Pages: [1]   Go Down

Author Topic: D 3 ISO  (Read 4044 times)

MARK WALLACK

  • Newbie
  • *
  • Offline Offline
  • Posts: 34
D 3 ISO
« on: April 03, 2008, 09:24:39 pm »

  The L  1.0 ISO indicates an ISO of 100.    Anyone do any testing or have any knowledge of how a 18X24 (same image, lens) would compare when shot at the 200 ISO setting ? I would guess that the difference would  be  very hard to detect.


Thanks
Logged

routlaw

  • Full Member
  • ***
  • Offline Offline
  • Posts: 160
    • http://www.roboutlawphotography.com
D 3 ISO
« Reply #1 on: April 05, 2008, 12:35:35 pm »

Just picked up a new D3 yesterday, and am starting to work through this stuff now. My gut level feeling is that it will only make a difference if making long exposures but could be wrong. Otherwise I doubt it would have much if anything to do with detail, resolution or color fidelity.

Not sure this answers your question though.

Rob

Quote
  The L  1.0 ISO indicates an ISO of 100.    Anyone do any testing or have any knowledge of how a 18X24 (same image, lens) would compare when shot at the 200 ISO setting ? I would guess that the difference would  be  very hard to detect.
Thanks
[a href=\"index.php?act=findpost&pid=186910\"][{POST_SNAPBACK}][/a]
Logged

MARK WALLACK

  • Newbie
  • *
  • Offline Offline
  • Posts: 34
D 3 ISO
« Reply #2 on: April 06, 2008, 10:13:17 pm »

Quote
Just picked up a new D3 yesterday, and am starting to work through this stuff now. My gut level feeling is that it will only make a difference if making long exposures but could be wrong. Otherwise I doubt it would have much if anything to do with detail, resolution or color fidelity.

Not sure this answers your question though.

Rob
[a href=\"index.php?act=findpost&pid=187244\"][{POST_SNAPBACK}][/a]

Thanks for the reply Rob.
I did some test exposures and prins and can not see any difference. If you do the "pixel peeping" thing there is a very slight difference but again nothing I could see in a print on 17X22 paper.
Would be interested in your experience with the D3
Thanks again.
Logged

jcote

  • Newbie
  • *
  • Offline Offline
  • Posts: 47
    • http://www.johncotephotography.com
D 3 ISO
« Reply #3 on: April 08, 2008, 11:51:16 am »

Mark,

A friend of me sent me this graph. He did not claim to have put it together and I hope I am not violating anybody's rights by displaying it. Whoever did this put in a lot of work and it is one of the best technical illustrations of what various cameras do at different ISOs. Hopefully it is of help with your question.

Logged

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
D 3 ISO
« Reply #4 on: April 08, 2008, 03:07:16 pm »

Quote
it is one of the best technical illustrations of what various cameras do at different ISOs
I'm afraid there is a serious problem with this illustration. It contradicts other findings, among those my own measurements, which indicate, that increasing the ISO by one stop increases the SNR by square root 2, not doubles it as the graphs show (except for the fake ISOs, for example 3200 and 6400 with the D300).

I need to say here, that I measured this on a Canon 40D, but I don't believe, that this camera is so much superior to the D300 and D3.
Logged
Gabor

jcote

  • Newbie
  • *
  • Offline Offline
  • Posts: 47
    • http://www.johncotephotography.com
D 3 ISO
« Reply #5 on: April 08, 2008, 04:43:44 pm »

Quote
I'm afraid there is a serious problem with this illustration. It contradicts other findings, among those my own measurements, which indicate, that increasing the ISO by one stop increases the SNR by square root 2, not doubles it as the graphs show (except for the fake ISOs, for example 3200 and 6400 with the D300).
[a href=\"index.php?act=findpost&pid=188023\"][{POST_SNAPBACK}][/a]

Not to disagree with you but this graph does not directly represent signal to noise ratio. It represents Dynamic Range. While I realize that the two are certainly related and intertwined, I am not sure that there is a 1 to 1 correlation...is there?
Logged

Panopeeper

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1805
D 3 ISO
« Reply #6 on: April 08, 2008, 05:19:51 pm »

Quote
Not to disagree with you

Why not? This should not be a consideration.


Quote
this graph does not directly represent signal to noise ratio. It represents Dynamic Range

The "photographic" definition dynamic range (as opposed to the "scientific"  one) is based on the acceptance of a certain level of SNR. This should not be the only criterion, but that's a different issue.

