-camera produced jpeg in PS). – Michael C Dec 30 '15 at 22:55
  • How long the burst rate can be maintained is almost entirely dependent upon the speed of the image processor, since pretty much any digital camera can maintain maximum burst rate much longer when saving images as jpeg files than when saving raw data files. And the specs of various image processors, although not heavily marketed, are available for those interested enough to look for them. At least Canon's are. – Michael C Dec 30 '15 at 22:58
  • This is crap. Really. The fact that ultimately burst rate does fail means that raw and jpg is falling behind in writing the raw data. Which means the sensor is irrelevant, the buffer is relevant. More ram before processing the sensor data means you can maintain burst rate longer. This is the logic level you should learn in primary school. And no, the buffer is not part of the sensor. Btw., nearly all cameras maintain burst longer writing jpg because WRITING speed is limiting - i.e. the smaller jpg files simply are written faster to the card. – TomTom Dec 31 '15 at 14:06
  • Of course cameras can maintain the burst rate longer because the amount of data needing to be written to the memory card is much less. But on the other side of that coin, the processor must be able to keep up with the heavier load when performing demosaicing and other image processing to convert raw to jpeg. The load on the processor is much less when the only conversion needed is to produce a thumbnail preview to add to the raw data along with metadata, rather than a full size, high quality jpeg. – Michael C Dec 31 '15 at 14:41
  • And while I'll grant I may have appeared to have overstated the importance of processor speed with regard to burst rate by not also discussing the need for the entire system to be properly matched, the fact still remains that 1) the OP was referring to processing the raw data in PS, not processing an in-camera jpeg with Ps and 2) processors are not "unspecified" - the performance numbers are publically available for at least some manufacturers including the camera company with the largest market share on the planet. – Michael C Dec 31 '15 at 14:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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