You've Got to be Kidding!
No – I'm Not
Needles – Algonquin Park, Ontario. October, 2008
Canon G10 @ ISO 200
18mm (100mm equiv) – 1/40 sec @ f/4
In mid-October, 2008 I found myself conducting a five day One-On-One workshop in Algonquin Park and the Muskoka region of north-central Ontario with a client from Ireland. George was shooting with a Sinar medium format digital system while I was shooting with my Hasselblad H2 and Phase One P45+ back. When I had a few free moments I was also testing the brand new Canon G10 and Nikon P6000 pocket digicams for an upcoming comparative review.
I had become very impressed with the Canon G10 after just a few days of earlier light-duty testing. Each evening that week I would sit with my 15" Macbook Pro reviewing the day's files. At one point I found myself looking at raw files on-screen and not being sure if I was looking at Hasselblad P45+ files or Canon G10 files. That includes at 100% onscreen enlargements.
Now, I'm no newbe. After some 50 years in this industry I know what I'm looking at, be it a screen blow-up or a print, and I certainly don't confuse how something looks on a 15" laptop screen (though properly profiled and calibrated) with how it will turn out on a critically produced exhibition-quality print. But nevertheless, I was curious about what I was seeing. In fact I was more than curious, I was somewhat amazed.
In The Field
The next afternoon I hiked into the forest with my tripod, Hasselblad H2 and P45+ with the Hasselblad 55-110mm lens attached, and photographed a lovely deep forest fall scene. As I stood there wondering where to go next I put my hands in my vest pocket (it was a chilly fall morning), and there was the Canon G10.
Humm! I wonder how the it would compare with the H2/P45+ combo in a critically controlled side-by-side comparison. Nahh! Silly idea.
But – nothing ventured, etc, and so without a tripod mount for the G10 I simply held it firmly braced on top of the Hassy and did an exposure, framing the shot as closely as I could to what I'd done on the Hasselblad.
That evening I looked at the files on my laptop screen, along with several other people, and we were amazed to see that the differences between the 39 Megapixel medium format system and the 15 Megapixel pocket digicam didn't seem that dramatic – certainly not as big as one might have expected. In fact, with the files in Lightroom I usually couldn't tell which was which without pixel-peeping.
Not having a printer on-hand, I decided to put any further evaluation on hold until I could do a critical processing of both images with my desktop system and 30" Cinemadisplay, and then make comparison prints at my studio.
The H2 was on a large Enduro tripod with RRS head. Mirror lock-up was used, along with 3 second self timer and cable release. Shots taken were critically focused by eye as well as with autofocus. White balance was achieved through the use of a gray card, and prime focus was on the large knot in the foreground tree.
The G10 was hand-held, carefully braced atop the Hassy on the tripod. I had no mounting plate for it, so couldn't use the tripod directly. Also, the G10 has Image stabilization, so I guess that evens things out a bit. Autofocus was used, as was autoexposure in Aperture priority mode. Overall if there was any technical bias in the test it was in favour of the medium format system. Both shots were taken at or close to each camera's optimum apertures to avoid quality loss due to diffraction.
At The Studio
Each file was processed from raw to the best of my ability. Neither file was processed in either Canon's or Phase One's software, but rather in either Lightroom 2.1 for the P45+ file or a beta version of another raw processor, which I prefer to Canon's software. I suppose it would have been more appropriate to use each maker's software, but I used what I prefer to work with, and in any event, the differences would likely have been a quibble.
I tried to match colour as best I could, but I didn't really sweat it too much. Different sensors reproduce colours differently, and without producing custom profiles there's not much to be done about it – and the difference wouldn't have materially affected the results in any event. I simply tried to make each print look as good as I could.
Hasselblad H2 with Phase One P45+ back
55–110mm lens @ 70mm
1 second @ f/11 at ISO 50
The above shows both files, sized to be roughly comparable, as one might see them side by side on a print viewing station.
Doing any comparison such as this is fraught with potential pitfalls because of the inherent differences in the file size and resolutions of the two cameras. The screen grab immediately above, taken from my 30" Cinemadisplay, is intended to show you how these files compare in Photoshop when imported at the same screen resolution. In other words, what you're seeing is 15MP vs 39MP translated into image size rather than resolution.
On The Viewing Stand
Over a two day period I invited photographers and local industry professionals to come to my print studio and look at a series of 13X19" prints from an Epson 3800 printer made on Ilford Gold Fiber Silk paper which were then hung side by side on my floor-standing print viewing box. This collection of seven people included experienced photographers, people from the commercial print industry, and other trade professionals. Between them there was at least 200 years of photographic industry shooting and printing experience.
In most cases I did not tell them what they were looking at, simply saying that I had been shooting with two cameras, and that they should divide the prints (about a dozen) into two piles – Camera A and Camera B. They were asked to judge resolution, accutance, colour reproduction, highlight detail, dMax, and any other factors that they wished to consider.