Anyway, the DR shown on this chart is based on the SNR limit 20, i.e. the standard deviation around the lowest accepted signal is 5% of the signal.

The chart shows for example re the D3, that the DR is reduced by one EV with every stop increase of ISO above 800. This means just, that the SNR doubles by increasing the ISO by one stop.

The consequence is, that ISO 200 with the D2X, ISO 400 for the D200, ISO 800 for the D300 in 12bit mode, ISO 1600 for the D3 and for the 1DsMkII are useless, except when creating JPEG images in camera. I can't test this for myself, but I find it hard to believe and it is certainly not so with my Canon 20D.

When the SNR increases only by sqrt 2 per EV, then the DR is reduced by 1 EV only per every two stop increase of the ISO. This is the result of my measurement.

If you have one of the cameras included in that chart, you can shoot test images and if you upload the raw files, I analyze them.
Logged
Gabor

jcote

  • Newbie
  • *
  • Offline Offline
  • Posts: 47
    • http://www.johncotephotography.com
D 3 ISO
« Reply #7 on: April 08, 2008, 06:19:09 pm »

Actually, now that I have considered what you say...it makes perfect sense. I guess that is why I didn't disagree right off the bat.

Thanks
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
D 3 ISO
« Reply #8 on: April 09, 2008, 09:29:38 am »

Quote
I'm afraid there is a serious problem with this illustration. It contradicts other findings, among those my own measurements, which indicate, that increasing the ISO by one stop increases the SNR by square root 2, not doubles it as the graphs show (except for the fake ISOs, for example 3200 and 6400 with the D300).

I need to say here, that I measured this on a Canon 40D, but I don't believe, that this camera is so much superior to the D300 and D3.
[{POST_SNAPBACK}][/a]

Since the SNR is determined mainly by photon noise, and photon noise is proportional to the square root of the number of photons, doubling of the ISO results in half the number of photons  being collected and the SNR decreases by 1/sqrt(2).

The relationship to dynamic range is less direct and depends on how the DR is defined. The engineering definition of DR is full well capacity/read noise, both expressed in electrons. When full well is not used with higher ISO, one substitutes the number of photons actually collected for full well. With the engineering definition of DR, the noise floor is very low and includes read noise. A photographic definition may set the noise floor higher and include mainly photon noise. See the graphic illustration on [a href=\"http://www.dpreview.com/learn/?/Glossary/Digital_Imaging/dynamic_range_01.htm]DPReview[/url]

Examination of Table 1b on Roger Clark's web site illustrates this. The SNR decreases by 1/1.414 for each doubling of the ISO, but the DR remains within the range of 11 f/stops between ISO 50 and 400. From ISO 800 to 3200, where read noise is relatively constant, each doubling of ISO results in loss of 1 stop in DR.

Roger's Figure 5 in a different post shows a graph including the DR of the IDMII at various ISOs and the contour is quite different from the one shown earlier in this thread, which comes from Bill Claff, who is quite knowledgeable. Roger is using a log-log scale, and you have to take this into account.

From Roger's noise model, it would be possible to calculate the DR for various other definitions of DR (e.g. the DR for a S:N of 20:1), using Roger's equation 2: N = (P + r^2 + t^2)^(1/2). For short exposures, one could ignore thermal noise and consider only read noise and photon noise.

Bill
Logged

Tony Beach

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 452
    • http://imageevent.com/tonybeach/twelveimages
D 3 ISO
« Reply #9 on: April 09, 2008, 01:23:46 pm »

Quote
  The L  1.0 ISO indicates an ISO of 100.    Anyone do any testing or have any knowledge of how a 18X24 (same image, lens) would compare when shot at the 200 ISO setting ? I would guess that the difference would  be  very hard to detect.
Thanks
[a href=\"index.php?act=findpost&pid=186910\"][{POST_SNAPBACK}][/a]

Testing with my D300 indicates adding up to 1 stop of positive EC at ISO 200 and then pulling the conversion in NX or Capture One to be the same as using the ISO Low  settings (the Low settings in fact correspond to positive EC .3, .7 or 1.0).  The only reason to use the ISO Low settings is for studio lighting where the ISO 200 EV is too high or to gain longer exposures than would be possible at ISO 200 -- the price of using ISO LO is a theoretical loss of DR, but realizing the extra DR of ISO 200 requires utilizing ETTR.  My guess is that this all applies to the D3 as well, but that the D3 probably has about one stop more negative exposure latitude (i.e. ISO 200 with no EC on the D3 would have as little noise as ISO 200 with + 1 EC on the D300).
Logged
Pages: [1]   Go Up