When I first heard about the impending release of the Canon 5DS I knew I had to have one. Those of you who have followed me here know of my love for large prints and the ultimate quest for resolution in order to achieve those big prints. In the film days my go-to cameras for serious work were my 4×5 and 8×10 view and field cameras while my commercial and portrait work was done almost entirely with those cameras and medium format. I rarely if ever used 35mm cameras. So since getting into the digital world I have always been on a quest to recreate the quality of those cameras.
Alas, on a Professor’s pay, a medium format digital system was simply out of the question and there were too many issues with the new Nikon and Sony offerings in the 30+ megapixel world to make me willing to have to buy into a whole new system for not all that much gain. So, following in the footsteps of Max Lyons and Paul Taraba, I settled for the equipment I built or purchased such as the “digi-view” and most lately the Rhinocam™ as well as with multi-row panoramas called “Mosaics” using spherical panorama heads to improve resolution.
But when we talk about “resolution” what are we really talking about? In order to tell if this new camera attains my resolution goals I need to define what is under consideration and what all in the “system” really has an influence on it. So this first part is mostly about what is this “resolution” I’m seeking with a couple of new shots, then the 2nd part will be mostly some in-the-field shooting.
“Resolution,” it turns out, is not as simple an issue as it might seem. The final resolution that shows up in your photograph is a result of a large number of variables, each having an impact on the other. The camera’s potential resolution is but one part of the equation. In the digital world, the regular 2-dimensional tabular array of photo sensors, creates issues far more complex than simply the number of them that can be crammed into a given space. It was the same with film.
Film grains, the actual grains of silver halides in an emulsion, are, in a very fine grained film, about 2 microns in size and they are arranged randomly and 3-dimensionally through the emulsion layer. But depending on the film/developer combination, those actual grains form clumps of from 10 – 40 grains, each clump of like grains, either exposed and developed and therefore black grains, or insufficiently exposed and/or developed grains which are removed by the fixing agent leaving their spaces “clear” (apart from the base fog on the film). Those clumps, based on developer, can have hard edges or soft edges which will have an effect on the apparent resolution via its “acutance” or sharpness.
Normal DSLRs of 8 to 20+ megapixels have photo sensors of generally 6 – 12 microns. (Yes, yes, I know; for all you science techies out there, the term “micron” is in disfavor for the more approved “micrometer” but in photography the use of “micron” persists and is understood, as is its symbol, “µ”). Interpolation reduces the actual potential resolution by about 1/3 – ½ and the anti-aliasing filter further softens the image slightly reducing the apparent resolution in a way that looks similar to a low acutance developer (though in fact is a very different issue). The size of the photo sensor influences its inherent sensitivity to light but that is a different issue. What is important to us here is how that can effect their proximity and therefore how much dead space is there between photo sensors? What happens to the light that strikes between sensors? Micro-lens arrays are now used to collect this light and re-aim (refract) it into the photo sensor. But this last moment re-alignment of some of the light has an effect on final resolution so the closer the photo sensors are to each other the better.
The 5DS has 50+ million photo sensors that are 4.1 microns in size, according to the Canon specs ( for comparison a human red blood cell is about 5 microns and an average human hair is about 25-75 microns (actually anywhere from about 15 microns to over 200 microns in diameter). The “R” model I ordered does not have the anti-aliasing effect which would soften the resulting image. The sheer numbers means they are very close together and the micro-lens array has less extreme refracting to accomplish. So apart from the loss due to interpolation the resolution should be very good, essentially 2X that of its best brand mates.
But that is just part of the resolution picture… so to speak. Enter the issue of the lens. As complex as the issue of film/sensor in the camera/film arena, the variables in lenses make that seem simple by comparison. In fact a mismatched lens can render sensor resolution pointless. In a lens everything contributes to the final resolution. The housing material, the gearing, the mount, the lens elements, the lens groups, the adhesive, the movement of those groups and elements, the chemical make-up of the glass itself, the arrangement of the groups and elements, the polish and finish (the so-called “scratch and dig” specifications), and, of course, the coating(s) ALL have a say in the lens’s final resolution. And when that lens is now aimed at a digital sensor, even the angles of the light rays will influence resolution.
