New Full-Frame vs. Old Full-Frame: Nikon D750 (2014) vs. Sony A900 (2008)

I don't own a normal lens for my old Sony A900—I used to use it as a portrait camera with an 85mm. Recently I had the idea that a new lens might breathe new life into the old beast. And I had never tried the Sigma 35mm ƒ/1.4 Art that's been getting so much positive attention since it came out in 2012. So I thought I'd put the two together and see what happened.

At the same time, I wanted to try the Nikon D750 and the new Tamron SP35mm image-stabilized lens.

The lightbulb lit up, and I decided to ask LensRentals for two identical Sigma Arts—one for the Sony and one for the Nikon. (Shown above, on the cameras.) That way, not only could I see how the Sigma did when paired with the Sony, but I could see for myself how the Sony's aging sensor stacked up against a newer sensor with the same pixel count in the Nikon (both are 24-MP full-frame DSLRs). I also reasoned that it's the kind of "test" (I call them trials or tryouts—I don't work hard enough for them to be called tests) that isn't often done on the Internet. Most Internet reviews are of equipment when it's new. So if read a review of the A900 from 2008 and a review of the D750 from this year, both will talk about the sensors and the image quality in terms of the standards that prevailed when the reviews were written. It's difficult to get a clear picture of how older things stack up against newer ones.

The weather hasn't been good until today (I'm going to get out to shoot very soon) but here's a rainy-day trial intended to stress exposure range a.k.a. dynamic range. Shot at ISO 800 with both cameras.

The Sony needed quite a bit of noise reduction (NR), here applied to make the large file look good. But I was surprised at how pretty the result was.

The Nikon needed less noise reduction in the shadows and retained slightly better detail around the light; it was also noticeably easier to work with and had more flexibility in responding to the controls in ACR. I'd call this a win for the newer sensor. But not overwhelmingly; the Nikon doesn't kill the Sony as I expected it might.

Here's another attempt to find an extreme of highlights and shadows despite the overcast weather—a darkened interior with a window. At ISO 3200, I let the Nikon pick the exposure and manually set the Sony to match—ƒ/5.6 at 1/10 sec. Out of the camera the scene looked like this:

Then for each picture I HDR'd to the max: –100 highlights, +100 shadows in ACR.

To my surprise, the Sony result wasn't half bad. Detail out the window was very good in particular, although that could be due to the vagaries of exposure, which is never exactly identical whether at the mechanical or the electronic level.

The Nikon did better in the shadows, although I feel the Sony result is more true to the "feel" of the real situation as I experienced it, for what little that's worth. (That kind of judgement is called a "perceptual" appraisal. Once you determined that's what you wanted, you could probably match with the Nikon.) It looks to me like the Nikon just exposed a little more for whatever reason. The Sony might also be biased toward greater highlight detail, something I've long suspected just from using it (and probably the reason I liked it).

But then here's a pretty clear difference:

(You'll be seeing it at 100% after you click on it.) That's a detail of the noise from the Nikon file, captured as a screenshot from ACR. You can see "grain," but bear in mind this is ISO 3200 boosted +100 in the shadows, with no added noise reduction (NR). You could clean and sharpen this up a lot further.

This is the same thing from the Sony file. And bear in mind that Sony's engineers baked some NR even into the RAW files (we can turn to Imaging-Resource again for a good discussion of sensor NR—although I sure wish they'd date their articles).

This probably makes you remember how obsessed everyone was with shadow noise during the middle years of the decade of the 2000s, and why. The Nikon example makes you realize why it's no longer such an issue.

Result of this trial: the Sony did better than I thought it would, but the Nikon's sensor is clearly better. As I recall, I used to limit the A900 to ISO 1600 in real shooting.

But now here's a different situation, one where the old Sony actually came out on top:

Here's the shot. I was wandering around with both cameras focusing back and forth from near to far objects to get a handle on focusing speed for the kinds of things I shoot. (Verdict on that: the Nikon's faster but it's mainly noticeable when the lens has a longer way to rack. The A900 is Alydar, always second but running a good race. I was never bothered by the Sony's focusing speed.)

