Nikon D3X DXOMark Numbers Blow the Sony A900 Away!

[Javelin]

Looks about right for the diference in price though. actually the A900 still looks like a bargain considering the Nikons 3x's the price. Is it weird though that in the plots the Nikon and Sony plots are pretty linear where the canon is curved? is there some kind of boosting going on for certain ranges and thats the results you see in the graphs ?

[David Kilpatrick]

Interesting figures, but doesn't relate to real life very accurately. They must be working from JPEGs, since DxO Optics Pro can not yet read D3X files, Alpha 900 files or Canon 5D MkII files! Really, coming from a software maker which has not even managed to get the basic function of recognising these three cameras in their raw conversion program, I don't see how the assessment can be valid. All three cameras apply substantial, different, raw to JPEG conversion processing. They don't even have reliably matchable presets; Nikon standard, Sony standard and Canon standard aRGB or sRGB conversions are entirely different, Canon has two alternatives called 'Faithful' and 'Neutral', Nikon uses film-type simulation by default.

All that the DxO labs tests prove is that it is POSSIBLE to test certain results from these three cameras and end up with their figures. What their tests do not show, at all, are how well the raw files compare by the same standards - or whether the raw files are better than the in-camera JPEGs.

Every test I've done, whether casual or planned, shows that the A900 doesn't clip at either end of the scale in conditions where the D3X will clip. So how they get the D3X having better dynamic range, I have no idea. Plenty of users and reviewers agree, the A900 is better at providing shadow and highlight detail in one go (even if the images look lower in contrast as a result).

I'll put more weight on DxO Mark when DxO bothers to get its own raw converter to work with the cameras reviewed.

David

[David Kilpatrick]

So what? You don't need to run any kind of lab test to know that the D3X is currently the best DSLR on the market. In my own report in the BJP published yesterday I say that:

"Conclusion
I’ve tried to compare the D3X with the cameras it is being pitched against by those who can’t afford to buy it – or justify it. The comparison is not easily made except in terms of detail. It’s easy to find a function, or a situation, where the 5D MkII or the Alpha 900 will be faster or easier to use than the D3X but the same applies in reverse.
The D3X is greater than the sum of all its parts, qualities and features. It does not hark back to the basics of film SLRs the way the Alpha 900 does, or look to a movie-clip shooting future with the 5D MkII. It offers everything a professional still photographer needs from a DSLR.
That makes it probably the best DSLR yet made. Or just miss out the ‘probably’. All you need is the work, and you can afford it."

I think the DxO labs tests are ridiculous in their attempt to quantify performance down to a decimal point of a bit depth value, and yes, they are the same company as DxO Optics Pro, and no, they have not released any kind of raw processing compatibility for any of these cameras yet. If they don't consider what they have adequate for public release, how can it possibly be adequate for lab tests intended to produce a once-and-for-all definitive set of FIGURES expressing the performance?

Since they claim to make a print to assess, they must be applying a transfer function to the raw conversion and creating an RGB image from the raw - presumably they simply use DCraw. And if they really do reduce everything to 8 megapixels, the entire exercise seems pointless. But, of course, they do in order to provide fuel for people like you to jump up and down in glee shouting "the D3X stomps all over the A900!". Juvenile. A) It doesn't, it just marginally betters it in a few specific ways B) It should at the price.

David

[douglasf13]

Interestingly, the noise between the A900 and D3x isn't much different. It's around 1db at all ISOs, which equates to only around 1/3 stop EV.

What's really interesting to me is that the difference between the D90 and D300 on Dxo Mark is somewhat similar to the difference between the D3x and A900. Nikon seems to have found some kind of new special sauce in hardware or software that is really working with EXMOR.

[douglasf13]

Sonolta, I said nothing hostile towards you. I was simply stating that the SNR measurement between the a900 and D3x wasn't far off. I agree that the DR is much different, but it is worth mentioning that the d90 has outrageous dr as well. I read about this stuff as mych as you do, but I rarely post here, because I like being involved in pleasant discussions rather than childish rants. You're truly one of the worst posters in all photography forums. Normally your comments are blocked to me, but I forgot to login first today At least your responce to this will be blocked

[David Kilpatrick]

Don, I am not claiming anything for the Sony. You are saying I am. That's entirely different. You don't read what I write professionally and you have no idea how much I have criticised the Sony (and previous Sony models). The various simple tests I have posted here (on the website) and on dPhotoexpert draw no conclusions and don't attempt to 'rate' cameras. They let the viewer judge. It's interesting to see from dPreview that each brand's forum appears to think these articles prove theirs is the best, apart from a few sensible people who comment that all the 'competing' cameras are clearly capable of excellent results.

