Pages: 1 2 [3] 4 5   Go Down

Author Topic: Eyes vs Numbers  (Read 30901 times)

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
Eyes vs Numbers
« Reply #40 on: February 03, 2009, 06:02:38 pm »

Quote from: fike
I don't understand how any algorithm can be so certain that it is only removing unwanted data--noise
Pls read the Canon White Paper

"Noise" is not a simple concept, though it sounds so. The on-chip noise reduction "cleanses" a single bit. The noise reduction you have to with affects ranges of pixels, based on their characteristics, for example on the irregularities where that is not expected (context sensitive noise reduction).

Quote
This stuff is interesting, but not really incredibly relevant to determining whether a camera is good or not.  Or, perhaps I should say, can you explain to me why this is useful or relevant information for the pro or amateur consumer to understand?

1. This stuff may appear to be irrelevant to you; it is very relevant to me (not the on-chip noise reduction).

2. Michael found it relevant enough to mention in "Eyes vs Numbers". *If* it is mentioned, then it should be correctly mentioned.
Logged
Gabor

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
Eyes vs Numbers
« Reply #41 on: February 03, 2009, 06:05:24 pm »

Quote from: Mort54
I find this comment hillarious, given that the D3X delivers more detail in deep shadows than either the 1DsIII or the 5DII at lower ISOs. So tell me again about cutting off a portion of the pixel data ???? Sounds like a rationalization to me.
I am sure you can present raw images demonstrating, that the D3X has the same level (not better) details two stops deeper than the 5D2. Pls use yousendit.com and post here the URL for downloading.
Logged
Gabor

Misirlou

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 711
    • http://
Eyes vs Numbers
« Reply #42 on: February 03, 2009, 06:06:03 pm »

Quote from: Chris_Brown
Michael R.,

Great topic! Thanks for stirring the pot.

For me, a camera is more than a sensor and its supporting electronics. Even though the DxO evaluations are informative, they don't tell us the whole story. Considerations of body design, lenses, portability and software weigh significantly on our choices as photographers (pros & amateurs alike). Some of these factors are given more weight in our decisions than others.

One thing I'd like to point out is that, at present, all camera manufacturers don't seem to be listening to the global user base very well. They seem to be focusing on cost per pixel at the moment, which is quickly coming to a dead end (as Michael's topic illustrates).

Can we say "Mirror lockup button"?
Logged

Mort54

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 590
    • http://
Eyes vs Numbers
« Reply #43 on: February 03, 2009, 06:18:43 pm »

Quote from: Panopeeper
I am sure you can present raw images demonstrating, that the D3X has the same level (not better) details two stops deeper than the 5D2. Pls use yousendit.com and post here the URL for downloading.
I never said two stops better. I just said the D3X is demonstrating better detail in deep shadows at low ISOs than the 5DII is, and no pattern noise to boot. For examples, log onto Digilloyd's subscription site - lot's of real world test images and field shots from both cameras there that compare shadow details.

Sorry, but there's no data to support your contention that Nikon is cutting off pixel data to avoid things like pattern noise. Real world images say just the opposite.
« Last Edit: February 03, 2009, 06:21:33 pm by Mort54 »
Logged
I Reject Your Reality And Substitute My

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
Eyes vs Numbers
« Reply #44 on: February 03, 2009, 06:58:38 pm »

Quote from: Mort54
I never said two stops better
The DxO report said so. You need to decide if you are defending DxO's statements of you make your own conclusion (pls make this clear in your posts).

Quote
I just said the D3X is demonstrating better detail in deep shadows at low ISOs than the 5DII is, and no pattern noise to boot
You need to learn the very basic of this issue:

no raw file = no proof

You can claim whatever you want to, and I can give less than a fig for that.

Quote
For examples, log onto Digilloyd's subscription site
Only suckers are paying for that amateurish review (this opinion is based on the "free" information of that site).

Quote
lot's of real world test images and field shots from both cameras there that compare shadow details
I am sorry, but I don't have time for analysing lots of shots (made by people, who understand nothing of a review). I spend sometimes hours with a single pair of comparative raw images, if they are worth of analysis.