Here's the detail from the Nikon test shot. You'll notice that I managed to handhold 1/10th of a second with the unstabilized Nikon in the window shot above, but here, at 1/25th of a second, I've gotten just a slight amount of camera shake. (This is a degree of camera-movement degradation most shooters would have been unaware of in the film years, by the way, before we could easily examine detail at such high magnifications.)

Look at the same detail from the Sony:

(Again, if you click on these you'll see them at 100%.) Much sharper. Note the bluish blotching in the black block letters, though.

I added a little speed for the Nikon shot so I could more easily hand-hold it—1/25th vs. 1/10th. There's certainly the possibility that I merely happened to hold the camera more steady for the Sony shot; in the window shot above, I handheld the Nikon at 1/10th and I don't see any motion blur. But given that the Sony shot here is at 1/10th and I can't handhold 1/10th as a rule, I'm going to credit the Sony's IBIS for the sharper result here. That has not been demonstrated by this single test, but that's my story and I'm stickin' to it.

One for the old guy.

Tentative conclusion for now
It appears from my few comparisons (these and several more) that the Sony sensor in the D750, with Nikon's processor and implementations, is superior to the then-groundbreaking Sony 24-MP sensor from 2008, mainly in the area of shadow noise and high ISOs—and it appears to have more in reserve for manipulation in post. But just as nine-year-old Lulu-dog occasionally gets to the ball first by being clever, despite two-year-old Butters' youthful strength, speed, and agility, the Sony still has a trick or two up its sleeve. The A900 is still in the game. (Lulu has a ton of heart too. She's always game.) So far, the Sony is losing but putting up a decently impressive fight.

I should add the possibly apocryphal admonition of the old Leica lens designer: "all tests are just shortcuts. The way to really get to know a camera is to use it for a year."

I got outside today and shot a bunch of stuff with one camera after the other. If there's anything to talk about from those comparisons I'll get to it this week. I have another reader portfolio review tomorrow (Monday), which I'm looking forward to.

Mike
(Thanks to Ctein)

Equipment for our posts is provided by
LensRentals or B&H Photo

Original contents copyright 2015 by Michael C. Johnston and/or the bylined author. All Rights Reserved. Links in this post may be to our affiliates; sales through affiliate links may benefit this site.

TOP's links!

(To see all the comments, click on the "Comments" link below.)
Featured Comments from:

Ulf Aagerup: "It’s great to see a sober comparison of cameras from different generations. The idea that photos from old digital cameras is comparable to those from new ones is interesting, and in line with my experience. I have the EOS 5D and the 5D Mark III. I prefer the look of the original’s files. I also prefer the look of the Fuji X100 compared to that of the later versions, and the X-Pro 1 files compared to the latest Fuji Xs. I don’t know why this is the case. If it were not for the newer cameras’ superior handling, I would be happy with my old cameras."

Mike replies: There was something sweet-spotty about that original 5D. I know what you mean about the files—they looked great, and no one could ever quite figure out why. Black-and-white conversions, too. A landmark camera for Canon, in several ways.

Comments

[Ash]

Hey Mike,
I was hoping you would throw some Fuji into a test like this. Can the X-T1 and 23mm 1.4 play on the same level as these full frame cameras?

[The Nikon has to go back tomorrow, so the Sony with the 35mm Sigma vs. the X-T1 with the 23mm is next on the docket. --Mike]

[Tuomas]

Thanks, Mike! Very interesting, and a great idea. I'd like to hear more about this pair of cameras.

And may I say, what little bits you've shown of your house, inside and out, it looks and feels great!

[Andrew C Eckert]

The one thing I sometimes feel is missing from my Nikon SLR + prime lenses combination is some form of stabilization other than my tripod. That Tamron 35mm f/1.8 is looking mighty tempting for that reason, though it seems a sub-par solution compared to the A900's IBIS.
The blue blocking in the black letters of the A900 crop looks to me to be a reflection. It's less apparent, but still visible in the D750 crop slightly to the left, which lines up with the blue reflection in the silver label being further to the left in the D750 crop.

