The Latest Sensors are Always the Best, Sharpest and Happiest, Right?

I know it's the Holiday Season and as an American blogger I have almost a sacred duty to inflame your lust for a new camera, lens or costly accessory so you will order said unit and I will have a bit more cash in my Christmas stocking from Amazon.com via their affiliate program (which, incidentally, costs you nothing extra...) but---I'm just not in the mood to be mercantile right now because I'm not feeling that it's really that important to buy any of the new gear that's out there  today when there is so much surplus "last season" stuff out there that is almost as good as the new stuff....or maybe better. 

I'm going off on a tangent right now that has very little to do with science and perhaps more to do with the emotion of seeing instead of the quantification side of judging things. As an example it's pretty much a given that on most computer screens (where the vast majority of people ingest and enjoy photographs) a sharp image from a lower megapixel camera will look better than a massive and much higher resolution camera file. Try it yourself with any of the cameras you own and you'll find it's true. Now, if you are using the cameras to produce mondo sized printed posters you'll definitely decide that bigger is better so newer is better. But honestly, how many of us even get around to printing the majority of our images?  If you are truthful you'll admit that mostly you share your images at about 2000 pixels on the long side, right? So while those with a scientific bent can show us that the bigger, newer sensor is quantitatively much better and can equal the on-screen look of the less populated sensor through the process of downsampling or binning it's really just theoretical for our every day use.

Why am I hesitant to rush out and buy a newer, bigger, more specification-glorious camera right now? Well, maybe it's because I'm coming to realize that for my uses those cameras might not be the best choice. The reason I have a photo of the Kodak DCS 760 at the head of the blog today is to serve as a reminder that some of the most wonderful digital portraits I ever took were done with this camera and that I often reverted back to it long after having "upgraded" to cameras like the D2x, etc. because I LIKED THE LOOK OF THE FILES BETTER. I didn't measure anything or go to DXO to get their considered opinion, instead I used a very complex method I learned long ago: I looked at the images. The slow, noisy, CCD sensor in the DCS 760 ( a whopping six megs) made skin tones look wonderful and had a feeling of depth that I don't usually see in the cameras we rush to buy today. 

Several times in the last ten years I upgraded cameras NOT because I needed better image quality but because I needed other unrelated features to make my jobs out in the field more flexible and accurate. I traded up from one camera that had a small, low res, uncalibrated LCD screen on the back to one that had a much bigger screen because it was easier to see (an more accurately) what I might end up with when I brought the images back into the studio to post process them. I upgraded to prevent nasty surprises.

Early on I had a Nikon D100 camera. It was a very nicely done camera. In time Nikon came out with a camera that had more resolution and a better screen but the biggest reason I felt compelled to upgrade was the fact that the weakest point of the D100 had nothing to do with the quality of the files but with the paucity of the buffer. If you shot raw you would get four images and then you would need to pause while the camera processed the files and wrote to the card. It was tedious. I am a garrulous and promiscuous shooter and the small buffer really cramped my style. But in terms of image quality both my D100 and my D2H were better photograph making machines than the D200 ever was, no matter how quickly it was able to whip its mediocre constructs through its internal process. (Can you tell that I loathed that camera? God it was awful. A real example of checking off the marketing boxes with tedious engineering).

Recently I jumped down another silly rabbit hole. I had a client give me a job the parameters of which I thought would be outside the performance envelope of the micro four thirds cameras I was trying to press into service for nearly everything. I did some quick research on the sharpest, highest resolution cameras I could buy that would also be cost effective and have a system that would be (for a portrait photographer) relatively easy financially to slide into. For better or worse the Nikon D7100 was the choice I made. If you need very sharp, very high resolution files I can recommend the camera with very few caveats. It's actually my idea of the best APS-C working camera out there right now.

Well, since the Summer I have used it pretty extensively (but please note that I shoot a lot and also constantly rotate other cameras into the mix as well), so much so that I decided I should have a back up camera of those times when its feature set suggested its use out on a remote location. You gotta have a back up, right? So I went back to my research and after researching every recent Nikon camera body I settled on a used D7000 for my back up. The first one I tried had some massive back focus problems so I returned it, waited a while and then found another one at a decent price. I immediately tested it and then AF fine-tuned all the AF lenses I intended to use on it. Then I took it out for a series of walks to check out the overall performance of the machine. It was nice. It focused as quickly as all of my current cameras. It handled low light levels at least as well as all of my current cameras, including the three year newer D7100. I came to trust that camera as a working tool.