What I find harder to understand is why you think this forum is the right place to highlight tests which, in your view, show the Alpha models in a bad light.

As you know, many different comments can be made about the DxO test results. For example - the Sony Alpha 900 is almost two-thirds of a stop more sensitive at ISO 100 than the Nikon D3X or Canon 5D MkII. Sony ISO = 119, Nikon = 73, Canon = 78. Of course, that's just one figure and the rest of the entire scale shows the A900 to be less sensitive than the Nikon - and more sensitive than the Canon. And I know this is all nonsense, and I have no idea how they got these figures which show that ISO 100 on the A900 is only 1/3rd stop slower than ISO 200, because whether I shoot raw or JPEG I can assure you a normal one-stop difference gives well matched exposures. And the Canon, at exactly the same exposure, is consistently brighter not darker than the Sony (about 1/3rd of a stop). I've checked this using in-camera JPEGs, the maker's own converters, various lenses, Iridient Raw Developer, ACR - the lot.

The DxO test is getting a lot of comment on dPreview on all the forums involved, and a fair number of 'but we know this is not so' comments included. This is the problem I have with the DxO results. They appear to prove things which in practice either turn out to be not the case, or apparently the reverse. Therefore I would be cautious making too many claims for the absolute authority of these, or any other, lab measured tests. For a start all such tests claim their own methodology is the best, and then they come out with different results. At the best you get some broad agreements, and guess what - most users could spot the same general conclusions without needing a lab. At the worst you get opposed findings.

Your links to these results are welcome and useful, but the way you present them isn't. You could equally well have headed this thread:

12-bit Sony A900 matches 14-bit Canon!

After all, it is more impressive that the 12-bit Sony with its smaller pixel pitch should manage 12.3 dynamic range against Canon's 11.9 with 14-bit. Nikon's 14-bit should beat Sony's and at 13.7 does pretty well. But Canon's 14-bit should easily beat Sony's 12-bit and does not. Why not make a song and dance about that statistic, instead of 'Nikon...blows Sony...away!'?

David

[David Kilpatrick]

Actually Don hardly any Sony 'fanboys' here are popping up with any results. The reason a large number of manufacturers and publications use DxO data is because it is offered entirely free on a contra link basis - end of story. They do not have to include ALL the data, they can cherry pick anything they like from the test results, and the cost to them is a website link or a printed acknowledgement. The gain to DxO is publicity for the real game, selling DxO Optics Pro. They will also sell you a complete replicated lab for testing, and the software, and a hefty annual maintenance contract charge; they tried this on me as a publisher about four-five years ago, but no thanks! I think they ended up with no takers from the UK.

This website has nothing do to do with my professional writing. This is a voluntary site run as a personal interest, for the benefit of Sony Alpha users. It is not even restricted to Photoworld subscribers (and has no restricted areas), and it does not solicit donations like Mr Rockwell. Fortunately my professional writing and editing, which you won't see anywhere on my own websites, is profitable and has been for 30 years.

David

[David Kilpatrick]

Don, I get direct press information from DxO about their products. Never once in the history of the company has a single release been issued stating that any camera company has purchased their hardware. Nor have I ever seen any photograph of an installation. Nor have I ever been given any information, from any camera maker, that they use such an installation. I know there ARE installations out there, DIWAlabs being one, but I also know that these installations are financed by contracts from... the major camera companies, cellphone manufacturers etc.

Sure, they are using the software and the tests, but just like any other industry this does not mean they all have in-house facilities. It just means they send their products to DxO or DIWA (etc) and get the tests done. Just the same way the motor industry does not own all its own test facilities but uses national labs built to serve their needs.

One thing you can be sure of, if the wording used does not state something directly and clearly, it doesn't mean it.