Quote
Sorry, but there's no data to support your contention that Nikon is cutting off pixel data to avoid things like pattern noise. Real world images say just the opposite
Honestly, I am becoming uneasy with this discussion. How do you come to make such a statement? As the matter of fact, I can prove any time what I stated, but I am a bit selective regarding the purpose of my demonstrations, for I doubt that you would understand anything of that.
Logged
Gabor

ndevlin

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 679
    • Follow me on Twitter
Eyes vs Numbers
« Reply #45 on: February 03, 2009, 06:59:54 pm »

I've found that, in digital, the ability to capture shadow detail is of way less importance than the ability to preserve highlights. The reason for this is linked to the 'expose to the right' principle. I want to get my shadow tones (the one's I've decided are the lowest in the scene I care to preserve) as far up the histogram as possible (ie: overexposing them).

A curve can then be applied in post (eg: very simply through the "Blacks" slider or shadow slider in LR). The 'higher' in the exposure these tones are, the better they will reproduce in the final image.  

Of course, the controlling factor is the burn-out of highlight detail. One can only overexpose the image to the point where critical highlight detail can be recovered, making a camera's highlight capture ability paramount to final image quality.

In practical applications, the really low tones get compressed even further in printing and viewing. Not that many images actually need gobs of low shadow detail (I'm talking zone 1 and the space bw zone 1 and zone 2). Nice solid blacks just look good. On the other hand, I find burnt highlights much more aesthetically troubling on a more regular basis.

- N.
Logged
Nick Devlin   @onelittlecamera        ww

douglasf13

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 547
Eyes vs Numbers
« Reply #46 on: February 03, 2009, 07:02:42 pm »

Quote from: Panopeeper
Only suckers are paying for that amateurish review (this opinion is based on the "free" information of that site).

  I am one of those suckers, and unfortunately, I agree  
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Eyes vs Numbers
« Reply #47 on: February 03, 2009, 07:18:45 pm »

Quote from: ndevlin
I've found that, in digital, the ability to capture shadow detail is of way less importance than the ability to preserve highlights. The reason for this is linked to the 'expose to the right' principle. I want to get my shadow tones (the one's I've decided are the lowest in the scene I care to preserve) as far up the histogram as possible (ie: overexposing them).

I am not sure to understand what you are saying here.

The ability to preserve the highlights is a matter of correctly exposing. All the DSLRs can preserve highlights when exposed correctly, even a Canon G10 can do that.

What matters then is the amount of noise present in the 1/3 lower bits of the linear raw data corresponding to the tones of the scene that are 7 to 10 stops darker than those highlights you have avoided clipping through correct exposure.

If those bits are clean, then a suitable curve can be applied to transform the linear raw data into a good looking file. If they are not clean, then the curve that has to be applied will produce poor mid-tones and highlights transitions.

So in the end, the ability to produce noise free darker tones is the single most important aspect of a digital sensor.

Cheers,
Bernard

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
Eyes vs Numbers
« Reply #48 on: February 03, 2009, 07:38:13 pm »

Quote from: BernardLanguillier
The ability to preserve the highlights is a matter of correctly exposing
LOL, isn't it strange that this has to be said again and again?
Logged
Gabor

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Eyes vs Numbers
« Reply #49 on: February 03, 2009, 07:38:51 pm »

Quote from: Panopeeper
1. All Nikon DSLR cameras cut off a portion of the pixel data, thereby removing pattern noise (one could not name this "noise reduction" without provocing the international Nikon community to violent responses). All Canon DSLRs keep the raw data "uncut". This difference may have caused the ridiculous assertion by DxO, that the D3X has two stops greater dynamic range than the 5D2.

Gabor,
Are you implying that DXO has been able to access the D3X raw data before it's been cut off and that the DR results for the D3X are therefore inflated because they include very noisy lower stops of data that simply cannot be accessed in practice when converting raw D3X files?

Quote
I'm afraid, in the DxO evaluation of the P45+ they used ACR for the analysis; if so, that invalidates all of their findings.

Why do you think there may be a possibility that DXO are using ACR? They analyze the raw data before demosaicing and conversion, don't they?


Logged

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
Eyes vs Numbers
« Reply #50 on: February 03, 2009, 07:51:51 pm »

Quote from: Ray
Are you implying that DXO has been able to access the D3X raw data before it's been cut off and that the DR results for the D3X are therefore inflated because they include very noisy lower stops of data that simply cannot be accessed in practice when converting raw D3X files?
I am not implying anything regarding the DxO results, except that they had been achieved at incorrectly. If they would publish their sources (i.e. raw files, like Imaging Resources are doing) and if they had published their examt measurements, like I am doing, then and only then we could talk about that factually. Until then their "results" are nothing but opinions in my opinion.

