« Hat Weather | Main | Random Excellence: Jean-Yves Lemoigne »

Friday, 16 January 2009

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00df351e888f8834010536d49319970c

Listed below are links to weblogs that reference Nikon's New SLR Leads the Pack for Sensor Quality:

Comments

re: the Nikon "The Price is So High that I am Going Back to Film - KR" D3X

Of course Nikon claims that their sensor is different from the A900's and supposedly they use 14 bits ADC etc., but we don't really know what goes on in the backrooms, do we? I can imagine the following scenario:

Sony: Hey Mr. Nikon, we can supply you the highest quality FX sensor so you can eat Canon's lunch.

Nikon: Sure, what's the catch, you little devil?

Sony: Simple, you position the D3/D700 at mid to upper range prices, and we will price our A900 at the low end of that, and you will price the D3X at nose bleed price. In the end, we will all eat Canon's lunch.

Nikon: Only if you cripple your A900 processing!

Sony: But of course, Bwahahahaahahahahaha

A commentary on the camera or on the test ?

Brilliant comment, Richard Man. And the worst is that it might possibly be not too far fetched from the reality.

Great. A camera that costs more than the total value of my three cars...

Very interesting. In the past couple of weeks two of the most prominent Nikon bloggers - Thom Hogan and Ken Rockwell - have said as much about the D3x's quality ("unparalleled," basically) while decrying its price (Rockwell has been encouraging a worldwide boycott of the D3x until Nikon drops the price).

Apparently Nikon knew what they had with the D3x (i.e., yet another home-run product introduction), and they knew what they were doing when they priced it at $8000: even in this economy, there are a lot of photographers who will pony up for a clear increase in image quality. Is the price too high? Depends whether one is comparing it to low-end medium-format digital backs or to high-end 35mm-sized cameras that are (now) measurably lesser in image quality.

Besides, Nikon probably wouldn't have sold a lot more DX3's at $6000; as I heard someone quote recently the line said to a desperate-to-leave-town Steve Martin in the movie "Planes, Trains, and Automobiles," "Any man who would pay $50 for a taxi surely would pay $75"--and he was right.

Things will get *really* interesting this summer if Nikon introduces the rumored "D700x," a more-affordable version of the D3x that uses the same sensor and technology for half the price.

I tell my students that a single number conveys no useful information. A number plus the associated error is informative and useful.

The problem with DXO mark number is that it is a relative value and we don't know what change in that number is meaningful (meaning that someone can see the difference). Without knowing the error (standard deviation) of each number they are as good as an overal rating by a good reviewer.

The plots in DXOmark do not show the error bars, therefore even there it is difficult to compare close cameras. Anyway looking at the actual plots in DXOmark and comparing the Sony 900, Canon 5Dm2 and Nikon D3x, we can see that "it seems" that the only area where the D3x may have a clear edge is in the dynamic range (below ISO400 regarding the 5Dm2, and at all ISO regarding the A900). All other plots are way much to similar to distinguish any of those models. Of course, this comparison can be extended to other cameras as well.

It is, therefore, not surprising that photographers are showing that all these cameras produce superb results.
I would be very much surprised if someone were able to atribute prints of the same scene, taken in similar circumstances and treated in similar ways, to any of the top cameras from those 3 manufacturers just by looking at them.

Buy any camera and start photographing...

I'm glad the D3X is the best 35mm on the market. It certainly costs a kings ransom and for some it will be worth it. I think for most of us we won't need that much camera even for pro work at that price. IQ wise it's makes you wonder what Sony has up the sleeve in FW updates and possibly newer models. They obviously have the skill to create top notch sensors in-house. The next 5 years should be interesting.

Regarding Richard's imagined scenario above, Sony's influence must be wide indeed! Thom Hogan said "this is the DSLR with the best image quality available at the moment" and "the best DSLR in terms of image quality so far." Joe McNally said "The detail of the D3X for me, obviates the need for a medium format approach to just about anything I would tackle" and "when a D3X NEF finally boiled to the surface of the screen, I cocked my head, my eyes got wide and my jaw slack..." and Dave Black said "By the way, the NEW Nikon D3X camera is AMAZING!" Or maybe, as Peter suggest, it's merely a commentary on their testing methods.