You will also already know that DIWAlabs testing of the cameras using DxO's own gear produced entirely different results to DxO's testing. That says a lot. If one operator or one lab set-up fails to match another, it does not mean there is a fault with the gear, but it does mean any results are subject to operator error or local conditions.

David

[David Kilpatrick]

Indeed:

"The DxO Analyzer has been installed in the Popular Photography test lab and joins the numerous instruments, databases, and computer programs that are used to generate objective test measurement data for the latest generation of digital cameras and lenses. Beginning with a feature on Anti-Shake technologies in the August 2004 issue, this DxO Analyzer data will assist the magazine’s editors in evaluating camera
and lens performance and image quality."

We received a similar proposal around that time, it doesn't mean you produce your own data or install a lab. Read the wording with great care. It means you buy the software and DO Labs send you the data, which you can then display and manipulate to produce charts and tables and stuff for publishing. Basically they bought a PC software suite and a contract to receive the measurements made by DO Labs.

This is nothing new - we had a similar film-era deal through TIPA (the Technical Image Press Association) to receive all the material from the Carl Zeiss Hasselblad testing facility, way back. You could assemble a lens test without ever handling any of the lenses, all you had to do was pick the set of lenses, get the data, produce your charts and graphs, and write it up. The cost at that time was minimal (about $500 a year for TIPA membership per magazine). I decided against it, because the amateur mags were using it - Practical Photogaphy included.

The camera makers have their own research facilities mainly used to evaluate suppliers' proposed components, as a brief visit to photokina would show you. What you think are 'Panasonic' or 'Sony' or 'GE' or 'Vivitar' etc are mainly third party products. The sensor-lens assembly business is the biggest source. At consumer camera level, the sensor and the lens are a module - sealed - which the maker buys and puts into the camera, adding firmware/processor and other stuf (like the name on the lens rim!) later. I remember seeing the first 10 megapixel consumer camera zoom lens plus sensor assemblies, before there were any cameras using them, and wondering where they would turn up first. Answer- really cheap rubbish! Presumably the big names tried them, and rejected the first wave.

David

[David Kilpatrick]

Don, DxO stuff costs nothing to use - read the website. All they ask is a link. That is, for their own reports. This is a relatively new move, in the past they wanted to sell data.

The A900 is not 'getting blown away' by the D3X. If you bother to study the DxO site closely, you will find that the A900 is placed higher up in the rankings for many of their tests than you would imagine. The D3X is certainly better, we knew that before DxO published anything - you would need to be blind not to be able to see the differences in 'pixel size' for noise, this stuff was apparent from the day the D3X first hit reviewers.

To be exact, the Alpha 900 is no 6 in the global ranking of all DSLRs they have ever tested. The order is D3X, D3, D700, 1Ds MkIII, 5D MkII - then the A900, though only .1 of a step below the 5D mkII, scoring 78.9 versus 79, so in effect you could call them equal fifth within any margin of DxO's error. Every single camera ranked higher is also more expensive, two by a factor of double the price or more - and the D3/D700 gain an understandable ranking given that megapixel count is not part of the DxO assessment.

In their 'Colour Depth' ranking, the Alpha 900 is no 3 behind the D3X and 1Ds MkIII - cameras at over twice the price.

In their Dynamic Range ranking, it's number 5 - but there is a catch, because two places are taken by the Fuji S3 Pro and S5 Pro which have dynamic range as a special feature. Actually, only the Nikon D3X and D90 beat the A900 for dynamic range in a conventional sensor (and the two Fujis are, really, only 6 megapixel cameras despite the 12 megapixel theoretical count - they are no sharper than the KM 5D and I used both the Fujis).

In low light high ISO, the final parameter they use, the 900 ranks seventh. That is surprising. I thought it would be much lower down, but it beats the EOS 5D, the 1D MkIII (!), 1D MkII, D90, D300, every single APS-C Canon ever made, and even beats the KM 5D and 7D. I have to say this surprises me as I would not have thought the high ISO low light performance of the A900 beat every single APS-C Canon, or the 6 megapixel KM models, in absolute terms. (This is one of the reasons I don't take DxO lab tests as gospel, I know for sure that some of the cameras they rate as worse for low light high ISO are actually better in practice).