In every case no one could reliably tell the difference between 13X19" prints shot with the $40,000 Hasselblad and Phase One 39 Megapixel back, and the new $500 Canon G10. In the end no one got more than 60% right, and overall the split was about 50 / 50, with no clear differentiator. In other words, no better than chance.
In fact it was the H2 system's narrower depth of field that occasionally was the only clear give-away. Some viewers eventually figured out that the prints with the narrower depth of field were from medium format, while other photographers chose the G10 images because with its wider depth of field it created an overall impression of greater sharpness.
Needless to say there was much shaking of heads and muttering. Could this be? Could a $500 digicam equal a $40,000 medium format digital system in image quality, at least in prints up to 13X19" (Super A3)?
Frost –Muskoka, Ontario. October, 2008
Canon G10 @ ISO 200
25mm (125mm equiv) – 1/60 sec @ f/4.5
The Take Away
Here's my analysis – for what it's worth. The top end of the industry has achieved very high image quality for some time. The improvements that we've seen over the past several years from medium format backs have slowed, though each new generation has given us not just more megapixels but also wide dynamic range, better high ISO performance, and other advantages. But, the high end has now gotten so good, that the incremental improvements achieved with each new generation are smaller each year. In other words, the graph has flattened.
On the other hand low-end cameras have had a long way to go. In the past they have been noisy, the lenses on digicams have not been the best, and resolution as well as other aspects of image quality have often left a lot to be desired.
Each new generation of cameras though gets better than the last, and with the Canon G10 that company appears to have taken a significant step forward from its predecessor the G9, and for that matter in my experience to any other comparable camera on the market. The sensor and lens used in the G10 offer a marked improvement over anything comparable that's come before, and it shows.
To Be Clear
Let me be clear though, this comparison is not by any means definitive. It was done for my own edification because I was having a hard time believing what I was seeing. Now that I have had my observations confirmed by several industry pros, I am more confident in what I'm seeing, and that those funny cigarettes that I smoked back in the '60's haven't totally befuddled my judgment.
Please note that what I'm describing here is really not new when it comes to comparing high-end 35mm DSLRs to medium format systems. We've all done such comparisons for years, and know that the advantages of large sensors and MF systems are best seen in large prints and in critical applications. The only thing that's different now is that instead of comparing an MF system with a DSLR I'm comparing it to a digicam, though a 15 megapixel one to be sure.
Be aware as well that these comparisons fall down when prints over about 13X19" are made. Once the output resolution drops below 200PPI the advantages of a 39 Megapixel sensor over a 15 Megapixels sensor become evident. And, even when smaller prints are made, cropping becomes an issue.
Also, though on prints up to 13X19" differences are almost impossible to see, on-screen at 100% one can fairly easily tell which files are from the G10. There are artifacts visible at the micro detail level and one can easily see other hints of what one is paying for.
But, where the rubber meets the road (or more to the point where the ink hits the paper), in medium sized prints it's been almost impossible for experienced photographers who I've shown these comparison prints to to tell the difference. Scary.
One final comment. Landscape and nature shots are one thing – models in a studio with fabrics, delicate skin tones and other challenging subjects are likely to be quite another. Also, I have no idea how well these files might hold up to CMYK conversion. We therefore need to keep expectations within reasons.
But, with all of these caveats, the take-away as I see it is that the new Canon G10 has crossed a threshold; one in which an inexpensive pocket camera can produce very high quality images, at least on moderate sized prints, which is what most photographers end up making.
Will I be selling my Hasselblad and Phase One back? No, of course not. Why would I? Each system has its place and specialized function. Indeed I'm really excited about testing the new Phase One P65+ and to acquiring one as soon as it becomes available. But, the next time I take a walk in the woods, or go on a family vacation, I know which camera is going to be along for the trip, nicely tucked away in my jacket pocket. The Canon G10.
Oh Yes – One More Thing
Please don't write to me asking whether I think camera X or Y is as good as the G10, or better than it, or how any of these might compare to a Hasselblad or Phase back. That's not what this is about. The point of all of this is simple. As the industry matures the low end is improving rapidly while the high end's improvements are slowing down. This is narrowing the gap, and that's good news for all of us. Don't read too much more into it than that.
Birch Tree. Algonquin Park. Ontario. October, 2008
Canon G10 @ ISO 80
18mm (100mm equiv) – 1/13 sec @ f/4
The interesting side note about this shot is that it was taken hand-held at 1/13th of a second.
The combination of built-in stabilization and small focal length lenses means that slow shutter speeds –
much slower than those possible with 35mm, and certainly medium format, are possible,
making hand-held shots at low ISO feasible, leading to higher image quality than expected in poor light conditions.