Quote
Why do you think there may be a possibility that DXO are using ACR? They analyze the raw data before demosaicing and conversion, don't they?
I think you should ask this question from DxO. What I do know is, that their results re the P45+ DR and ISO characteristics are telling a tale of their ineptness.
Logged
Gabor

Mosccol

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 88
Eyes vs Numbers
« Reply #51 on: February 03, 2009, 07:54:28 pm »

Quote from: BernardLanguillier
I am not sure to understand what you are saying here.

The ability to preserve the highlights is a matter of correctly exposing. All the DSLRs can preserve highlights when exposed correctly, even a Canon G10 can do that.

Hi Bernard

I guess that what he meant related to the now well trodden 'expose to the right' approach: when shooting in RAW (and ONLY in RAW), then overexposing preserves more information in the highlights. It requires post processing (back to the 'left') but seems to be worth it.

Here is Michael's article on the subject:

http://www.luminous-landscape.com/tutorial...ose-right.shtml

François
Logged
Canon 7 & L Glass + iPhone 6 ... Where n

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Eyes vs Numbers
« Reply #52 on: February 03, 2009, 08:00:05 pm »

Quote from: Mosccol
Hi Bernard

I guess that what he meant related to the now well trodden 'expose to the right' approach: when shooting in RAW (and ONLY in RAW), then overexposing preserves more information in the highlights. It requires post processing (back to the 'left') but seems to be worth it.

Sure, I am familiar with the ETTR dogma.

EETR can sometimes reduce the negative impact of noisy shadows, and is a way to optimize the available DR of a given sensor, but but the fact remains that what drives the quality of the file in the end is shadow noise.

Cheers,
Bernard

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 13983
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Eyes vs Numbers
« Reply #53 on: February 03, 2009, 08:07:28 pm »

Quote from: Panopeeper
LOL, isn't it strange that this has to be said again and again?

Somehow it is understandable.

- The camera histogram does often not reflect the actual saturation of the photosites - which gives the illusion that highlight recovery is real,
  - Impact of WB,
  - Internal jpg vs raw

- The raw converters often provide the ability to reconstruct some blown channels from non blown ones and surrounding data  - which again gives the impression that highlight headroom is a reality,

Cheers,
Bernard

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
Eyes vs Numbers
« Reply #54 on: February 03, 2009, 08:13:02 pm »

Quote from: Mosccol
when shooting in RAW (and ONLY in RAW), then overexposing preserves more information in the highlights. It requires post processing (back to the 'left') but seems to be worth it
The problem of a whole (ok, a half) world of digital photography is compressed in this statement:

what does overexposing mean?

Overexposing does not preserve but destroys information in the highlights, which may or may not be "recovered" (in cleartext: guessed). The question is: when does overexposure start?]/i]

Edit: Bernard, I started writing this post but did something else in the meantime, came back and posted it, then I saw your post; that's the reason of overlaps.
« Last Edit: February 03, 2009, 08:16:13 pm by Panopeeper »
Logged
Gabor

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Eyes vs Numbers
« Reply #55 on: February 03, 2009, 08:14:12 pm »

Quote from: Panopeeper
I think you should ask this question from DxO. What I do know is, that their results re the P45+ DR and ISO characteristics are telling a tale of their ineptness.
 

How do you know that if you don't have the cameras or the raw files to compare?

You probably recall that I went to some trouble in Bangkok comparing the Nikon D3 with my 5D at high ISO. I came to the conclusion that the high-ISO advantage of the D3 was in the order of 1/4 to 1/2 a stop, at various simulated ISOs above ISO 1600. This result seems to correspond with DXO's result which places the DR of the D3 at 1/2 a stop greater than the 5D at ISO 1600 and above.

What interest me is that DXO claims there's an even greater DR difference between the D3 (or D700) and the 5D at base ISO; more than one stop greater DR. This I did not test when I had the opportunity. But I now have the opportunity since I have both a D700 and a 5D.

