« True or False? Snopes the Source of 'Fauxtography' | Main | A Wise, Iconoclastic Voice »

Thursday, 14 May 2009

Comments

I can't decide which ironic comment is more appropriate:

"In that case, I'll go buy one."

or

"In other news: my cat's breath smells like cat food."

(No offense intended.)

Hi Mike, one question about that. I find the DxO mark info incredibly valuable, but I'm curious about something: are all the measurements done in reference to a unit of sensor area, or to the whole image produced?
Because, as I get it, a big sensor, besides the technical advantages, produces something like Ctein's article on noise pointed out, it averages a LOT of variables, noise, etc, into a much better whole outcome than another smaller sensor that might have the same or better ratings measured on a resolution-per-square-millimetre basis, for example.
That changes a lot of things.
For example, if this back's performance is better than the average DSLR in terms of resolution per square millimetre, the advantage on the whole image must be brutal. It would be even if they had the same rating for that matter.

Hey, Phase One: the Big Ten called. They said to stop ripping off their logo.

"Hey, Phase One: the Big Ten called. They said to stop ripping off their logo."

JonA,
At least Phase One doesn't have the numeral "2" incorporated in theirs...what's with the "11"?

Mike, not a fan of college athletics

What I'd really like to see is a graph of cost per point of DxOmark rating. Could be done per-camera, but it'd be more interesting if it was the averaged camera price per five-point spread, with any strong outliers noted.

mcananeya,

You should buy two, as you will need a back-up.

Funny, my dog's breath smells like cat food.

Graydon,

Search on Michael Reichmann's site, he did something exactly like what you describe a few weeks (months?) ago. Although he did it on a two axis graph. IIRC, the Sony A900 (and Nikon D700) scored *very* well by that metric, and understandably the MF digital backs didn't do as well.

Although I can't really tell what the point of the original post is (hmmm: newly revised binning implementation eh?), I can say that if I were to get close enough to my cat to smell its breath, the scent would be nothing like anyone's idea of food, except perhaps a dung beetle's.

Well I should hope so. Geez.

As the late comic Pat Paulsen once observed, "For centuries to come years will pass."

Mike -- since 1990, the Big Ten has had 11 members. Rather than changing their name, they did the "clever" thing with the logo.

Matthew,
So which of the 11 members is not big?

Mike

Mike,
I like your idea of putting a 2 in the Phase One logo. It gets people asking questions about the logo rather than about the $40k.

My vote for the college that isn't big is the one that requires their players to know the difference between "Ten" and 11 upon graduation. The way things are in college sports, I think that's a low priority.

Peter --

Thanks, but that's not quite what I was thinking of. (It's at http://www.luminous-landscape.com/essays/quality-vs-value.shtml if anyone else is interested.)

That essay is making the sensible point that you determine value by the ratio of price and performance. The thing I'm interested in -- since I already knew that was how one figures out value :) -- is how value scales with price.

So, pulling numbers out of the nether regions, the median price for a camera that gives a DxO score between 50 and 55 is 450 USD. The median price for a camera that gives a DxO score between 60 and 65 is 1100 USD, and so on. It's that graph, ideally that graph by release year and scaled by resolution in some way, is the one I'm thinking is particularly interesting. Just the curve shape would say a lot about where we are in terms of the digital camera development trajectory, and the existing DxO graphing functions are way too busy (and too time-invariant) to provide that sense of shape.

"So which of the 11 members is not big?"

- Excellent. I was actually laughing at my desk when I read that...

Best,
Adam

I have no idea where the technologic advances in digital imaging will end up but I have been finding the blog writings of my friend Kirk Tuck, a fine professional from Austin Texas, very comforting. Actually I have never met Kirk but what I see of his work and his writing are enough to make me want to be his friend. Dropping $40k on a camera in absolutely not an option for me and I doubt it ever will be. Thank to Kirk, I have found a cure for upgrade fever. Maybe others will find some comfort in his wisdom.
http://visualsciencelab.blogspot.com/2009/04/everything-old-is-new-againphotography.html

Dear Max,

DxO image quality is measured with reference to the whole image, presented in two ways. The first is "screen" which is essentially pixel-peeping: what you get if you view everything at 100% magnification. The second, and of more practical use, is "print"; images of different pixel counts are scaled to produce the same size print. While it's a derivative measurement, it's a better match to what you'd actually see.

I don't think I understand what you would expect to get out of a measurement that was scaled per square millimeter of sensor. This would not correlate in any way with what you saw in a photograph. It's got no more obvious relationship to what you'd see from the sensor than, say, electron mobility does. It would be a truly useless, and outright misleading, metric for photographers.

It might be vaguely interesting from an information theory point of view; sometimes physical information density is important to doing calculations. I'm noodling around with some of that stuff right now. (In fact, if any of the folks reading this are adept at doing photometric conversions, I could really use some help converting illuminances to luminances. And, no, I don't need pointers to instructional material, I need someone who will actually do some conversion calculations for me. I screw them up. E-mail me if you're volunteering.)

But it's of no value to photographers, so far as I can see.