Granted, many have questioned it's value. So far, though, DxO hasn't devised a test for that.

(BTW, does anyone actually listen to KR? Just curious.)

sure its the best camera out there, but i hope people do realize that that comes with a price tag. So people who do need those size of images can do dump close to 10K on it, & it will do them good work. However, those who just want more cropping room, can get the Sony/Canon for 3K and enjoy almost the same resolution. My conclusion, its not the best camera, the files are to big, the price is to big, its not all that great.

ok, that was just my rant about how this camera seems to be justified by everyone, i just wonder weather you could see a 3 times difference on a normal size print (say 24 by 36) at a few feet away.

The DXO evaluations seem somewhat rough, and the overall score does not always seem to match the sum of the individual scores.

For instance, the 5DII seems to equal or beat the 1DsIII in every category, yet ends up with a lower final score.

The 5DII has higher dynamic range than the D3X from 1600 onwards, but the D3X has clearly higher dynamic range from 400 and lower - a nuanced evaluation would take this into account in its numerical final scores. Unfortunately, only the maximum dynamic range is recorded in the final numbered scores, even though this clear advantage only exists from 400 and below.

The comments at the end of the article about how many full frame chips can fit on a wafer interested me, since this seems to be one of the main factors in the high price of FF sensors. According to one method of estimation I found, a 200mm wafer can hold 61 APS-C chips and 21 FF chips. A 300mm wafer can hold 156 and 59 respectively. That's 3/2.5 to 1 ratio depending on what size you choose (doesn't take into account defect rates).

source:
http://en.wikipedia.org/wiki/Silicon_wafer#Analytical_die_count_estimation

Anything that scores an 88, *I'm* buying. I can see myself out on the street, the camera hung proudly around my neck, saying, "My camera got an 88. What'd yours get?" and then watch them creep away in dismay. By the way, here's a weird thought: 88 in American underground slang stands for HH (the eighth letter of the alphabet) and HH means Heil Hitler; that's why you see skinheads with 88 tattooed on their scalps. Just a coincidence? Maybe.

JC

Dear John Roberts,

I agree with your featured comment overall, but there's a minor false assumption in there that I've seen repeated a lot lately. Namely, if you make really huge prints, you will pick out differences in image quality that you wouldn't see with smaller prints. That's incorrect.

Two aspects of image quality are affected by print size: resolution and noise. Two cameras may very well look identical in these respects in small prints but not in larger ones. The thing is, you don't need to go bigger than 16 x 20, ever. There is no DSLR on the market (and not likely to be one for some time) whose resolution capabilities exceed that of the human eye or photo-quality inkjet printers at a 16 x 20 print size. They don't even come close. At that scale, also, with current pixel numbers, differences in noise will be readily visible. Doesn't mean they will be significant, but if they are you will see them.

Other aspects of image quality aren't even affected by print size; in fact you may have more trouble perceiving them in larger prints than smaller ones. Things like exposure range, color fidelity, curve shape, and the like are as easy to see in an 8 x 10 as a 16 x 20. For many people it's even easier to see them in small prints, because then your eye isn't distracted by "noise" in the visual system (unsharpness, edge artifacts, actual image noise, etc.).


~ pax \ Ctein
[ Please excuse any word-salad. MacSpeech in training! ]
======================================
-- Ctein's Online Gallery http://ctein.com 
-- Digital Restorations http://photo-repair.com 
======================================

While I realize there are many out there who are more interested in the technology involved in photography rather than the practice of photography, I think my head is gonna explode over all the "next big thing" hub-bub in digital cameras. It's pervasive these days on every website to the point it's...well, pointless.

I come from the days when Nikon made an F and Leica made an M and they made them for fifteen years with only minor changes and photographers used them for two decades with only intermittent service and/or repairs. And even though there were no major upgrades or revolutionary developments during that time, it never seemed to inhibit anyone's ability to make good pictures. So it makes me wonder who really benefits from Nikon and Canon's constant improvements. Answer: Nikon and Canon. And, of course, those equipment geeks who talk about sensors, pixels and signal-to-noise ratio rather than actually taking pictures with the stuff they discuss.