Recently I shot an evening event for the Texas Appleseed Foundation at the Four Seasons Hotel here in Austin. Most of the images would be used on the web and repurposed as five by seven inch prints for attendee/donor gifts. I decided to use the D7000 as my primary "grip and grin" camera for the event mostly because it's largest Jpeg file was just right. A good intersection of high image quality but not so much information as to bloat the files. I used it with flash and the flash was modified by a Rogue bounce apparatus. I stuck an 85mm lens on the D7100 and used it for about 20 % of the shots. When I started to compare the files in post production I found that I preferred the color and tonality of the portraits from the D7000 every single time. 

That led me to start using the D7000 as the primary camera and keeping the D7100 camera as the back up. And that interested me. Was there something about the ever increasing resolution and dynamic range of the ever newer sensors that, while measuring better, is aesthetically at odds at least with my perceptions of what is good? To discover more I called a friend who sometimes sits as a model for me and we made a bunch of tests. Same lens, same light. 

It's hard to quantify the actual differences but I'd say that the D7100 works sheer force. By that I mean it relies for its impressions of sharpness and quality on endless assemblages of endless dots. But it feels a bit muddier than the D7000. The files feel almost to thick for me. The D7000 feels more open and crisp. As I say, it's hard to put into logical words but the D7000 feels more like it's making images to me where the D7100 feels like a lot of the little dots are just wasted filler that ends up giving you more detail as you increase file size but also makes the files seem frizzly and less substantial as you increase on screen magnification. It may just be that my lenses aren't up to the challenge but..... it's all in the way the image looks not in the way it measures.

I also talked to a friend who is in the fashion industry and who shoots with a large assortment of cameras. He shot with Olympus EM-5 cameras among many others and, upon announcement he rushed out and bought an EM-1. He says the EM-1 is a much better handling camera but that he much prefers the colors and especially the handling of flesh tones from the sensor in the EM-5 cameras. I had wondered if my experience with the Nikons was nikon-limited until I heard from him. With this new information (and four EM-5's in the drawer) I decided to borrow an EM-1 and compare for myself. What I found tracked what I saw on the Nikons as well as on the Olympus cameras. The older model had a palette and overall rendering that I preferred. 

But now, in late 2014, the reasons I might have reflexively upgraded in years past are no longer nearly as valid. The finders aren't worlds better. (yes, the EM-1 has a much better EVF) they are better but not in a "make it or break it" way. The buffers are pretty much invisible to me on all four of the camera under discussion. So I have to ask myself, "is the upgrade actually to downgrade?"

In the same way many people thought that the Olympus e1 and a few other 4:3 cameras that used Kodak's CCD imagers rendered a much more pleasant image than the following generation of cameras that used CMOS sensors. Again, the newer cameras measured better  but whether the holistic image was superior is one of those subjective questions that can probably only be answered by the users.  There are non-camera analogies all over the place. The handling of mature generation rear wheel drive cars versus the early generations of front wheel drive cars. The look of finely tweaked tube computer monitors versus the first few generations of flat screens. Old Coke v. New Coke. But what I am seeing in comparing the Nikon cameras really has nothing to do with CCD versus CMOS because both cameras use CMOS imagers and both are imbued with the same color profiles by their makers. What I think we are seeing is an unintended artifact of a more highly populated sensor versus a more loosely populated sensor. And it may be that some people will prefer one over the other while others won't. And there will be a third contingent that sees no differences at all. But we don't care about that group...

So, where did I finally fall in on all of this? I decided that I liked shooting most stuff with the D7000 better than the D7100. I like the look of the files. 16 megapixels is more than enough resolution for everything I shoot and when I want to go over the top I can always trot out the D7100 and put it on a tripod and lock up the mirror. Smaller files means I am more often disposed to shoot raw files or, when already shooting raw files I am more likely to go "platinum level" and shoot them at 14bits instead of 12. I know I won't run out of card space nearly as quickly. 