[Dennis ng]

I was attacked to use sony camera without using zeiss lens - call eating the veggie due to the Cantonese sound.

The tuning is for zeiss lens and not far ...

Not sure it is true.

[Manuel]

So now we have further proof that most of Sony's development effort went into improving noise performance of their sensors. Not exactly breaking news.
We also become aware that photography was pushed too far into artificiality, as the files taken with the Sony camera indeed look more true to life, despite the greater amount of noise. (Though noise reduction seems a bit overcompensated in the Nikon files.)
You know what? If this test were absolutely conclusive, allowing me to make a choice confidently, I'd take the Sony over the Nikon. Despite the poorer noise performance and the (very marginal) dynamic range limitation, the pictures simply look more authentic. That's no surprise, because at this level the evolution is leading towards more detail ("resolution") and more of what people call "dynamic range", making pictures look more clinical and unauthentic. A bit of noise seems to be an acceptable tradeoff if the pictures have a little more soul. (And maybe that noise can work favourably in black and white by emulating film grain, who knows...)

[Louis McCullagh]

When hand holding at low shutter speeds use the drive option and shoot a sequence of images. Invariably you will find a near perfect image. Digital is cheap and easy to delete.

[Lou]

As a film shooter, I am never too worried about high ISO performance, since 1600 is way above the usual film ratings. I find that the Sony A900 to be perfectly usable today, still way way better than anything but pro cameras. The Zeiss glass is lovely, as well as the 1.4/35mm from Sony which has a little something special to it. What I really like is the dynamic range, which is very good, close to black and white film. I have lugged mine on various trips, and wear starts to show, but it is still perfectly working. At the end of the day, when counting keeper files of the past year, many of them are with the Sony.

[Rob Spring]

Did you have the vibration control activated on the Tamron?

[Hi Rob, I didn't use the Tamron for this post. Both cameras had the Sigma 35mm f/1.4 Art lenses on them. --Mike]

[Gato]

Thanks for the comparison. Most interesting.

Like a couple of other posters I have been a bit unhappy with the photos from my latest generation of cameras compared to the ones they replace. (Panasonic, in my case.) To my eye the the modern files are a bit "crunchy" or "edgy" - they look a bit oversharpened to me. Also the noise reduction seems overdone, even after I turned it down. The old files have more happening in the darker areas which to me gives more life and appears to give a tiny bit more detail. On close examination it seems to be about half detail, half noise, but it still looks better in a print than smooth flat gray.

From what my friends are doing and what I see online this seems to be a general trend. I wonder if the drive for sharpness and low noise may have gone too far.

[John McMillin]

Thanks for a backwards glance at a camera some would write off as "obsolete," simply because it isn't optimal any more. My Sony a850, recently sold, was a noisy beast when viewing he files at 100%. It wasn't smooth, grain-like luminance noise, but blotchy color noise. But it never was apparent in prints, erased by either downsizing or noise reduction. Image quality wasn't the reason I left it behind. It needed repair twice in two years, and no successor OVF camera was coming to support my choice Maxxum lenses.

Sensor progress is real, though. My new DSLR proves that, with little noise showing at ISO 3200. Even stepping down from 24 to 16 megapixies was no loss, since the new camera loses the AA filter that's been blurring our images since digital's dawn.

My camera change had no costs in missing features, except for missing the widescreen view of the a850's lovely FF OVF. I didn't have to give up stabilization with all lenses, either. Nikon's in-lens VR approach may suit the supertelephoto specialists and the single-zoom masses, but it doesn't satisfy my needs. Like the moment on an Indian reservation (no tripods, please) where with a long exposure, the Sony photographed softly blurred tourists walking past timeless adobe buildings, all with an ultrawide lens. The new camera's IBIS even works with my few stray MF Takumars I sometimes used with the Sony. And soon, next year, I might even get that big viewfinder back.

Can you guess my new camera? The Pentax K5IIs, also "obsolete" and superseded by two new models. That's life in our real world- few of us, usually only novices, are choosing between several brand-new cameras. Instead, we're weighing that new one against the ones we already own. That's why comparisons like this are so useful, but rare, because professional reviewers only follow the newest gear.

