6 Years of Digital Camera Performance

Image quality is extremely important and one would imagine this to be primary driving factor for technological advancement among digital cameras. Partly, it is. The problem is the other part, the desire to market cameras with more megapixels. Novices can easily understand that a number of megapixels is higher than another, causing cameras targeted at that audience to keep racing for megapixels. The good news is that camera makers are also aware that advanced users demand image quality above megapixels. Hence, cameras like the Nikon Coolpix P7000 can be fitted with a 10 megapixels sensor, while other Nikon compacts currently have up to 14 megapixels.

Image quality among cameras is made up of many factors: sharpness, color, contrast, white-balance, dynamic-range, exposure, optics and noise. Noise is usually the focus of quality comparisons because it limits usable print sizes and even if reduced by software-processing it destroys image details. Color and white-balance correction for example have very little impact on image quality. Optics is also important but many modern cameras, including the P7000, optionally correct for distortion using internal processing.

Here is where 6 years of improved digital camera sensors have brought is. On the left, the Coolpix P7000 which can actually reach ISO 6400. On the right, the Dimage A2 which has a limit of ISO 800:

Nikon Coolpix P7000	Konica-Minolta Dimage A2
At the lowest setting, image quality was already rather good 6 years ago and we can see that it stayed mostly the same. ISO 64 and 100 from these cameras are just as usable. ISO 200 is only a little noisier on the older camera.	ISO 64
ISO 100	ISO 100
ISO 200	ISO 200
ISO 400	ISO 400
ISO 800	ISO 800
ISO 1600	The higher ISO crops clearly show where improvements have been going. Looking at the ISO 800 crop from the Nikon P7000, it compares to the A2's ISO 200 crop. This represents a 2-stop improvement in high ISO quality. Note that noise-processing left a lot of color-noise back then, noise is mostly luminance now. Incidentally, people are more disturbed by digital noise than film noise, because of its color-component.

The above crops certainly show that there was excellent progress relative to image noise at higher sensitivities. Probably the second most important image quality factor is dynamic-range. Just like with image-noise, failing to capture a certain part of dynamic-dynamic range makes it unrecoverable by processing, although certain cameras capture more dynamic range than they store in JPEG images. While Fuji takes this problem more seriously than anyone and has designed sensors to address this problem directly, most modern cameras implement a software approach to deal with dynamic-range. Nikon calls it Adaptive D-Lighting.

The big question is where does that dynamic-range come from. There are two answers:

• It does not come from anywhere because no dynamic-range is actually added to the image. This is exactly like using the brightness, levels or gamma settings in Photoshop or another image-processing software. The dynamic-range in the image does not change but is moved around which can make shadow details more visible.
• It comes from the sensor's RAW output. All digital cameras produce JPEG images by transforming the RAW sensor data into an 24-bit color-space. While this is done, the entire dynamic-range of the sensor is not generally used. If it was, many output imagesmostly the ones with low dynamic-range subjects, would look dull and washed out. Adaptive D-Ligthing, like Sony's DRO works by adjusting the RAW to JPEG transformation to include a little more or a lot more of the sensor's dynamic-range. Some cameras even have an Auto mode that guesses how much dynamic-range should be brought into the JPEG.

Although we do not have way to measure dynamic-range, it seems from tests of various scenes and targets that the P7000 and A2 have nearly exactly the same dynamic-range, at least to a 1/3 EV difference. The A2 does exposure with more highlight headroom and the P7000 leaves more shadow headroom. The Adaptive D-Lighting feature improves things with up to one full-stop more shadow details when set to HIGH.

Speed has been a concern for digital cameras since the beginnings. Although top speeds and shooting rate have increased, speed is still concerning for too many cameras. The smallest ones particularly suffer this problem. DSLRs are fast enough for most uses, action photography keeps pushing the limits. There are many factors that contribute to camera speed: shutter-lag, autofocus speed, shot-to-shot speed, burst rate and responsiveness of all functions.

