How come the difference?

[semilog]

^------- ;-)

Rockwell knows what he's saying, and if you read his reviews closely he's actually technically competent. But he values different things than DxOmark do. Specifically, he values the lack of an AA filter, and he days so in his review. For him, that trumps being able to shoot at ISO 25000. He also is looking at system-level performance, and he likes the size-performance ratio of the M system vs. say a 5DII.

 

[semilog]

That, and he writes in Truthiness. I believe that opinions are not facts, so I'd believe tests from a methodological means of comparative testing (even if it were a little biased) more as facts rather than for-profit opinionating (new word).

By the way, Rockwell is not the only well-informed person who takes issue with aspects of DxOlabs' weighting & interpretation of sensor parameters -- specifically, their weighting of pixel-level vs. array-level sensor performance. And don't make the mistake of thinking that DxO are not in it for the Bejamins.

 

[Bill Pierce]

An interesting interview with a Kodak employee back in 2007.

http://pluggedin.kodak.com/pluggedin/post/?id=664294

He talks about CMOS sensors catching up with the CCD. He talks about low dark current pinned photodiodes and a four-transistor pixel architecture, which provide lower noise readout methods among other things. And, interestingly enough, he is talking about small sensors.

 

[MartinP]

One of those things measured a few features of the sensor and software, the other is an opinion about the entire camera - both could be 'true' (for an indefinable value of 'true').
🙂

 

[former member]

The CMOS/CCD article is an interesting read. New technology for sensors will be interesting to watch.

Kodak has just about gotten out of the CMOS business, and appears to be in the process of selling the unit. They invented digital photography, had a line of innovative professional camera, then appear to have packed it in. Their alliance with Leica at least keeps them in the high-end camera business. This is a company that needs some direction.

 

[rpsawin]

If they weren't testing raw data directly off of the sensor, they were testing the signal processing algorithms of the respective cameras.

Kodak publishes the long data sheets for their sensors, most other companies do not. It's hard to know the real signal to noise ratio of a camera without the specifications of the detector.

Jeeze I love my MP even though it sadly lacks "signal processing algorithms"...😀

Best,

Bob

 

[Peter Klein]

DxOmark is measuring things that can be measured objectively. What they don't measure is how photos taken with the measured sensor actually look to humans. That would be subjective, unscientific, and subject to ridicule and scorn on Internet photo forums.

So I might trust DxOmark sensor ratings to tell me that, in the lab, the Nikon D90 used to give the best overall performance (as they define it) of APS-C DSLRs, and the Pentax K5 now has that mantle. Or that if I really want to take photos in the very dark, I should seriously consider a D700. But they don't tell me what kind of noise I'm getting, whether a bit of it makes the picture usable or not, or how well it can be reduced in a good noise reduction program.

They don't tell me how well the camera will work for traditional available light work, as opposed to sports shots with a slow tele-zoom where you need to stop the motion. They don't tell me that by exposing at 640 with the exposure compensation set at -1 stop, my M8 can get an equivalent ISO 1250 that is a bit better than the actual ISO setting.

And they don't tell me how the camera feels in my hand, or whether the menus and buttons are fast/intuitive or slow/fiddly/confusing.

So if I was in the market for a new camera, I'd probably peruse the DxOMark ratings for a general idea of things. But I would rely far more on actual sample pictures, including a bit of dpreview's test shots, and a lot of other people's real pictures. I would also look at actual prints if possible--and if not, I'd view shots not just at screen size and 100%, but also at 50% and 33%, where they give a much better idea of what a print will look like.

I did all this when I was deciding whether to buy my M8. And I decided that it was good enough that, with the lenses I already had, I would be just fine. I sometimes wish I had a stop or two more high-ISO, but not enough to carry a D700 around. Your mileage may vary.

Now, if somebody came out with a camera that DxOMark rated with D700-like high ISO performance, and it had M-lens compatibility and a rangefinder, I'd consider it seriously. But I'd still want to see the pictures.