[David Brown]

A few years ago (maybe 2010 or 11?) I commented on one of your blog posts about dslr's that all I really wanted was a Minolta SRt with a digital sensor. You replied that I should buy a Sony a850 or a900. I bought an a850 body and have used nothing but Minolta a-mount lenses on it (mostly that I already owned).

It was good advice, and I have no reason to upgrade until this camera dies. Although when I do, it's likely to be a much smaller mirrorless, somewhere between the Sony and my little Lumix LX7!

[Ed]

The question is not do they compare, they shouldn't and couldn't, sensor development goes in leaps not in strides. The question is do you notice it in an exhibition, with a skilled editor and great printer (both man and machine). What you see in a comparison will get lost, both camera's will turn out outstanding pictures for their respective owners.

Greets, Ed.

[Stephen Scharf]

I can't say I am at all surprised at your results, Mike. What this tells me is that we reached a point of sufficiency for resolution and overall performance a while back. In my Six Sigma parlance, I would say that you observed a statistical difference between the two cameras, but probably not practical one.

Regarding the Canon 5D, fully agreed. That camera did something magical to files, and my Fuji X-Pro1 was the first camera I shot with that recreated that same magic.

[Nigel]

Sensors might not have improved >all< that much (though for those of us shooting m4/3, I think they probably have over that period), but the responsiveness and resolution of electronic viewfinders has seen them go from near unusable to indispensible (for some of us)...
...which is why the otherwise splendid Nikon, its siblings, and its Canon competitors are of no interest to me.

A personal view, I acknowledge. And an interesting comparison nonetheless.

[Bill Mitchell]

These days, they're ALL good.
Just find one that's comfortable to use.

[Alan Carmody]

The gifts of latter day digital photography are usable high iso ranges and image stabilization, in-body or otherwise.

These let us take good pictures in marginal lighting that would have been technically impossible ten years ago.

The curse of latter day digital photography is high resolution, which makes us zoom in at 100% and makes us technically critical and dissatisfied even if there in no discernible aesthetic impact on a photograph. Also, high resolution encourages a 'crop later' mentality, which severely corrodes visual discipline in the photographer.

For this reason, it seems to me that cameras with 12 or 16 megapixel sensors are a good compromise, whether the expensive Sony A7SII, the now discontinued Nikon D700 or a variety of Dx and micro-four thirds cameras.

Me, I miss my 6 megapixel Nikon D100, with which I learned discipline in framing, and which encouraged me to seek out good light, which is usually synonymous with brighter light. Available light is, well, available, but almost always unsatisfactory from a light quality standpoint.

[tex andrews]

Still like the ergonomics (and look, even though that is not so meaningful)of the Sony (I had the A850---partly from your recommendation, Mike...). The A850 was one of the nicest handling cameras I ever had. A shame I had to trade it in.

Watch for the new Pentax FF. I've seen it in person (seen, not handled...). Looks to be quite impressive.

[Geoff Wittig]

I've recently gone back to some of my older digital images captured more than 12 years back with the original Canon Eos-1Ds. Yes, anything above ISO 200 shows pretty dreadful noise. But it's amazing how well the files actually stand up with some careful processing. And it's remarkable how much better Photoshop CC is for 'developing' these files than the release I was using back then.

No question I can print a lot larger with new images shot at 50 mp, but it's the whole imaging chain that matters in the end, from the glass in front of the sensor to the solid tripod under it. And most of all the eye behind the viewfinder! I find I have to be shooting for at least a week before I get back to any semblance of skill.

[Darlene Almeda]

What Ed said ^^^ ...

What you see in a comparison will get lost, both camera's will turn out outstanding pictures for their respective owners.

[Marty]

Really a comment about the featured comment and TOP reply; I finally parted ways with my 5D yesterday after eight years of pleasant and trouble-free friendship and some decent earnings too. My take on the 5D; No digital file since those produced by that camera have better replicated the look of film. Strange, I know, but that's how I felt about it. It was a digital/film platypus.