Additionally resolution is more than an issue of how fine something can render detail (usually measured in Line Pairs per Millimeter or LPMM), it is also heavily effected by contrast. True black and white line pairs are easier for us humans to resolve than the same sized line pairs that are shades of gray. This combination of detail and contrast “resolution” is represented in graphs called MTF (Modulation Transfer Function) curves that show the effect of contrast and also of lens distortion and aberrations that occur in ALL lenses. But even THAT is potentially misleading unless you read and understand the fine print and manufacturer’s spin on the data. What is being “measured” and analyzed is light transmission and how well the lens can focus the individual light waves.
As the lens is focused, individual points of light from the subject are represented as tiny circles (since no lens can actually render a point of light as a “point”) which are called “Circles of Confusion.” Film is indifferent to the size of those Circles of Confusion and simply shows less and less “sharpness” or less apparent focus as the circles get bigger. If the circle of confusion on the print being viewed is 1/200 of an inch or smaller most humans will see it as sharp but if bigger will see it as soft or out of focus (and it is from this illusion that Depth of Field comes in to play). But on a digital chip, if the circles of confusion are bigger than a photosite, and some of that projected point of light, consisting of the 3 wavelengths of RGB light creating that point’s specific hue, spills over into an adjacent photosite, the result is chromatic aberrations and a loss of apparent resolution.
So it ought to be simple to measure that, right? Well, it depends. Are you measuring the projected “resolution” with the lens on a camera, and if so WHICH camera since minute variances in the precise placement of the lens mount, the imaging chip, etc. can have an influence on the results. MY camera may actually have a slightly different response to a given lens than YOUR camera of the same make and model. And ANY influence on that projected image is negative. That influence will show up faster with a decrease in the size of the photo sensors; for example many lenses may look great on a camera with 6-8 megapixel chips but be far less satisfactory on a camera with a 20 megapixel chip.
So often manufacturers measure the projected image with the lens dismounted. There they are measuring an “airy disc” or the images projected through the air against a target but without any issues of photo sensors or film grains/clumps. That eliminates all issues of variations in cameras and also variations in lens mount placements. It does show the lens’s real potential but not necessarily its actual performance mounted on your camera.
Bottom line is twofold: the smaller the photo sites the more critical is the issue of lens performance; and the only performance that really matters is that of a specific lens on a specific camera body. There are many technical forums where “experts” argue over which is more important: lens resolution or camera resolution. The bottom line seems to come to this: It is camera resolution that has the greatest effect BUT a mismatched lens can seriously reduce that effect.
However, Photon, the “god” of photography, is capricious and for every good thing sent your way something negative will be there to match it. For all of the wonders of resolution gain, because it is created via smaller photo sensors (unlike true medium format digital where the chip itself is bigger and has more room for more photo sensors) any – ANY – movement during exposure that would cause a projected circle of confusion to swipe across photo-sensors, will cause aberrations and loss of detail and contrast, i.e. loss of resolution.
So, final bottom line, we have a small format camera with medium format resolution that needs to be handled like a large format camera.
Resolution is all about not only the sensor capability but also the match between the sensor and the lens. In the end we are talking about detail, but we are also talking about having that detail easy to see by us humans.
OK, Back to the Story
Whew, anyone still awake? So I pre-ordered the 50 megapixel beast and also ordered the 5D SR model for that extra measure of sharpness. Slated for delivery in early to mid June, the days ticked by at glacial speed. I wanted to get it and wring it out before the Bristlecone Pines trip in early July. You’ve heard the old adage about a watched pot never boiling? Well, I can tell you, a watched calendar or mailbox is just as resistant to getting on with it.
But finally the day arrived and at my door was THE long awaited package. I have quite a few lenses I want to test with this camera, and I certainly want to test it on the Rhinocam and Digicam rigs. So where to start? I will be headed out to shoot this weekend but of course I could not just sit and look at the new beast sitting proudly in my case… But what to shoot?