If I were to repeat the tests in more ideal circumstances and demonstrate that the D700 really is 1/2 a stop better at ISO 1600 and above, and really is a whole stop (or more) better at base ISO, would you accept that DXO results are accurate?  
Logged

otbricki

  • Newbie
  • *
  • Offline Offline
  • Posts: 3
Eyes vs Numbers
« Reply #56 on: February 03, 2009, 09:05:23 pm »

Quote from: John Camp
I honestly don't think this will ever happen. If you look at Sean Reid's review site [a pay site], you'll see that he often refers to a way a lens "draws." So in a review of high-end normal lenses, he gave excellent marks both to a Nikon and a Zeiss 50mm, said either one is capable of doing any kind of professional work, but he said that they draw differently and he prefers the Zeiss. So in this case, part of the "quality" equation is measuring Sean Reid's preferences, which may change from time to time. On the Online Photographer recently, Mike Johnston, who knows a lot about lenses, panned a lens because of its distortions, but actually liked the "look" it produced. In audio, you have large numbers of people who agree (apparently) that some amp/speaker systems sound better than others, despite measures to the contrary. It seems to me that the problem is with the engineers, not with the ears. Engineers have a bias toward measurement, which is a real bias, because if only 20% of a system is currently measurable and quantifiable, it's possible that the other 80% is more important in every way -- yet the engineers will tend to go with what they can measure, because that's their bias. It's the old "bumblebees can't fly" problem -- bumblebees CAN fly, the engineers just didn't have the ability to quantify the characteristics that allowed the bumblebees to do so.

The DXO problem is that they insist that they have a measure, but it's not a real measure. They have a whole bunch of measures, but those measures don't include the whole universe of possibilities, and without the whole universe, you can't reach a valid conclusion. And it may be inherently *impossible* to measure the entire universe of possibilities. So, if a lot of people with good, trained photographic eyes say that a large P65 print is better than a large D3 print, and a DXO measurement says it's not, then the eyes are right, just as they're right when they see a bumblebee fly.

JC

The problem is not that a person "prefers" a certain look or not, or that they have some ability to achieve a holistic view of image quality, but rather when that person is shown images produced by that same process without telling him what the process was, can he reproducibly pick the images that were produced with a given process? Do you really think photographers can look at an image and reliably determine a priori what camera and lens produced that image under normal viewing conditions? My guess is right now the only way to do that would be under conditions where a camera's particular limitations are present - pattern noise in shadows and the like.

It is the same thing with amplifier/speaker preferences. Some people claim to have these preferences, but when they are confronted with a listening test between two measurably equivalent amplifiers where they do not know what the amplifier is they cannot identify which amplifier is in use when the amplifiers are exchanged.

It is only the placebo effect in play here; that any perceived differences are purely the result of psychological bias, rather than some mysterious paranormal sensory ability that transcends measurement science.

I don't think most people have a clear understanding how advanced measurement technology actually is. Today a chemist can detect a single molecule in a sample containing more molecules than there are stars in the universe, and determine its structure in detail at the level of individual atoms; a physicist can measure time to one second in 100 billion years. Mere human sensory ability is effectively infinitely crude in comparison.

In the audio field nobody has been able to hear audible differences in a controlled test that are not explained by electrically measurable phenomena for decades now. There have been offered prizes up to one million dollars for anyone who can do this. These prizes go unclaimed. Unfortunately the charade of the golden ear continues, but it is only that, a charade perpetuated by audio salesmen and high end companies that pander to these beliefs.

There is no reason whatsoever than photographic images should not be subject to exactly the same rules. In fact I think it should be far easier for photographs than audio because the reproduction is a static digital file rather than a transient collection of pressure waves in space.

DXO is off to a start, however there is work to do. I am optimistic that the day will come when measurements will provide a reliable guide to image quality.

Logged

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
Eyes vs Numbers
« Reply #57 on: February 03, 2009, 09:18:50 pm »

Quote from: Ray
How do you know that if you don't have the cameras or the raw files to compare?

There are tale tell signs.

Quote
If I were to repeat the tests in more ideal circumstances and demonstrate that the D700 really is 1/2 a stop better at ISO 1600 and above, and really is a whole stop (or more) better at base ISO, would you accept that DXO results are accurate?
I can not relate the results of your comparisons to anything; we are comparing the digital images in different dimensions. Your cat shot an ISO 156,800 on a 2310 Mpix 6x6cm camera illuminated by a 60W lamp in the kitchen looks ugly compared to my dog shot at ISO 100 on an APS-C camera in daylight if printed 20mmx30mm on an Epson 9900 printer.  (This was meant to illustrate how I see your comparisons.)