I know, it's a natural and a good thing to try to strive for improvement. But it seems to me that we're so caught up in improving the tools of the trade that we often forget to consider the trade itself. One of the great things about TOP is the discussion of Photos, Photographers and Photography. You don't get that everywhere--maybe nowhere else.

Okay. I said it. And my head didn't explode after all.

"I honestly could not see a noticeable difference in print quality among the entries even though I knew that prints in the show came from people using top of the line pro gear, so-called 'entry level' DSLRs, and everything in between."

I've said almost the exact same thing on my blog. And even if there is what seems to a pro to be a big difference, the general public won't see it.
And yet, funny enough, I feel a little urge to defend "numbers". I think we are many who are a bit schizophrenic about these things.

RE: Thom Hogan, he is not decrying the price per se. Instead he is addressing the issue of perceived value. As a camera, it is at the top right now, but in today's market, he argues that Nikon did a poor job of handling the public presentation of the D3X. As a result, the company created conditions in which the nay-sayers are able to dominate the discussions of relative value. He therefore states in his evaluation that Nikon should have done other things like offering better NPS service to enhance the value of D3x package at that price point.

This is a direct quote from his website:

>I know I've been criticized by others for criticizing Nikon's launch of the D3x, but I'm not going to back down on this: it was just plain wrong strategy to just dump the D3x out with a press release and a US$7999 price. The higher the price, the more you have to do to set customer expectations in your marketing. Customers have long memories. The way the D3x introduction was handled is going to be remembered, and subsequent announcements are likely to analyzed even more carefully. If Nikon announces another camera with what appears to be a price premium, I think things will be settled in customer's minds: Nikon charges more. That raises the bar in telling the customer why you charge more. And as I've written several times over the last two months, the real problem with the D3x launch wasn't the price, it was that Nikon did nothing to set expectations nor did they make much of an attempt to explain or support their decision. The higher the price you charge for something, the more you need to handhold the customer and establish a close relationship with them. Nikon did neither. This is a classic marketing mistake.<

Alex

Link: http://www.bythom.com/nikond3xreview.htm

Goddangit! This aint gonna be easy, but put my name down on that D3x boycott list!

The D3X scores an 88? I'm reminded of Spinal Tap: "This goes up to 11."

Interesting comments as usual.

Everyone seems to think the D700x is an odds-on cert. But maybe the last laugh will be on Nikon if this camera never emerges or not until well into the product life-cycle of the D3x (by which time, Canon will probably have moved a few millimeters ahead again with their newest releases).

Certainly if I were Nikon's head of marketing, I'd be tempted to stick with the 24MP sensor in the D3x alone for as long as possible: "if you want the best IQ in the market boys, you've gotta buy this baby". After all, the D700 seems to serve up enough IQ for all but the most demanding mortals (and still beats the 5D II in high ISO performance).

Personally, I'm already enjoying the spoils of this full-frame war. I bought a brand-new, fully-warranted 1Ds III for 3.500€ the other day, a camera I thought I could not afford until it became a museum piece. And once you've used a 1 series, nothing else from Canon (at least) quite does it. . .

I agree with Dogman to a point, but there was a time when the Leica M and Nikon F were the new developments and photographers discussed their relative value. They just didn't have the internet so everyone could join in at the same time. The gap between upgrades may have shortened, but the benefit was there then, and helped get us to the point we're at today.

I could not care less if DxO marks make the D3x worth $8,000 to Joe Pro. That's his call for his profession. I think the real value for the majority of happy amateurs is the trickle down of technology.

You might not be able to tell the difference in a good 11x14 from a D3, 5D, or 20D at ISO 400, but the percentage of keepers and situations where you manage to get any image goes up with the improvements in technology. A stop of light or heavier cropping from D3x technology? For what I shoot (a lot of surfing in low light), I'll take it when it trickles down.