I liked the D7000 so much I bought one more this week on Amazon. It was new in the box with free delivery for a whopping $525. While that's a bit beyond my "pocket change" category it's a minute price to pay for a back up camera that contains a sensor I like but which will almost certainly be discontinued in all new cameras. And what artists like most is choice. My wife is familiar with the way I think. My belief is that when you find something you really like you may as well buy a second copy because you can be certain that the more you enjoy whatever the product is the more likely it will be quickly and unceremoniously discontinued and replaced with some similar product that, for a myriad of reasons, you will like much less. This is why finding a shirt that looks fantastic and fits perfectly is a logical situation for duplication. If you don't feel that way about your current camera you may not be using the camera that's perfect for you. Or you may be one of those people who believe that all things just get better and better. Tell that to the owners of classic 1966 Lincoln Continental, complete with suicide doors. They'll laugh you out of the garage.

There are always trade-offs. You gain some things in a new camera and you lose stuff. Some of it is handling. Some of it is ephemeral and personal and some of it is subjectively aesthetic = the difference between accuracy and "pleasing" in the rendition of the files. This seems to be the year that I discovered cameras I loved just as they hit the sweetest part of the price curve which means in a few months they'll be gone forever and available only in the used market. Caveat Emptor. And, Be Prepared.

One little non-gear ad: It's for the Kindle version of the novel. 



G Gudmundsson said...

Interesting. Food for thought.

Anonymous said...

Thought provoking article. A question, at what level do you see the differences? Can you see them in prints (8x10 for example), web size like in your blog, for screen size that you would send clients, 100% blown up on screen, etc?

It seems sensor technology is getting to the point of Hi-Fi audio where the quantitative measurements are secondary to the ascetic we see or hear.

Ken A said...

Great post Kirk! I've noticed the same thing. Looking back, "holistically", my favorite images are almost always from the Olynpus E-1 and the 14-54 f/2.8-3.5 or my original Fuji X100 from 2011. The Oly was something special and sometimes I wish I still had it :-)

Ray said...

If I could buy an E-M1 equivalent camera with an E-1 CCD sensor, I'd have one. Whenever I look at my old photos taken with the E-1, I can't help but appreciate the depth and tonality of those shots; everything that followed was so much colder and lacking character.

Mind you, I do like the results from the E-M1, as they are much 'nicer' than the E-5, for example. I think the Sony sensor in the E-M5 is what differentiates it from the Panasonic sensor in the E-M1.

Dave Jenkins said...

Too bad we can't buy duplicates of our wives. . .

Pat said...

I too own the D7000. I always love when someone agrees with me (picture quality). I believe (as Thom Hogan explains) with less pixels, each pixel is bigger thus allowing more photons of light to enter each pixel. According to Mr. Hogan- imagine buckets in the rain. Each bucket is a pixel. Each rain drop is a photon of light. The bigger the bucket, the more rain drops it catches. This is the reason you're seeing differences in images between different pixel counts. Also perhaps the reason the pro Nikon D4 doesn't compete in the Great Pixel war.

I don't work for Hogan but his book on the D7000 which also covers how digital cameras work is easy to understand and a great tool.

Tarjei T. Jensen said...

Buy more of them and wrap them in plastic bags and store them in a freezer.

Then they will keep longer. The reason : electronics might degrade slowly even if you are not powering the devices up.

Patrick Larson said...

Kirk, have you ever used the D700? I have not, but I've seen so many images from that 12mp camera that just blow me away. I love my D610. It's got the same megapixels as the D7100 but in full-frame. Almost the same exact setup. Also, you will notice on the D7000 that the wheel on the left doesn't lock. If you don't watch it you can easily bump it to a different setting. Watch out! It's one of the main reasons I got a D7100.

Kirk Tuck said...

Hi Patrick, I did have a D700 back when they first came out. I kept it for a year but something about the sound of the shutter just drove me nuts and I finally got rid of the camera. That being said the sensor was great and the general look of the files from the camera was very, very good.

Can't stand noisy, mechanical sounding shutters.

Godfrey DiGiorgi said...

Shutter noise doesn't bother me too much, and I can't find a better body to use my Leica R lenses on than a Sony A7 (other than the Leica R8)... :-)

But there's a reason I sold the E-5 and kept the E-1, and would sell the E-M1 before I sold the E-1. And despite the great CCD vs CMOS foofawraw, I've kept my M9 instead of trading up for the M-P, even though the M-P is actually better on every count. I just still like the M9 too much.

In 2015, I intend to go radical and reduce the gear pile by 50% or more. What I keep will puzzle some of my friends, and not surprise others. Because the goal I seek is to do my photography, and my photography seems to require less and less gear with every passing eon of my thought processes.