Over the last few years, huge strides were made with continuous shooting speeds due to the use of CMOS sensors. In this respect, the A2's 2.8 FPS is left in the dust by the likes of Casio Exilim EX-FH25
Casio Exilim EX-FH25 which can shoot at up to 40 FPS using an electronic shutter. Cameras like the Panasonic Lumix DMC-FZ100
Panasonic Lumix DMC-FZ100 can shoot at up to 11 FPS using a mechanical shutter. The P7000 is unfortunately not a good example of progress here because it only shoots at 1.3 FPS, it can do so for up to 45 JPEG images thanks to better I/O. The reason is that it still takes a CCD to reach the highest-levels of image quality but a CMOS is required to reach top burst-rates. For this reason, the advanced compacts and even medium-format cameras use CCD sensors.

Arguably the most important aspect of camera is speed is the shutter-lag. That is the time it takes, after focusing and metering, for the camera to react to the shutter being pressed. Between the A2 and P7000, no difference can be noted. They both take about 0.1 seconds to take the picture from the half-press position. There are some new models that claim to do better though.

Well, before going from the halfway point to the full-press, one must wait for the camera to be read. The time it takes there is dominated by the focusing-speed. This probably an area where there has been the least progress among compact cameras. Focusing using contrast-detection is actually a complex operation and requires interaction with the motors that move lens-elements. Even DSLRs which use it in Live-View are slow at it! After measuring focus-speeds of both the A2 and P7000, suspicions were confirmed. Both these cameras focus at roughly the same speed, although with an edge to the older A2. Wide-angle focusing is between 0.3 and 0.6s for the A2 and between 0.5 and 0.8s for the P7000. Telephoto focusing is always slower with up to 1.5s for the A2 and up to 2s for the P7000. The A2's focusing-system is absolutely more reliable than the P7000.

It certainly does not look good for focusing and responsiveness improvements. Casio has been pioneering advancements here with their CMOS-based cameras. One simulates zero shutter-lag by using pre-buffered continuous-shooting. When the shutter is pressed, it simply stores the photo which was already captured when the shutter was pressed. The obvious catch is that any photos taken before autofocus was locked is pretty much useless, so one still has to wait for autofocus.

Shot-to-shot speeds determine the speed at which you can try again when you miss a shot. Shots, particularly of people, are easy to miss and expressions can be fleeting. For this reason, shot-to-shoot speed is quite important. A camera's performance here often correlates to its continuous drive speed, one is basically an automation of the other. So, there was not much surprise when we measured 1s shot-to-shot speeds for the Konica-Minolta Dimage A2 and 2.5s for the Nikon Coolpix P7000. Although this is the case here, CMOS sensor cameras probably do better in this area. Other types of responsiveness have not progressed much either, most of the A2's menus and buttons respond as fast or faster than the P7000's, with the exception of panning and zooming in playback mode. The A2 can also capture TIFF files which causes it to lock-up for 25s! Memory bandwidth clearly has progressed these days. It is one of the major factors which enables HD video.

Years of changes among digital cameras have passed yet a lot remain unchanged. We've used the Nikon P7000 and Konica-Minolta A2 as examples due to the similarity of their specifications and their flagship status at the time, but a lot applies to cameras in general. The last six years brought great improvements in terms of image-noise at higher-sensitivities and much more advanced video capabilities. The viewing experience among cameras has also changed with much larger LCDs, up to 3.5" on some models now. Image-processing is much faster and more capable now, giving rise to things like better automatic white-balance, face-detection, distortion correction and more.

There were apparently neglected areas too. Speed, other than burst rates are seemingly unchanged. Our example cameras actually show a reversal here with the newer model focusing a little slower and showing much slower shot-to-shot speeds. On the ergonomics, certainly Konica-Minolta new what they were doing when designing the A2 which to this day has no equal in usability among fixed-lens cameras. The efficiency of controls is all about how long it takes to set up the camera for the next shot. There is hardly anything to criticize about this among modern DSLRs but the strive for smaller compacts with limited feature-sets have caused some regression among compact cameras.

As advanced users we must urge camera makers to emphasize operating a camera above gimmicky features. We must realize though that price is an important factor and that the Konica-Minolta Dimage A2 was quite an expensive camera to manufacture with all its buttons, dials, moving parts (both LCD and EVF tilt) and sensors for both eye-proximity and grip-holding. Software features like distortion-correction or art-filters are essentially free to manufacture. What has changed today is the price of DSLRs puts a difficult barrier to break for advanced compact cameras. Very few fixed-lens cameras can sell above the price of an entry-level DSLR.