Responding to Dogman’s comments, yes, a serious photographer could shoot with a Nikon F or Leica M for years at a time and you actually did see cameras show some age like the beautifully brassed M a few posts down on this blog but, boy, did we try every film available under the sun! Those were the days of finding a film you simply fell in love with only to use it for a few months and then have the doubt return. You were never 100% positive you had the film you really wanted when grabbing the camera. You never felt at ease with the idea of burning several rolls of several films with only the goal of comparing them. There were too many conversations about cheap Polaroid 35mm film being made by Agfa in Germany, Kodak packaging this B&W film in that other B&W box, questions on why a film came back purple, etc. Looking back, it was madness combined with sadness. Raise your hand if you ever found yourself standing in the film aisle at the drug store or camera shop staring at films and reading cartons. At the time, it was a lot of fun and, granted, quite a bit cheaper than trading DSLRs every few months but it was no way to develop a craft. The more things change...

I salute the artists who chose one film and used it the rest of their lives. It's possible though they chose one particular film not because of love but rather exhaustion from the film merry-go-round.

Great. Now folks are talking about a high score on the DxO scale as having significance.

We don't even know how they tally up the score!

Michael Reichman published a blinded comparison of large prints from a Canon G10 compact vs a medium format digital back, and a large group of experienced photographers could not tell the difference between the two.

Granted, the lighting conditions were favorable to the G10. But the kicker here is that the Nikon D3x has a Dxo score integers higher than the G10.

There is no way to relate a Dxo score to the actual image quality of a particular photograph, or to the actual image quality potential of a particular camera.

This is a study in meaninglessness.

Posted by: Dogman: "I know, it's a natural and a good thing to try to strive for improvement. But it seems to me that we're so caught up in improving the tools of the trade that we often forget to consider the trade itself. One of the great things about TOP is the discussion of Photos, Photographers and Photography. You don't get that everywhere--maybe nowhere else.

Okay. I said it. And my head didn't explode after all."

Hear, hear Dogman!

The Internet photo crowd is overwhelmingly, although certainly not exclusively, guys more interested in "stuff" than in stuff byproducts. Otherwise everyone would just sit in front of their computers all day reading Jorg Colberg's "Conscientious" monologue of "genius". Or the rather less egocentric LensCulture.com which also deals exclusively with photography and photographers rather than photo equipment.

But let's face it; even purist guys who don't consider anything that weighs less than 40 lbs. a "real" camera love a good stuff derby. (Much less so women.) And the dead of winter is an excellent time for such a derby!

Uh oh. I feel an unusual pressure building in my ears and sinuses. I hope it's just a cold coming on.

"This is a study in meaninglessness."

Only if you don't study it. Have you read the whole DxOmark site? I think they do a good job of specifying exactly what they're quantifying. As with any single index (as with, say, a lens rating), it's more useful to break it down into its component parts and weight most strongly what matters most to you (in the case of the Michael R.'s G10 comparison, for example, you would discard the low-light metrics), but that doesn't mean it's meaningless.

Mike J.

I do not own an inkjet printer, so my comments should be taken with a grain of salt. I have seen many discussions on the internet that have the same structure: "Camera X has much higher overall image quality than camera Y, but when you print them out, you can't reliably or easily distinguish which prints come from camera X and which come from camera Y."

I realize that there are many factors that come into play, but is it possible that inkjet printers have a "flattening" effect on image quality differences? Is it possible that there is something about inkjet printing technology that compresses image quality differences that might otherwise be readily apparent?

Dogman,
I'm too young for the days of my youth to become "the good old days," transformed by their distance in the past into glowingly golden wholesomeness. You're forgetting that we have ALWAYS nattered over comparative differences--in those days, pixel-peepers were called "grain sniffers," and people obsessively compared the graininess of different films in a plethora of different developers (I think Arthur Kramer of "Modern Photography" magazine tried every single developer that existed, or close); struggled to figure out the ingredients in commercial chemicals (remember Patrick Dignan?); tested enlarging papers (remember all the fuss about Dmax?); argued about enlarger light sources (remember Zone VI's cold light heads?); and tested and compared lenses--both taking lenses and enlarging lenses--till the proverbial cows came home. At least I did, and a lot of my friends did.

All that sturm und drang *does* seem less meaningful today, because there does seem to be less difference between raw digital results, and there is certainly more such activity today--probably simply because there are more shoppers, shopping more often, who can communicate with each other more directly. But it's not like the only choice in the old days was whether to buy an F or a Nikkormat and then live happily ever after for twenty years. Or were you one of the few who actually did?

Mike J.

