Pages: 1 ... 5 6 [7]   Go Down

Author Topic: D800E -Truly Amazing Low-Light Performance  (Read 46138 times)

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Five ISO measures of a sensor/camera sensitivity; some ignored by DxOMark
« Reply #120 on: November 27, 2012, 09:56:24 am »

I've yet to see any tests which demonstrate that, when DXO claim that camera A requires twice the exposure as camera B, for full well saturation, that this is in fact incorrect.
you are not dealing with "wells" - you are dealing with raw files containing the data written there by firmware (and so does DxO) - so just take a proper software like rawdigger and see the raw histogram, that's it...
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Five ISO measures of a sensor/camera sensitivity; some ignored by DxOMark
« Reply #121 on: November 27, 2012, 10:00:09 am »

According to DXO, most manufacturers exaggerate their ISO sensitivity
not most, just some.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: Five ISO measures of a sensor/camera sensitivity; some ignored by DxOMark
« Reply #122 on: November 27, 2012, 10:14:22 am »

you are not dealing with "wells" - you are dealing with raw files containing the data written there by firmware (and so does DxO) - so just take a proper software like rawdigger and see the raw histogram, that's it...

I'm not familiar with rawdigger and have no experience whatsoever with it. I you do, and think it demonstrates that DXO's ISO comparisons are inaccurate, then please show us the results so we may discuss them.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: Five ISO measures of a sensor/camera sensitivity; some ignored by DxOMark
« Reply #123 on: November 27, 2012, 10:30:05 am »

not most, just some.

My experience in relation to the cameras that interest me is, most. It is a rarity that a camera has even nearly the exact ISO specified by the manufacturer. MFDBs are way out. Most Canon and Nikon cameras are somewhat less in ISO sensitivity. It's a rarity to find a camera, such as the Sony RX100, for example, that has a DXO-measured ISO of 101 at its nominated ISO of 100. I also notice that the Panasonic P&S FZ150 has some very accurate ISOs. But such cameras are in the minority.

There are lots of cameras that are significantly less sensitive than the nominated ISO. There are a few that are pretty exact. There are none I've seen that are significantly more sensitive, such as a DXO tested ISO 120, as opposed to the manufacturer's claim of 100.

If I've missed such examples, then please enlighten me. I'm here to learn.
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Five ISO measures of a sensor/camera sensitivity; some ignored by DxOMark
« Reply #124 on: November 27, 2012, 11:22:37 am »

I'm not familiar with rawdigger and have no experience whatsoever with it. I you do, and think it demonstrates that DXO's ISO comparisons are inaccurate, then please show us the results so we may discuss them.

I'd suggest you to get familiar = http://www.rawdigger.com , it is a nice program that replaced Gabor Schorr's (RIP) one

The test is very simple - get a raw file from a camera where DxO says manufacturer's ("nominal") ISO is less than "measured" ISO and identically exposed (aperture, exposure, etc) raw file from a camera where DxO says manufacturer's ISO is pretty much the same as "measured" ISO, both shot with the same "nominal" ISO set in camera (ISO setting is not a part of exposure)... open both in rawdigger and compare the raw data histograms between both raw files in terms of how close the data is to clipping.
Logged

Vladimirovich

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1311
Re: Five ISO measures of a sensor/camera sensitivity; some ignored by DxOMark
« Reply #125 on: November 27, 2012, 11:27:29 am »

MFDBs are way out. Most Canon and Nikon cameras are somewhat less in ISO sensitivity.
well, that was exactly what I was talking about - there is a difference between 1) MFDBs where high ISOs are just values in some raw file tag, but raw data is unchanged numbers wise     2) camera's like m43 from Olympus, where you have a constant sensor undersaturation to safeguard highlights for JPEG shooters      3) the rest of the pack that is either exact match or just "somewhat less in ISO sensitivity"
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Five measures from the ISO 2006 standard for digital still cameras
« Reply #126 on: November 27, 2012, 11:47:23 am »

BJL,
No-one gives two hoots about these ancient standards of ISO. With modern DSLRs, it's sensor saturation which becomes the standard.
"Ancient"?! ISO 12232:2006 Photography -- Digital still cameras -- Determination of exposure index, ISO speed ratings, standard output sensitivity, and recommended exposure index
http://www.iso.org/iso/catalogue_detail?csnumber=37777
was published in 2006 (as the suffix indicates) and is specifically for digital sensors, not film (as the title states).

Since you are apparently ignorant of this current standard, and buying the full document from ISO is a bit expensive just to try to win an internet forum debate, you could try reading secondary sources like this slideshow presentation from someone who is involved in some of this standards work:
http://www.rps-isg.org/DF2008/DigitalPhotographyStandards.pdf

Even Section 4 of the misleadingly named Wikipedia article
http://en.wikipedia.org/wiki/Film_speed
would help.

You would then learn that if indeed DxO claims that "most manufacturers exaggerate their ISO sensitivity",
then DxO is utterly wrong: instead, most manufacturers are following the current ISO standard, using measurements according to either "Standard Output Sensitivity" or "Recommended Exposure Index", as is in fact required of Japanese camera manufacturers. They are simply not using DxO's preferred choice of equating "more than the standard-specified _minimum_ amount of highlight headroom in raw files" with "overstating the sensitivity".
« Last Edit: November 27, 2012, 07:06:14 pm by BJL »
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: Five ISO measures of a sensor/camera sensitivity; some ignored by DxOMark
« Reply #127 on: November 29, 2012, 05:38:05 am »

I'd suggest you to get familiar = http://www.rawdigger.com , it is a nice program that replaced Gabor Schorr's (RIP) one

The test is very simple - get a raw file from a camera where DxO says manufacturer's ("nominal") ISO is less than "measured" ISO and identically exposed (aperture, exposure, etc) raw file from a camera where DxO says manufacturer's ISO is pretty much the same as "measured" ISO, both shot with the same "nominal" ISO set in camera (ISO setting is not a part of exposure)... open both in rawdigger and compare the raw data histograms between both raw files in terms of how close the data is to clipping.