If you shoot a very clean gray scale card, *very* evenly illuminated, underexposed two or three stops (highlights are irrelevant), I can take measurements from those shots (if you throw in a color checker card too, even better). Equal exposure of the shots between the cameras is not relevant; low exposure is essential.

A really exciting shot for measurement looks like this Stouffer wedge (but it is not easy to make this shot correctly).
Logged
Gabor

kwalsh

  • Full Member
  • ***
  • Offline Offline
  • Posts: 101
Eyes vs Numbers
« Reply #58 on: February 03, 2009, 09:56:16 pm »

Quote from: otbricki
I don't think most people have a clear understanding how advanced measurement technology actually is. Today a chemist can detect a single molecule in a sample containing more molecules than there are stars in the universe, and determine its structure in detail at the level of individual atoms; a physicist can measure time to one second in 100 billion years. Mere human sensory ability is effectively infinitely crude in comparison.

In the audio field nobody has been able to hear audible differences in a controlled test that are not explained by electrically measurable phenomena for decades now. There have been offered prizes up to one million dollars for anyone who can do this. These prizes go unclaimed. Unfortunately the charade of the golden ear continues, but it is only that, a charade perpetuated by audio salesmen and high end companies that pander to these beliefs.

DXO is off to a start, however there is work to do. I am optimistic that the day will come when measurements will provide a reliable guide to image quality.

I had something similar to some of your comments to say, but I've got to play devil's advocate here first!

Based on the above snipped quote I think you might be missing a little bit of the point.  The article isn't saying that you can't measure image quality.  The point I believe is that when your chosen numerical metric deviates from what the community views as "good" it means you have failed to come up with a useful metric.  Doesn't matter how accurately you measure it, you are measuring the wrong thing.  Michael gives a number of reasons why he thinks the DXO metric fails the "smell" test from the community.

Similarly, I'll take issue with your comment on the audio field.  I also think the Golden Ears are mostly full of it (was going to be the point of my post) but your post again made me think you aren't getting the right message from Michael's article.  The point is that THD in fact is an awful metric for audio quality and this is well established for exactly the issues Michael outlines (odd vs. even harmonics). Blind studies clearly show that listeners easily prefer an amplifier with a high THD that has even harmonics to an amp with much lower THD that is composed of odd harmonics.  In fact, listeners will tolerate well over an order of magnitude more even harmonic distortion than odd.  These aren't idiots comparing thousand dollar AC power cables and claiming they can tell the difference, they are effects trivial to measure with a spectrum analyzer and easy for an average person to hear.  This issue is that the industry choose a metric that poorly correlated to perceived audio quality and it was obvious to everyone in the community.  There are two separate issues here.  One is crazy folks claiming they hear things that are placebo effects.  The other is engineers telling people one thing is better than another because of some numbers a spectrum analyzer spit out and not realizing their metric was totally screwed up because they had the wrong psycho-acoustic model.

Now - that all said - I was going to make a point about the audio analogy as well.  And that is much along what your comments are with the regards of the crazy Golden Ear set not understanding what a double blind study is and advocating people spend $500 on a wood knob for their volume potentiometer.  If measurements are useless when the metrics they produce aren't correlated to community perceptions then one must be equally careful with the methodology used to gather data from humans.  The Golden Ear's are world renown for their complete failure to understand human subject methodologies.  From all the things I've read from Michael over the years I don't think he suffers the same fault.  For example in his oft cited G10 vs. MF test he presented images blind to his subjects, and he clearly knew the limits of what they were comparing.

Ken

P.S. First post here, short intro.  Photographer, amateur, have shot 35mm and 4x5 film and now all digital.  Tend to do web and modest prints so I stick to entry level DSLRs but still am a sucker for high dollar lenses.  Have always liked the tone of conversions here so I finally registered.  That said, some of the discussion in this thread has started to tend towards the DPRish .
Logged

Leping

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 87
    • http://www.lepingzha.com
Eyes vs Numbers
« Reply #59 on: February 03, 2009, 10:04:32 pm »

I tried not to jump to the discussion, since I am trying to finish an essay that in many way echoes Michael's words.  But it is not very often to see level of ignorance in some of the posts.  So, just a very few quick points.