--Peter

 

[PKR]

I don't trust Ken.

From his page:
Begin Quote
. I occasionally weave fiction and satire into my stories to keep them interesting. I love a good hoax. Read The Museum of Hoaxes, or see their site. A hoax, like some of the things I do on this website, is done as a goof simply for the heck of it by overactive minds as a practical joke. Even Ansel Adams kidded around when he was just a pup in the 1920s by selling his photos as "Parmelian Prints." I have the energy and sense of humor of a three-year old, so remember, this is a personal website, and never presented as fact. I enjoy making things up for fun, as does The Onion, and I publish them here — even on this page.
End Quote

I understand the above may have been met as a cute disclaimer. I just don't have time to ferret out his "facts".

http://www.kenrockwell.com/about.htm

 

[PKR]

If they weren't testing raw data directly off of the sensor, they were testing the signal processing algorithms of the respective cameras.

Kodak publishes the long data sheets for their sensors, most other companies do not. It's hard to know the real signal to noise ratio of a camera without the specifications of the detector.

Sony won't supply Sensor Data? Is this correct Brian?
pkr

 

[Paul Luscher]

I guess my thought is "who cares what Dxo thinks about the M9?" Works good enough for me.

Reminds of the review I read of the Olympus E-3 some time back, which complained about the "slight softness" of the camera images. And the reviewer was right--once you had blown up a segment of the image by 20x or 30x...but in reality-land....

Sometimes I think we can't see the forest for the pixels....

 

[semilog]

Sony won't supply Sensor Data? Is this correct Brian?
pkr

They certainly supply full data for their scientific sensors. I don't know about the consumer stuff.

 

[former member]

I cannot find the data sheets for the sensors that they use in consumer cameras. You can find the "Glossy Brochure" type stuff, but I have not seen the traditional data sheets.

If you find them, post the links here:
http://www.rangefinderforum.com/forums/showthread.php?t=106571

Interesting enough: Sony's listed Data Sheets covers their CCD sensors, not the CMOS sensors.

http://www.sony.net/Products/SC-HP/datasheet/index.html

But you cannot just download the spec sheets without printing out and signing a user agreement and returning it to Sony...

I had an NDA with Kodak in the 1980s, but that was for products still in development. Datasheets for released products, you can just download.

 

[semilog]

So, it's an interesting thing that in scientific imaging CCD is king, and has been for a couple of decades due to better pixel-level performance and higher uniformity... but let's just say that this morning I'm going to see a man about testing a newly-developed high-end scientific CMOS array detector...

 

[Gabriel M.A.]

By the way, Rockwell is not the only well-informed person who takes issue with aspects of DxOlabs' weighting & interpretation of sensor parameters -- specifically, their weighting of pixel-level vs. array-level sensor performance. And don't make the mistake of thinking that DxO are not in it for the Bejamins.

Being well-informed does not make one's opinions factual. Opinions are opinions, and factual research results are factual. Opinions vary, as well as research parameters. You cannot (sorry --should not) be biased against methodology and be biased towards opinion and then believe that one makes one more true than the other.

 

[semilog]

You cannot (sorry --should not) be biased against methodology and be biased towards opinion and then believe that one makes one more true than the other.

What on earth are you talking about?

 

[kbb]

Last year, digital was junk; a film Leica was the best camera created on the earth, according to Ken.

 

[semilog]

^----- 4 days ago, @kenrockwell: "LEICA is the best. There is no substitute, in fact, even LEICA sadly tends to make cheaper and cheaper cameras as time goes on. Although the prices go up, the intrinsic quality and function has been slipping due to a weak market for rangefinder cameras ever ever since the highpoint of the 1954-1967 LEICA M3."

You don't have to agree with him, but Rockwell is not being inconsistent on this point.

 

[Bike Tourist]

I like to read Ken Rockwell when I need a smile or maybe a laugh. He's entertaining. I would put his expertise, however, about on a par with mine, which is to say, none at all.