If the test is so simple, Vladimirovich, why don't you do it and show everyone the results. I'm currently travelling with just one camera, the D800E and two zoom lenses. I don't currently have the time nor the opportunity to carry out such tests, but I will emphasise that whenever I come across anomalies or puzzling aspects of camera performance with my own equipment, that suggest a prevailing or authoritative opinion on performance appears to be incorrect, I do my best to test the issue for myself and report on the results, showing my test images on forums such as LL.

The last series of tests I conducted as a result of what appeared to be some anomaly that I'd observed in DXOMark's ISO sensitivity rating, was in relation to the Canon 50D. DXOMark's ISO Sensitivity curves show this camera to have a base ISO of 200. ISO 100 appears to be an expanded ISO, and therefore DXO does not show any DR, SNR or Tonal Range results for ISO 100.

That seemed very odd to me because Canon make no mention that ISO 100 on the 50D is an expanded ISO. Also, my impression is that expanded ISOs below base ISO tend to lose some aspect of performance and gain in other aspects. For example, if the exposure is double, at an expanded ISO below base, one would expect at least some slight loss of highlight detail, or a greater tendency to clipping of highlights, but also one would expect slightly cleaner shadows and better SNR at 18%.

My tests indicated that there was no image quality difference of any practical significance between an exposure at ISO 100, and 1/2 that same exposure at ISO 200, therefore, from my perspective, the issue was settled. I use ISO 200 on my 50D as though it is base ISO, and get the benefit of a possibly sharper image due to the faster shutter speed. I have DXO to thank for pointing this out, even if, as some claim, a mistake was made on their part.

As I've mentioned before, the value of DXOMark for me is the objectivity and consistency of their results. Whenever I've compared the performance of two or more of my cameras, testing for qualities such as DR and noise, I've never found that DXOmarks relative ISO Sensitivity ratings are wrong. It's the relative values that count for me, because at least I have some control over them. I can decide buy a camera with a higher base sensitivity, which has equal or better performance at that higher sensitivity, if I'm disatisfied.

If you have evidence that DXO's misrepresentation of ISO values, or understatement of them is causing a problem, then please explain why and show us your test results. We might all learn something of benefit to our photography.


Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: Five measures from the ISO 2006 standard for digital still cameras
« Reply #128 on: November 29, 2012, 05:59:07 am »

You would then learn that if indeed DxO claims that "most manufacturers exaggerate their ISO sensitivity",
then DxO is utterly wrong: instead, most manufacturers are following the current ISO standard, using measurements according to either "Standard Output Sensitivity" or "Recommended Exposure Index", as is in fact required of Japanese camera manufacturers. They are simply not using DxO's preferred choice of equating "more than the standard-specified _minimum_ amount of highlight headroom in raw files" with "overstating the sensitivity".


That's a very good point, BJL, upon which I think you should amplify. If DXOMark use the terminology, ISO, then the reader would naturally assume that DXO are abiding by the accepted ISO standard, whatever than may be.

If you have discovered that DXO are not abiding by that standard, and have in effect their own standard, and that this is the explanation for the variance between the manufacturer's nominated ISO and DXO measurements, then that's a good point to make.

In itself, the point may be of little practical relevance to the photographer using his camera, but may be of some general interest and could be the subject of a new topic.

If you do decide to raise a new topic on this issue, I have the following questions for points of dicussion, which I would raise in advance.

(1) If DXO are using a slightly different standard than the ISO, is it because they think that their standard is better, or more appropriate within the context of their testing procedures, and/or in relation to the digital medium.

If this is the case, what are the relative merits of the two standards? My guess would be, despite my ignorance, that the true ISO standard is more conservative than DXO's standard, in order to accommodate in-camera jpeg processing which, in my experience, always tends to lose highlight detail.

(2) If DXO are in practice using a different sensitivity standard to ISO, is there perhaps a good reason for their not renaming their standard along the lines of, perhaps, SSSS (Sensor Saturation Sensitivity Standard)?

When placing the cursor over the ISO dot on the DXOMark graph, which produces two readings, the manufacturer-nominated reading of ISO 100 and the DXO measured reading of ISO 87, for example, would it be a good idea for the two readings to be ISO 100 and SSSS 87?

I get the impression that many folks don't go further than the overall, weighted scores for the camera, perhaps because they are a bit intimidated by decibels and EVs. Perhaps DXOMark have decided that such people would be even more intimidated by completely new terms such as SSSS, so they continue to use the widely recognised term ISO which is generally more meaningful.

(3) Does it really matter what standard is used provided that, whatever standard is used, it is consistent? Is that not the only legitimate criticism, a lack of consistency?

When I buy a camera after careful comparison of its specs, my main consideration may be that my choice of camera has a higher base ISO than another, without any compromise in performance with regard to noise and color sensitivity.

Whatever standard is used may be irrelevant as long as the standard is consistent. I don't care if someone gives me a meausrement in centimetres or inches. I can work it out.

Now I hope you can follow this up, BJL, so we can perhaps learn something useful and practical.
Logged
Pages: 1 ... 5 6 [7]   Go Up