Friday, September 6, 2013

Canon EOS 70D Review: Part 2 - Where do We Go From Here?

As previously mentioned, the Canon EOS 70D is a well-sorted camera that is competitive with its nearest rival, the Nikon D7100. Much has been made of the 70D's Dual Pixel phase detection architecture, but one of the questions that isn't being asked is if this technology is actually revolutionary, or if it's merely very good. Here's some food for thought:

  • The Olympus OM-D M5 focuses faster with contrast detection when taking stills. A lot fast actually.
  • The Nikon V1 and Fujifilm X100s also use PDAF, but without the added complexity of having double sets of read-out interconnects in the sensor bed. 

These are actually shortcomings against other on-chip phase-detection technologies, but they are also red herrings. The OM-D M5 and the PEN E-P5 are ridiculously fast at single-shot image-taking, but the 70D, though a tad bit slower, does better with tracking moving subjects. The other red herring is speed: we're at the point now that most continuous PDAF systems track faster than what is reasonable for well made video. (Hint, pan slower than what you think you need to, your viewers will thank you for it.)

In the most obvious ways, the 70D sensor is a success, but it's a success that reminds you of the skeleton in Canon's closet: the EOS M.

Ask any retailer... the people who actually sell the cameras... what they think about the EOS M in North America and you'll get something along the lines of indifference mixed in with an exasperated eye-roll. The Hybrid CMOS technology didn't live up to its promise, and what's worse, wasn't significantly improved by a firmware update released in the summer of 2013.

via Canon Inc.
If you read into the original literature, the Hybrid CMOS AF worked by using the phase detection elements to quickly ascertain distance information to obtain general focus, after which the contrast-detection ability of the sensor took over in order to bring the picture into fine-focus. Sounds great in theory, but the problem is that the CDAF routine added to the final part of the focusing operation will slow down the process considerably. However Fujifilm and Aptina managed it, their solutions work on a similar principle, only faster.

That the 70D performs better than the EOS-M isn't necessarily just an indication of better technology, but a more mature understanding of its application on the part of Canon's engineers. Even though they go about it in different ways, the basic principle behind the PDAF operation in the two cameras is the same. The only difference is that the 70D is faster, and the entirety of the sensor area can be used for phase-detection. That's a pretty significant jump in capability. However, the downside of the Dual Pixel architecture is that the complexity of the chip is increased. Each photodiode requires two read-outs as opposed to one in a conventional chip. This inevitably results in some blockage of light to the substrate, though it can be mitigated by improved manufacturing processes (thinner interconnects, copper fabs instead of aluminum, etc.) is reporting a quantum efficiency figure of 45% for the 70D. This is an improvement over the 60D (40%), but is still less than the Nikon D7000 (48%) and the Nikon D7100 (52%).

One wonders why Canon could never get their original on-chip PDAF technology to work as well as the competition, but given the lead-time necessary in developing sensor technology it would be safe to presume that the 70D sensor was in development at the time the EOS-M was launched. Yes, they launched a product knowing full well that it would be inferior on the market with the knowledge that something better was coming down the pipeline. Welcome to the world of product marketing; sometimes you hit it out of the park, other times you move on to the next thing and hope the sales department is as good as they say they are. The easier thing would have been to replicate the Fujifilm/Aptina model, with a limited number of PDAF elements on chip and the rest of the sensor being conventional. There might be patent issues with that approach, which is beyond the scope of this blog. However, the Dual Pixel technology is a more ambitious approach, and ties in better with Canon's long-terms strategy of developing a coherent videography experience in with their DSLR's.

Looking forward into the immediate future, it's not hard to see that the next EOS-M will use the 70D chip. The buzz that the 70D has gotten will give a boost to the next generation of the EOS-M should this sensor be used, and will remove some of the memories of the first generation. In fact, based on past behaviour, Canon will be introducing this chip into as many cameras as it can. From a production standpoint, this is a fiscally sound strategy as it spreads the development and manufacturing costs of the new chip across a broad range of cameras. The downside is that three-four years from now, the whole line-up will look a bit long in the tooth if the same sensor is still in use. To put into perspective; the 7D was launched in 2009; that 18mp sensor is still in use today, whereas Nikon has move from 16mp to 24mp with AA filter and then 24mp without AA filter. Regardless of whether or not this is relevant to real world photography, it much have been an irritant to Canon's marketing department. However, in terms of real-world usability, a 20mp Dual Pixel version of the T5i with great video is a better value for the budget/compact DSLR shooter than a the 24mp great-stills-only Nikon D3200.

Looking at more blue-sky application, a Dual-Pixel version of  the 5DmIII would fit in with the legions of wedding photography studios. The Mark II brought cinema-quality videography to the masses such that shallow DOF high-production videos is a staple of many studios today. The 70D would make for an excellent documentary camera, but for the added gloss that goes with the commercial wedding market, the larger full-frame format gives that extra bit of oomph to justify wedding-day prices.

However, most people think that we have reached "peak-DSLR" and are now on the back-half of the course, mirrorless being the next big thing. Canon is now certainly poised to enter the era of "big mirrorless" with the 70D sensor, but I would caution against looking for too much change. First of all, both Canon and Nikon are heavily invested in DSLR's and would rather sell you that than something else. The second point is that both companies have significant muscle in terms of retail shelf space; if m4/3 has done as well as it could have in North America, this is also a reason why. If for whatever reason the DSLR market completely collapses in the next two years (highly unlikely and don't stand where the comet is presumed to strike oil), at the very least Canon now has an APS-C sensor going forward: Nikon can do a mirrorless APS-C, but they are a step behind in that they currently don't have one with PDAF. That doesn't mean that they wouldn't be able to scale the Aptina V1/V2 sensors to APS-C size; they could, it's just that we haven't seen it yet.

However, as good as the commercial potential for the 70D sensor is, it still doesn't solve the issues that the traditional serious-enthusiast/semi-pro crowd want in a camera; lower image noise and more dynamic range. There's a difference between "good-enough" and "competitive". The 70D is better than the 60D, but it is still edged out by the D7100. If you are the type that wants to keep up with the Joneses, then "good enough" isn't in your blood (not that I actually advocate this as a reason to buy a camera). However, if you earn income from shooting, then the stakes are higher. Yes, you would want equipment that suits your needs, but you also need to have competitive equipment with the studio down the street even if you are a low-volume weekend-only wedding shooter. The 70D is good, and it even opens up creative doors in adding video to a photographic portfolio, but it doesn't address those basic issues of improvement for traditional stills photography. That's why I see two possibilities; either the 7DmII will use a different, more traditional photo-centric sensor, or the category of professional APS-C really is comatose (not dead, I hope!). This goes also for the D400, which has been like a mirage that never materializes for the Nikon camp. Arguably, entry-level pro was the 5DmII, of which you can still find plenty of relatively inexpensive copies, but until the picture gets clearer for the 7D/D400 market space... this isn't the semi-pro APS-C sensor that you are looking for.

No comments:

Post a Comment