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

I have to question the academic standards of the Big Ten. I would reduce their football scholarships by 10(11) every year until they change to the Big 11 or boot a school.

The results from DXOMark seem to get more and more curious as time goes on. Looking through the charts comparing the P65+ and the D3X, there isn't a single instance at low ISOs where the P65+ is shown to measure higher. I realize the sensor resolution is taken into consideration and mitigates the lower per pixel performance, but it is still curious.

I also read LL, and realize Mr. Reichmann's comments regarding output from MF as compared to smaller sensors. In reference to sharpeners vs. levelers, I think it may be a matter of imperfect measurement, as compared to something immeasurable.

I was reminded by Stereophile's reposting of the challenge made to Bob Carver. It's not so much that measurement can't quantify a particular "sound". Just that the proper measurement and interpretation is required to determine the underlying source of a "sound". Bob circumvented the whole thing in his challenge, but the point can be drawn from his work.

Ctein --

Performance per pixel isn't so much interesting in the context of "how does this camera doing?" but in the context of "where are cameras going?"

One of the things I'm very curious about is if there's an x86 effect going on with camera sensors (Are all the arguably better clever architectures getting hammered out of existence by sheer commodity mass?); another is if there's a consistent cost scaling for relative performance changes across or within manufacturers. (I note that the performance differences of the Pentax cameras listed at DxO are not significant, for example.)

Ummm, can anyone lend me $40,000. I'm good for it, I swear.

I would be much happier if, rather than making a 65 MP camera, someone would PLEASE make a luminous-only sensor. There are enough people buying black and white film to show that such an option would be well received. Won't someone, PLEASE, think of the children?

agree with mike, sensor for B&W is what a lot of people are waiting for!

A B&W sensor will only give you some what sharper photo's--you will have to do the old color filter bit to get the densities to change in different color areas. Your best bet would be 3 photo sensors with RGB in front of each. cutting out the need for the diffusion filter. It sort of works just fine at this point in the technology of digital cameras & PS CS3 or 4 and the B&W mode. Things will get better in the next 100 years or so. Can't wait-- Shoot B&W film, I have some R,G & B filters you can have.
The Sigma photo sensor should give you better B&W but my tests did not show any major difference.
I have always found that the subject matter & the photographer's creativity & skill is always more important then the camera you are using.
Sorry Mike, I don't see a need for a B&W only sensor--If the Sigma sensor was a true 12 MP I think that would get us 90% of the way there and have some color control with out filters.

Ctein,

Is that MR image also from his fabulous new printer? If so how many really's in the "Really, really, really purdy." are due to the 65+ and how many to the 7900?

I don't know. Call me old, a dinosaur, a Luddite... I work digitally, for both commercial and personal. MF and DSLR. And having viewed another group of clinically perfect, 3x4 footers the other day, I left the exhibit with the feeling that science has no soul.

Dear Mike and Leon,

Your wish is already granted! Such things exist. Megavision makes monochrome medium format digital backs. Kodak produced a monochrome version of the DCS 760. Phase One did a limited run of monochrome digital backs.

All of these are/were very expensive products. They always will be. The reason is that black and white is an extremely small part of the market. 25 years ago, only a few percent of film sales were black and white. The number has dropped since then. I'd be surprised if it were more than 1% when film sales started to collapse about five years ago.

The situation is even worse for digital than film, because with film it only cost you a couple of bucks to switch between black and white in color. With digital, people have to buy a whole 'nother camera.

Put it all together and the inescapable conclusion is that sales of such a camera would be a tiny fraction of what they would be for a conventional color camera, and that means the cost per unit is always going to be high.


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

"All of these are/were very expensive products. They always will be."

You're no fun.

Mike

Dear Tom,

Wrong question.

The right question is how much of it is due to Michael being a really good photographer and master printer.

==============================

Dear Mike,

If someone had access to both cameras (I don't), it would be interesting to see if the DCS 760m actually produces better black-and-white photographs than a black-and-white conversion from today's several-thousand-dollar 15-25 megapixel DSLRs. Wouldn't surprise me if the current "full-color" cameras can run monochrome rings around a dedicated black-and-white camera from several years ago.

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

There is a solution for the Black and white sensor problem.
It also has the advantage that it is cheap, you can buy multiple sensors that give you images that range from almost grain free to golf ball size clumps. In the field they are simple to use (the sensors are sealed when you replace them)and they are easy to carry.
Very few issues with archiving and image conservation, and there are a wide range of printing options.
I was field testing the system yesterday, as I have been for some time. Must say I find the results excellent.

And after extensive testing I have to say that my favorites are two of the Ilford sensors followed by one of the Kodak ones.

:-)

While it's interesting to se how the ~$40k P65+ compares to the Nikon and Canon products, I'm more curious to see how it compares to the (currently) ~$25k Leaf Aptus-II 10. It'd also be nice to see Jenoptik (Sinar) MF backs on the DxOmark website.

And after extensive testing I have to say that my favorites are two of the Ilford sensors followed by one of the Kodak ones.

Any dust problems seem to disappear after one shot too!

The comments to this entry are closed.