We should all calm down and to try to understand what DxO really said.  There is a wealth of information in the DxOMark site, but poorly organized and presented.  How many bothered to read their technical definitions, formulations, testing methods, and the most important measure, the "3D" "Full SNR" charts, that was inproperly burried deep?

Nikon's clamping the black point at the mean read noise level is not an NR, and does not affect the noise in the shadows, as in the DxO SNR measurements.  The lower read noise of D3x is confirmed by eyes in many carefully conducted tests, when shadows are pushed up 2-4 stops, and the different is anything but subtle.  Weather the SNR over the 0.1% luminance region important is though totally a different question.  If you do not push the shadows up, if you do bracketing on every shots, then the answer is clearly no.  But the numbers and the pushed-up shadows do tell the differences the careful numerical analysis could reveal, even it is only an academia issue.  And the carefully measured and carefully tested did happn to agree almost perfectly.  

Are you aware that, despite of the common quick believe, actually DxOMark revealed even at low ISOs the Canon's noise is better than the D3x's from the midtone up, as noise performance can not be described as a single number even for a fixed ISO?  Did you see DxOMark actually said the 5DM2 is the least noisy camera at high ISO?  DxO does fail in several very important fronts, for example, to address the resolution issue (so the D3/D700 received overall scores improportional to their image quality, mostly the true resolution and the "looking"), to analysis not only the standard deviation but other noise measures such as luma/chroma noise spectrum and energy-spatial frequency distribution, and to intelligently construct sensible performance merit scores out of these raw numbers, which, exactly as MR said, by themselves can be valid while tend to grossly mis-lead (such as the 18% gray SNR).

When DxOMark said they used ACR for the Phase One P45+ analysis?  Calm down, read and think more, understand what the numbers really mean, and do they really help or harm your school of arts.

"Only suckers buy paid reviews", so all the magazine subscribers and buyers are in that category?

Also very interesting to me (but most did not agree), the MF lens quality is a very important contributor to the "MF look" -- my Pentax 67 prime and zoom lense raised my 5DM2 image quality one big step up.

I just want to stop here, as we seem all agree pretty well that DxOMark only provides one side of the story.

P.S.: Happen to be also an audiophile, I fully understand what MR said, and totally love LPs and tubes.  Many great photographers are musicians, Ansel Adams and Charlie Cramer included.  Photography is a form of art, not science about the numbers, although from time to time we need ways to justify the amount of money we ponder weather to spend, so we want to spend time to fully understand some numbers rather than treating them as total nonsense.

Quote from: Panopeeper
I read following statements in Eyes vs Numbers:

I feel following needs to be said, for Michael's statements may be taken by some readers too literally; this is not for or against any camera but for the better understanding of these issues.

1. All Nikon DSLR cameras cut off a portion of the pixel data, thereby removing pattern noise (one could not name this "noise reduction" without provocing the international Nikon community to violent responses). All Canon DSLRs keep the raw data "uncut". This difference may have caused the ridiculous assertion by DxO, that the D3X has two stops greater dynamic range than the 5D2.

2. Noise reduction in the sense most people understand it will be performed on raw data only by Sonys, at least by the A700 and A900. This step could not be disabled with an earlier firmware version of the A700, but now OFF means OFF, apparently.

3. On-chip (on-sensor) noise reduction occurs with all CMOS sensors, which are inherently more noisy than CCD. However, that is single-pixel noise reduction, which has absolute nothing to do with the noise reduction in the sense as it is done in raw processing software or by the mentioned Sony cameras, namely based on the context of pixels.

4. The raw data of the Phase One cameras is far from being really raw.

Finally, there is nothing wrong in analyzing raw data and making certain conclusions based on that; at least I am doing just that without feeling guilty. In fact, one can make certain findings only based on the raw data. However, one needs to be careful with those findings. I'm afraid, in the DxO evaluation of the P45+ they used ACR for the analysis; if so, that invalidates all of their findings.
« Last Edit: February 03, 2009, 10:21:29 pm by LEPING »
Logged
Leping Zha
www.lepingzha.com leping@att.
Pages: 1 2 [3] 4 5   Go Up