Well, it turns out I had something I needed to shoot and this would be perfect for it. On my last trip to Santa Fe I had ordered a custom hat from the O’Farrell hat makers and to set it off, a custom hatband from my old friend and master silversmith and engraver, Jim Neely and I promised him pictures of it in place. So I set them up out on my patio with some “props” to fill in the blank space (I sometimes can’t help being me…), and went to work. These shots are all natural afternoon light through the wrap-around patio windows with some reflectors and flags to control the light. I’ve uploaded these and left them fairly large so you can click and enlarge them to see the detail.
The first shot is a side view of the band so you can see the horsehair base and the silver elements (conchos and buffalo skulls) adorning it.
Custom Hat by O’Farrell in Santa Fe showing the custom hat band by Jim Neely, Santa Fe. Shot with Canon 5D SR and Canon 70–200 f4 lens. Click on the image to enlarge. Click twice for full size.
The second shot is a detailed shot of the buffalo skull clasp on the band showing off some of the engraved “drawing” on it.
Detailo shot of the buffalo motif clasp on Jim Neel’y’s custom hatband on my custom O’Farrell hat. The skull is about the size of a US Nickle (5-cent piece). Even though small it looks like it took two tiny shots to bring it down… Shot with Canon 5D SR with Tamron 180mm f3.5 macro lens. Click on the image to enlarge. Click twice for full size.
Cynthia found the little feather near the Acoma Pueblo during Spring Break. It seems like a perfect addition to the band.
Both shots were from a tripod using the cameras shutter’s 1-second delay setting since I had left my remote trigger in the car. The files convert from RAW into 287 Mb PSD files at 16 bit, full size, Pro-Photo files. Fortunately Canon kept the same RAW format as with the 5D MkIII so I did not need to scrounge around for an updated converter (although it comes with Canon’s own RAW converter). These images are at 100 ppi to make them at least semi-manageable for online transfer and viewing; at full resolution of 300 ppi they are REALLY impressive. It showed me that those two lenses work well with the camera but they were always sharp, especially the Canon 70-200.
It will be wild to do a mosaic with the Rhinocam which I expect to do this weekend using the Hasselblad-Zeiss lenses.
Meantime I’m really impressed with the detail and with the tonal smoothness from the 5D SR’s 50.3 Megapixel sensor. I’m pretty sure I’ll be keeping it!
Addendum… There are lots of custom hat makers in the world, but after looking into a lot of them, and having a hat rack full of hats, Kevin and Scott O’Farrell in Santa FE are among the best in the business… maybe the best. They have this antique device that looks like it is from a medieval torture chamber that is used to not only measure the head but to precisely define the shape at the hat line. This is the very first hat I’ve ever owned, including some custom ones over the years, that I did not have to “train” to my head. Their stuff is not cheap by any stretch of the imagination but if you want the best, check them out (www.ofarrellhatco.com), then give them a call, ask for Syndi, and tell her I sent you.
And when it comes to silver work, Jim Neely has been among the best anglo smiths for decades. He and a half dozen of his compatriots, anglos all, were credited in the early ’70s with reviving many of the old techniques and style of the southwestern Native Americans that were being slowly lost through non-use. That is not politically correct these days but I was there and watched it happening; the truth simply is what it is whether it steps on someones toes or egos or not.
An old “free trapper” in spirit and temperament, Jim epitomizes the old time hand craftsmen/artisans/artists. He is also from ranching stock and we were in art school together in Colorado in the early ’60s — we we go WAY back. Those who know me have actually seen a lot of his work including my bracelet, a buckle, some lapel/hat pins, a scarf slide, and now this hatband. He does not have a computer but if you were thinking of a custom piece I could connect you. He is currently working on a megabuck hatband featuring 18K gold elements for a high roller and is often swamped with work since he refuses to mechanize his approach. But I can tell you the wait is definitely worth it. He designed and made a ring specifically for Cynthia that is beyond stunning!!! I’ve seen a lot of southwestern and Indian style silver in my time but never anything quite like that ring…
So in a few days I should have Part 2 of the 5D SR reviews done.
Travels with Rocinante 