Dogman said:

I think my head is gonna explode over all the "next big thing" hub-bub in digital cameras. It's pervasive these days on every website to the point it's...well, pointless. I come from the days when Nikon made an F and Leica made an M and they made them for fifteen years with only minor changes and photographers used them for two decades with only intermittent service and/or repairs."

The Nikon D1 was introduced almost ten years ago (June 1999) and since then we've gone through three full generations of cameras and many more partial (x and h) generations. There has been some tumult.

But if you think this has been hectic, imagine the ten years between about 1840 and 1850, when they went from *no photography* to a widely practiced art...From the invention of the paper negative around 1839 to aerial photography fifteen or twenty years years later...

This last ten years..this is nothing.

DxOmark seems to do a good job of quantifying the properties of the image making guts of cameras. They certainly use more rigorous methods than just about anyone and take the time to explain what they are studying. It's a great resource to balance out the mostly JPEG-based reports presented by the large technical review sites.

In the case of the D3x, they are finding class leading dynamic range at base ISO. Even greater than the Fuji S5 Pro in this regard! True that the 5D II has marginally better DR at high ISO, but an unbiased observer will look at those graphs and easily discern that the big difference is at low ISO. The color depth is also significantly greater for the D3x than for the competition.

In practice, this ought to mean that an ISO 100 D3x NEF (RAW) will hold up to tremendous manipulation, eg, pushing shadows without noise or banding.

There's no D3x in my future, but I can appreciate the progress of technology!

Mike said:

""This is a study in meaninglessness.""

"Only if you don't study it. Have you read the whole DxOmark site? I think they do a good job of specifying exactly what they're quantifying."

---

They specify what they are measuring, but not how they calculate the ultimate number assigned to a sensor. And the article was about how the Dx3 number (88) meant something significant compared to numbers from other cameras (80ish).

But without knowing how Dxo comes up with the final number, the statement that the Dx3 sensor "stands apart" is meaningless.

The Dx3 has a sensor score which is 2 or 3 times higher than the sensor in the G10. What does this mean? Can you tell me?

According to Michael Reichman, the G10 and the Dx3 ( well, to be precise, a medium-format digital back), under the right conditions will produce indistinguishable photographs. What does this say about the meaningfulness of the Dxo score of 88, if even a three-fold difference in sensor score makes no difference in the photographs the cameras actually generate?

What does it say about the Dx3 and, say, the 5DII whose scores are closer? Will the 5DII or the Sony 900 take photographs that will consistently have lower overall IQ than the Dx3 or the D700?

Does the fact that the D700 and the Dx3 both perform chrominance noise reduction on chip before RAW generation poison the Dxo results? Is this taken into account in the final Dxo score? If not, is it fair to say that the Dxo score favors chips who do NR before RAW production? If so, how meaningful is the Dxo number if the final printed image of cameras with different Dxo scores can be the same after PP?

How can we even answer these questions without knowing the proprietary algorithm that Dxo uses to generate the new Holy Grail of sensor quality, the Dxo score?

And if the Dxo score becomes generally recognized as of crucial importance, will we see manufacturers designing sensors to maximize their Dxo scores so that they can advertise their new cam as having a score that "stands apart"?

Will the Dxo score become the new watts/RMS or the new signal/noise ratio?

Dear Philip,

There are definite problems with print fidelity, but I don't think they're limited to just inkjet printers. There is something in the workflow/signal processing somewhere between the RIPping engine and the final output that costs some visible quality. I've discussed this in previous columns: it looks as though subtle gradations get lost. I *suspect* that it has to do with eight-bit workflows, but until we have uniform 16-bit RIP, profiles, and printer drivers and engines, it's hard to be sure.

The thing is, the printers are still very, very good. They do not have perfect fidelity, but no print ever has and never will.

I doubt very much if that the printers are masking the differences. It's more just the perceptual differences between looking at images on a monitor (especially at 100% scale) and looking at images in a print. Different visual expectations, both conscious and unconscious, fundamentally different color spaces, etc.

You might think of the situation as being analogous to audio. You'll never have a perfect set of speakers, but even a moderately decent set of speakers will let you hear differences in the quality of the recording and studio processing.