[Kenneth Tanaka]

It's always interesting to overlay the DXOMark ratings against personal observations.

Sensors most certainly have improved in recent years. In this case it seems that the low-light performance of the Nikon represents the biggest measurable improvement. No surprise there, as that's where most sensors have seen their greatest generational improvements.

Overall I think we've reached an asymptote on the curve of sensor performance. We're unlikely to see the performance / resolution leaps we've experienced over the past decade.

[Dillan K]

And here I soldier on with my trusty 5D. Its 10 year old technology is good enough enough for me--except for the auto focus, but that's not what we are discussing here. The files I get from the camera are just fine. Properly exposed photos are quite manageable at any ISO. The key, though, is 'properly exposed.'

[Gildas]

A few years back I nearly nearly bought an A900 new... Probably due to your review.

Why? Not on megapixels and other purely technical merits, but on the image results. I know this is mostly down to the photographer, but A900 has a way with sunlit skin tones and translucence that is really appealing. Much in the same way the original 5D has a great tonal range and the FUJI's make skin lovely.

Still might get an A900 second hand with 85mm...

[Nigel]

Overall I think we've reached an asymptote on the curve of sensor performance. We're unlikely to see the performance / resolution leap..

I'm not sure that's true; resolution might well increase until the area of a four pixel RGB array is below the diffraction limit even for the fastest lenses,.

And as far as 'performance' is concerned, it depends what you mean.
Note the under sampling/oversampling conundrum & potential solution(s) outlined here:
http://spectrum.ieee.org/biomedical/devices/how-neuromorphic-image-sensors-steal-tricks-from-the-human-eye

[Kevin Purcell]

The A900 (and A850) used the Sony IMX028 CMOS sensor that was also used in the Nikon D3x (and Nikon seemed to get more out of the same sensor).

http://www.dxomark.com/Cameras/Compare/Side-by-side/Nikon-D3X-versus-Sony-Alpha-900___485_371

http://sensorgen.info/SonyAlpha-850.html
http://sensorgen.info/NikonD3X.html

Both came out in 2008 after Sony's CMOS EXMOR "great leap forward" with on-chip column ADCs after they moved from the previous "Super HAD" CCD sensors with off-chip ADCs.

That's the big dividing line between "old cameras" and "new cameras".

Comparable APS-C Sony CMOS sensor IMX021 (with rather similar sized pixels) being used in the Nikon D90, D300 and Fuji X100 have similar performance (i.e. better than most people expect).

The takeaway is Sony CMOS sensors started to get "really good" in 2008.

Although the sensors are not the current state of the art they are very usable even though the cameras seem "old" to many. This is especially true if you shoot in "film-like" lighting conditions for film-like exposures (think ISO 200, 400, 800 or even 1600 but no higher).

The main difference to more recent sensors is improved read noise that will show up at higher ISOs (3200 and above) and the (smaller) difference in QE that will show up in midtone noise. The latter is rather less noticeable.

[Tom Bell]

As with Ulf ... I love the original X100 Fuji files ...very filmic .... Subtle ... And of course a Bayer Sensor and "only" 12 mp.

I love the results and personally like the fuji lenses with the Bayer sensor. For that reason the XA1 is a bargain and I hope they will bring out a XTA 2 with a ? 20-24 mp bayer sensor.

BUT I will keep the X100 indefinitely

[AHC McDonald]

I don't understand the meaning of these tests, and these ludicrous comparisons probably sit better on other sites, where bad photographers compare noise levels on their terrible photos to no effect. How often would the blue blotching on the jar make the difference between a good and bad photo? The number would approach - probably reach zero. what does it mean exactly that this lens with this camera has a slightly different response? The times when any of this crap made the slightest difference to any kind of photograph is never. Zero. It has never made any difference and it never will.

[Wim Ter Voert]

Very useful and above all real life observations instead of the usual graphs. I use the Sony A900 all the time and even added an a850 now that these animals can be bought from US$ 550-700 here in Europe as almost everyone seems to climb over the fence where the E-mount land seems so much greener. I will probably stick to the old ones for some time to come though and this review certainly supports that feeling.