I'd love to invite you over to my gallery to see these prints, but alas that isn't possible. So instead I'm going to suggest that if this little experiment interests you that you make some prints for yourself. I am therefore making the files shown for this article available for download. These files are JPGs at maximum quality setting, and are rendered in ProPhoto RGB colour space. Each has been processed the way that I normally would for printing, including the usual adjustments of tonality, colour and the like. They have been input sharpened using Lightroom 2.1, but they have not yet been output sharpened for printing. These are NOT the raw files and each has also been slightly cropped for esthetic purposes.
Please note that these image files are intended for personal use by the viewer and may not be displayed, reproduced or used in any manner, whether in print or online, other than for the private use of the person doing the download. These files are not public domain, and are Copyright 2008 by Michael Reichmann.
Each of these files is between between 10 and 18 megabytes in size, varying depending on content and JPG compression applied.
The anticipated bandwidth that will be used by tens of thousands of people downloading 70 Megabytes of files during the first 30 days is in the order of many Terabytes. This is expensive for us, and though we are not charging for these downloads, we would appreciate it you would visit our online store and consider purchasing one of our download tutorials as a way of expressing your support for this site. Thanks.
My full review of the Canon G10, as part of a comparison with the Nikon P6000, will appear on these pages by the end of October. Watch for it.
I'm not a moralist, in that I don't believe that everything is here to teach us a lesson. But there is a lesson to be learned from this tale.
In other pursuits, let's take audiophile equipment for example, we have long known that it takes larger and larger amounts of money to achieve smaller and smaller incremental gains in sound quality. Once you reach a certain plateau, the curve gets very flat, very quickly.
This is now happening (as it was inevitable that it would) in the field of digital photography. Up until the last year or so it meant spending $5,000 to $8,000 to get the best DSLR image quality. The next step above that were medium format backs at $15,000 to $30,000. Improved image quality to be sure, but at a serious cost.
Now, within the past six months, we have seen that trend accelerate more than ever before. The Nikon D700 brings the D3's stunning image quality to a camera costing half as much as before. Shipping in November will be the Canon 5D MKII, which promises to offer image quality comparable to the 1Ds MKIII at a quarter the price. This trend is inexorable!
So why should we be surprised that the just-released Canon G10 does something similar – pushing the price performance ratio even further downwards. As Ray Kurzweil has illuminated in The Singularity is Near, once Moore's Law states to kick-in in earnest the rate of change becomes almost exponential. That, my friends, is what we're now starting to see in the camera industry.
We can now find DVD players at the check-out counter at Best Buy for $20. Imagine what the price is leaving the factory. That $20 DVD player's retail price includes components, manufacturing, R&D, packaging, documentation, licensing and royalties, shipping half-way round the world, import duties, and retailer margin. Only a few years ago DVD players cost $1,000. Now they're $20. The camera industry isn't going to be much different soon.
The lesson here, especially for newbies and amateurs (the pros have always understood this) is – stop fussing over each new camera's image quality. Now, even sub-$1,000 cameras are able to produce exceptional image quality – likely far better than most photographers are able to execute. Buy high end gear if you need it or want it. But, don't think that the better camera is going to make you a better photographer. As I've written before – most cameras are better than most photographers. This has always been true, and now simply more so than ever.
And finally, to my friends in the industry – I intend no disservice or harm by this comparison. I am not saying that a Mercedes and a Honda Civic are comparable, but I am saying that for some people they'll both get you where you're going. The photographic industry is now the computer industry, and that means that Moore's Law largely holds sway.
It may appear at first read that I'm taking a swipe at Phase One, or Hasselblad, or the medium format industry in general. Trust me – I'm not. In 2001 I wrote that the 3 Megapixel Canon D30 produced files competitive with scanned 35mm film at smaller print sizes. I was derided and vilified then, but of course was ultimately proven correct – digital simply outperforms film, and few people today think otherwise.
Now we have a situation where pocketable digital cameras – at least so far in the form of the Canon G10 – have come of age and are able to produce professional quality images in some situations. If I don't point this out someone else will, and this will happen sooner rather than later. So be pissed off and disagree if you will, but I can't do otherwise than state what I see, and what I know other knowledgeable observers will soon be seeing as well.
It's been said that if Moore's Law was applied to automobiles we'd all be driving Rolls Royce's that cost $10 and got 1000 miles to the gallon. As a more relevant analogy, I bought a one Terabytes drive the other day for less than $200, while it wasn't that many years ago that a one Gigabyte drive cost $50,000, was the size of a refrigerator, and needed to be housed in an air conditioned room. Thats an increase of 1000:1 in storage capacity combined with a reduction of 250:1 in price.
Why should we therefore be surprised when similar improvements are seen in the photographic industry, which is now as much about silicon as anything else?