Now those 'facts' from DxO tell a very different story from your knocking copy. They tell the story of a full-frame DSLR which, at its price, punches above its weight.

David

Your sole interest here seems to be to put a spin on stuff to diss the A900 at all costs. God knows why! It's never done you any harm.

[Javelin]

Don why wouldn't you post this on a Nikon site instead of here? They got their own fans you can abuse? or fight the fanboy fight on a general photography forum. seems counterproductive to go to brand enthusiast sites and accuse everyone of being fans.. don't ya think?

Edit: Before you ask no I didn't read the middle part of this thread I just read you post usually because you have interesting things to say when your saying them and not ranting ... at least they are interesting when your right that is.

[Javelin]

to be honest yes I haven't been following the comparison threads. I'm not all that interested in them really. I'm not shopping for a camera and frankly I can see some of the defects you guys talk about in the pictures (don't know where to look unless it's pointed out) just whenever I run across ssome posters in the main list of threads I try and read them and too often i've been hitting you in full bloom stucatto rant mode.. just this last one struck me as silly.

[Dusty]

Every single camera ranked higher is also more expensive, two by a factor of double the price or more - and the D3/D700 gain an understandable ranking given that megapixel count is not part of the DxO assessment.

Now those 'facts' from DxO tell a very different story from your knocking copy. They tell the story of a full-frame DSLR which, at its price, punches above its weight.

David

Which is why I originally bought Minolta nearly 30 years ago, and why I bought Sony this time around. It's a price/performance trade-off. I am good friends with several millionaires. They can afford (but don't always buy) anything they want. There are time I wish I could do that, but I can't. When I make money off of my pictures, it's usually to pay a few bills piling up, or to offset the cost of the extravagance of photography. Yes, it's an extravagance, if it's not your business. The fact that so many in the world can afford it now speaks volumes for human progress.

With my A350, I get more pixels than most other DSLRs, good glass available from several makers, and anti-shake built into the body, keeping future lens costs down. If I had more cash than brains, (or enough brains to make LOTS of cash) I would but a Phase One, or maybe the Leica S system. But I'm on a fairly limited budget, and if I can get a decent 14MP camera for about what I paid (inflation adjusted) for my X700, I think I'm doing pretty good.

Is it noisy at 3200? Depends upon your perspective. 10 years ago, when I was shooting inside various Pharaoh's tombs, I was using a very grainy Konica 1600 film. That was fairly new film at the time, and the highest color film readily available. BTW, the guy that had an expensive (but probably not top of the line) digital had poor resolution and S-L-O-W storage, AND he couldn't shoot in such low light!

I'm very happy with my A350, because I remember not only grainy 400ASA film, but I remember looking at my Uncles Kodacromes from the '40s. That was good stuff, but at ASA 8, it wasn't for action!

If I could afford a A900 and had to "suffer" with it's poor performance for the rest of my life, oh well, I'd learn to live with it!

Dusty

[dynaxfl]

Don, after reading all of these posts and others, I feel like you are the fanboy of anti-sony-fanboys. If anyone who praises Sony or says something like Sony is better than Nikon or not far behind Nikon, I will see your following post bash them. In fact, you claim others that bashes this and that, but I would say you bashes everyone here if you feel they are the sony fanboy. I have a black friend called him a coconut, because he has dark brown skin but he feels like he's white inside... I don't know how to call you, lol... a guy who owns 5k in Sony but with heart of Nikon? I'm just kidding. and if my post offends you, i'm ready to be bashed by you:). But I still respect you and enjoy your post and arguments. I have never seen anyone like you who can "bash" others with almost no constraint yet calls other people in guilty of bashing others. For me, It's entertaining and learning at the same time.

[bfitzgerald]

Hmmm nearly 13 stops of DR..for the NIkon

(in reality the usable figure is way lower)

I would honestly want to see more than the DXo test.
Find and dandy etc, but it's telling me next to nothing regarding real world use.

Have to doubt them a bit, I said this before but it went un mentioned..the Km5d rates .1 higher for DR than the A900, can anyone honestly say these numbers are accurate?

I see the Nikon D90 12.21 at base ISO v 11.5 odd DR that is.
I have not used a D90..but DK has...maybe he can comment on DR with this one..

Plus the D90 didn't get amazing DR scored on the reviews I read.