~ pax \ Ctein
[ Please excuse any word-salad. MacSpeech in training! ]
======================================
-- Ctein's Online Gallery http://ctein.com 
-- Digital Restorations http://photo-repair.com 
======================================

Gingerbaker,
I presume the single-number rating is a weighted average of some sort, but does it really matter? If it bothers you that the single number isn't transparent, then don't look at the single number. If you don't think the breakdown indices are significant, then don't look at DxOmark. Who cares? This stuff is supposed to be fun, not crazy-making. They're only cameras.

Mike J.

Yeah, and remember when Kodak came out with Tech Pan film in the mid-1980s, and the trick was to find the best developer to get continuous tones with max ASA out of this extremely fine-grained film? It went on and on; Kodak came out with Technidol developer and I settled on HC-110 dilution F, cause I used HC-110B for Tri-X (even T-MAX), etc. Well, it's all the same- that quest for the ultimate IQ. Analog or digital, the quest will probably never end.

But then experimenting with Tech Pan WAS cheaper than buying a new digital SLR every few years. I made some neat pix with Tech Pan, too.

Dear Ginger,

The single DxO ranking number is not the Holy Grail of measurement, it's "ranking for dummies." It's for people who can't or won't use the individual test results.

You don't need to know the precise algorithm they use to compute it. They have given you the approximate one. For dummies, that's enough. For you, it would just give you more to argue with.

DxO is reporting on RAW files as they come from the camera. How those files are generated internally is not of concern to photographers unless they're planning to hack their cameras.

It is not supposed to be 'fair' in terms of designs. If that favors one design over another because that design yields cleaner RAW files, then so be it. Again, photographers don't care; they only need to care what the RAW file is like.

Like Mike, I think you really haven't taken the time to understand all the content on their web site, because most of your objections are clearly addressed by the published data.

And, like Mike said, if you're so dead-set against it, ignore it.

pax / Ctein

I am fond of two postings: 'Mine goes to eleven,' and "How To Hold A Nikkormat."

That reminds me of the book that came out when I was in college: "How To Pick Up Girls."

One problem with that 88 score: what are the people who bought that Nikon going to do when another model scores 95 in six months - which will happen.

Finally, earlier you posted a video comparison between color negative film and a D700. The results were blown up to banner size, and the D700 won. How much better do you need? And the D700 is "only" 12mp.

The only time I EVER could tell a format by a print was when I looked at a landscape and
said out loud "That was taken with an 8x10." I was right. How many in this audience are going to start using large format? How many Clyde Butchers are there in this audience?

I thought about LF - I have the skill for it. Then I thought "Do I want to go through all of this just to take a picture?" No. I'm sticking to 35-based cameras. I'm leaning to that new Fuji/Cosina folder, which is as close to LF as I want to get.

Quick prediction (not particularly revolutionary prediction) on direction of photography in the near future.

I think the cameras are going to stabilize and the display media are going to "revolutionize". We have big screen TVs at 1080p resolution.

From what I see, 100 dpi (about 40 dots per cm) is about the upper limit of linear screen pixel density at the present time. I don't see this changing anytime soon.

The price per square inch of screen space is dropping radically. I could easily see 6000x4000 screens in 60x40 inch (152x102 cm) at 100 dpi by the end of the coming decade.

The current "realistic" contrast ratios of current screens give 10 stops of dynamic range. There isn't really a need but I could see it increasing to 12 or 14 stops in a decade.

The bit depth will probably go the same route that camera sensors did: 8, 12, 14, and finally 16.

Anyway...

The moral of the story is that the current crop of "full-frame" sensors will be sufficient for image display for most people in the near future.

The greater dynamic ranges of display screens will trump paper images for most people. Only die-hard print fans will prefer higher resolution to higher contrast.

Anyway...

The last point is that of all of the pixel peepers will get the last laugh since the quality of the image at the pixel level on such a display will become critical to selling images.

El ingles - "a camera I thought I could not afford until it became a museum piece. And once you've used a 1 series, nothing else from Canon (at least) quite does it. . ."


Yep, and it's not the image quality that you don't see until later anyway. I just enjoy using mine. There again, I love my ZI as well and everyone says that 35mm film is dead...

Mike

The comments to this entry are closed.