Good post, Kirk. Thanks.

Joseph Kashi said...

I have had a very similar experience with the Pentax K-5, which uses the same sensor as the D7000, compared with the later K-3, which apparently uses the same sensor as the D7100. The K-5 remains the preferred camera, with better image crispness and contrast.

HF said...

The rectangular rain bucket analogy is quite old. But now assume you just use the water from four buckets and replace them with one (downsizing or -sampling). You will get the same amount of water as you got for the larger buckets. This is why downsizing images increases DR and S/N ratio. The D7100 (newer sensor) seems to do better after downsizing (dxo individual measurement curves) and should lead to more details due to the larger detail and information available for downsizing. At very high ISO, however, the wiring (bucket border) reduces the amount of photons compared to the total number more significantly, and other factors like read noise etc. come in.

Jim Simmons said...

As a man with nine copies of the same shirt (four different colours) and who built models of the '66 Continental as a kid, I could not agree more. The choice of film was similar. It didn't matter that Kodak said T-max was "better," it wasn't tri-x dammit. So I bought a few copies of that too over the years. But if Fuji ever stops making 120 Acros, I'm gonna cry me a river. Thank goodness for my freezer.

Anonymous said...

Interesting article, thanks. I'm just thinking if this could be a consequence of making a project according to the inputs of the marketing department.
It is easier to convince people not so deep into photography to buy a newer more megapixel camera, where the sales man in the shop can give a figure (18 is better than 12!) than convince possible client of the better "holistic" look of a photograph. Justthinking about.

Brad Nichol said...

I own and regularly use a Sony a900 for paid work, increasingly I get quizzed by students and other photographers about my reluctance to upgrade.
Simple, I love the way the files render, especially when combined with some of the older low contrast Minolta glass.
I own other cameras and have used pretty much everything out there, but none give me that almost ananlogue on steroids look that the a900 does when shot at low iso.
The other factor, rarely soken about are the subtle rendereing differences you get via different Raw converters.
Thing is, photographers are quick to jump to the new gal, without actually having developed a great relationship with the current one, there is much to be said in favour of the long term marriage.

Corwin Black said...

Its rather simple. Its about Kodak. Be it SLR/n or Leica DMR, or 760 you owned. They knew how to make colors "right". Its not about CCD vs CMOS. Its more about Kodak vs rest.

And unfortunately "rest" still didnt pick up where Kodak left. Well, maybe except Sony A900. :)

Otherwise, for example Kodak SLR/n (and other Kodak cameras too, probably) had per-pixel color correction matrix to ensure colors arent deviating too much between them.. Also sensor itself was somehow tuned to respond different way to individual parts of light spectrum (much like BW film). And CFA used in those was a masterpiece.

Only remotely similar things from other manufacturers is A900 (A99 isnt bad either after some color profile tweaks) and my old Fuji S5 Pro (Fuji can make nice colors too).

CCD vs CMOS isnt much important as SLR/n had CMOS, DMR for Leica had CCD and both have colorwise output to die for..

JGR said...

My Oly E-1 has the best sounding and quietest shutter I have ever heard in a DSLR. Great photo quality too, with that Kodak sensor!

Gato said...

I've noticed some of the same things, though I've blamed it on the in-camera processing. Seems like there is a preference for brighter, punchier color and crunchy sharpening out there.

Specifically I seem to see this across my 3 generations of 16MP Panasonics. So I'm not sure it's all down to pixel count. It could be the manufacturers trying to appeal to the taste of younger/newer photographers who seem to go for a style that to me is garish and oversharpened.

Anonymous said...

A lot of ink has been spilled about the apparently inadequacies of the super high resolution sensors in the latest generation of cameras. The D7000 is much loved by its owners and delivers a file robustness that the D7100 doesn't quite match...deep shadows are artifact free, there's no color shifting of pulled shadows, etc. Further, modern cameras, high res or no, have had their imaging chain tweaked for high ISO sensitivity - at the expense of color discrimination. Blame the CFA for that. There should really be no difference between a high resolution sensor whose well capacity scales with the pixel size and a lower resolution pixel of the same technology generation from the standpoint of tonal depth; when downsized and printed or viewed, the same frame-level performance would result. As long as noise levels held up.
That being said, a lot of photogs have commented that 16MP is about as far as one wants to go for most things. The D810 is the resolution king, and can deliver spookily immersive files with the appropriate lenses, but many like the lower res FF sensors better.
I would argue that I like working with my 12MP D90 a lot better than my 24MP D7100, even though the latter camera is clearly better in many aspects and can deliver fabulous images under the right conditions. The problem is that I'm not able to take advantage of that higher resolution in most situations...it requires base ISO and a tripod. Part of the fun of the u4/3 stuff appears to be a constructive combination of just right sensor characteristics and super good primes, beyond what's normally mounted on DSLRs by most users these days.
Personally, I like the file quality of my D7100 better when pixel peeping than my D90, because its noise signature is smoother and slightly better. But my D90 is more fun.
Many fellow Nikonians with a D7000 or D5100 have passed on the D7100 for many of the same reasons as you, Kirk...too much muchness. We've cleared the IQ bar, but our trailing foot is dangling a bit. Careful not to knock the bar down.

granitix said...

Pentax will soon be out of the sweet-16 Sony sensor that you are enjoying in the d7000 (d7k?), so sensor fabs are driving us to the edge that you eloquently describe here. It is time to think of when you were most happy with both process and results and stockpile that gear.

In 2008 tests my 'best' images came from the E500 Kodak ccd and not from the E510's LiveMOS.. but new stuff like stabilization kept me searching.

Anonymous said...

Rewind not just 10 years, but 30 years back. We will be lamenting the clinical precision, sharpness, and colors of Nikon glass, and wistfully missing the organic full-bodied luminescent glow of Leitz lenses.

Now fast forward 10 years from now. Wistfully we'll look back at the human level rendering of the Nikon D810, the approachable friendliness of high megapixel D7100s. the gorgeous three dimensional skin rendering of the EM-1.

neopavlik said...

You're not helping my D600 vs. D800 decision ! :)

Recently using neutral density filters on my 24mp d3200 to bring the aperture down 2 stops has shown me some tremendous sharpness increases. The pixel pitch of my D3200 and your D7100 is tighter than the D800 !

I remember reading a lot of posts similar to this about the Canon 5d Mark I to Canon 5d Mark II increase of megapixels. I think that might be why the Canon 5d got the "5D Classic" name. Large fat pixels that were very satisfying, people wanted more pixels like those pixels...not a bunch of small pixels.

5-7 years ago I jot down my "dream" 35mm DSLR and 16-18 megapixels was what I listed. For a bit more completeness : 24 x 36 sensor, 1/500 sync speed, 1/8000 shutter, at least Fuji S5 dynamic range, large viewfinder size of F5 but body the size of F100, IN BODY crop / compose modes, a focus point that could move anywhere ( supposedly an early Konica Minolta digital body had this ), were the big ones.

Anonymous said...

I can relate. I found myself in a quandry two years back when the D7100 came out. I had been using the D7000 for a couple years and assumed all would be better. Bought the D7100 but felt the photos weren't as good. They had a lack of crispness...almost like the clarity slider had been slid to the left. Enlarging to 100% it seemed the edges of lines and such were highly compressed squigglies rather than edges, and it did seem like there was more noise than the 7000 at all ISOs. I was in a jam. i had already sold off the 7000. Pisser! I liked all the feature changes in the 7100, but didn't like the look of the images. Wish Nikon would have offered the 7000 sensor in the D7100 body and feature set! Craig

Anonymous said...

Perhaps the differences in perception between the D7000 and the D7100 images are due to the fact that 7000 used a Sony sensor with anti-aliasing filter, the 7100 uses a Toshiba sensor without an AA filter.

Richard Swearinger said...

This is kind of ironic because I LOVED the skin tones out of my D200 and hated them from my D7000. Partly the problem is that everyone has different color discrimination, but I wonder if local lighting conditions also play a part.
You're down there in Mexico Norte where the sun is like an angry sky god punishing you with an unceasing deluge of burning photons...I'm up here in Iowa where sunshine comes to us like a flickering candle seen through the window of a distant house.
And it might also be that the sensor engineering community has changed its philosophy. We all loved Kodak's digital interpretation of reality because they based it on their film cameras, it was what we grew up on. Perhaps the generation of engineers designing today's cameras have different ideas about what reality should look like.

Kirk Tuck said...

I'll agree with everything you said. Thanks.