Testing DxO DR numbers with an experiment

Started Nov 7, 2010 | Discussions
Horshack Veteran Member • Posts: 8,544
Testing DxO DR numbers with an experiment

In the K5 DxO thread that just reached 150 posts, Iliah Borg suggested shooting a newspaper with gradually decreasing exposure and looking at the text readability as a means of confirming DxO's DR numbers. I took that suggestion to heart and tried the experiment on my D5000 @ ISO 200. DxO reports 12.31 stops of DR for this sensor/ISO. I used an ISO 12233 chart.

I shot with a 18-105mm lens @ 35mm f/5.6, with exposures ranging from 5-seconds to 1/1600 seconds (14 stops). I would have liked to had better light to avoid long-exposures but couldn't arrange it. Long-exposure NR was on. Lighting was a non-so-great kitchen fluorescent with a LR3-measured WB of 3400 +26. All photos have been set to that WB in LR and there is some color shift for the faster shutter speeds as a result. Admittedly the lighting for the experiment is very far from ideal.

All photos shot RAW on tripod with timer release and delayed exposure, processed in LR3 with default settings (Adobe Standard 2010, NR: Chrominance +25, Luminance +00). I pushed the exposures in LR starting at the 1/6 photo (6th photo) by one stop and increasing by one additional stop thereafter. Focus was manual and locked for all shots.

Stops 1-6: (f 5 seconds through f 1/6)

Stops 7-12: (f 1/13 seconds through f 1/400)

Stops 13-14: (f 1/800 seconds through f 1/1600)

-- hide signature --

Kodak Instant Camera
Kyocera 1MP Camera phone (pre-paid phone plan)
http://horshack.smugmug.com/

rhlpetrus Forum Pro • Posts: 26,131
Re: Testing DxO DR numbers with an experiment

I'm not getting images.

One question: how did you set clipping point?

-- hide signature --

Renato.
http://www.flickr.com/photos/rhlpedrosa/
OnExposure member
http://www.onexposure.net/

Good shooting and good luck
(after Ed Murrow)

 rhlpetrus's gear list:rhlpetrus's gear list
Leica D-Lux (Typ 109) Nikon D7000 Nikon 1 V1 Nikon AF-S DX Nikkor 12-24mm f/4G ED-IF Nikon AF Nikkor 35mm f/2D +4 more
_sem_ Veteran Member • Posts: 5,033
Re: Testing DxO DR numbers with an experiment

The test does illustrate the forensics involved in searching the bottom DR stops and their general photographic uselessness.

Their uselessness would be even better illustrated with a colour target (colour info gets lost even earlier than BW)... In other words, if you you care for colours in the shadows you can lift less than if you're satisfied with rather monochromatic shadows.

The K5 and D7k do show some improvement compared to the D90 generation - not at the same ISO but by extending the curve to lower ISO. The D3s is much ahead at the same ISO.

OP Horshack Veteran Member • Posts: 8,544
1st photo is completely clipped, 2nd is at clipping

So the 1st stop for DR purposes is the 2nd photo.

rhlpetrus wrote:

I'm not getting images.

One question: how did you set clipping point?

-- hide signature --

Kodak Instant Camera
Kyocera 1MP Camera phone (pre-paid phone plan)
http://horshack.smugmug.com/

Stujomo
Stujomo Senior Member • Posts: 1,629
Re: Testing DxO DR numbers with an experiment

After performing the test how many stops of useful dynamic range to you feel the D5000 has.
--
Stuart...

http://www.sjmphotography.co.nr/

http://sjmphotography.wordpress.com/

 Stujomo's gear list:Stujomo's gear list
Nikon D1H Nikon D700 Nikon D800 Sony a7 Nikon AF Nikkor 50mm f/1.8D +6 more
OP Horshack Veteran Member • Posts: 8,544
10.5 - 11 stops on a luminance basis

Stujomo wrote:

After performing the test how many stops of useful dynamic range to you feel the D5000 has.

"Useful" is the key word here. The last stop in the shadows is a tossup and its usability depends on the particular photo and how much detail/color retention is needed. On a luminance-only basis I would say the DR measured in this lighting setup is 10.5 - 11 stops.

rhlpetrus Forum Pro • Posts: 26,131
Re: 10.5 - 11 stops on a luminance basis

What conversion settings did you use?

Horshack wrote:

Stujomo wrote:

After performing the test how many stops of useful dynamic range to you feel the D5000 has.

"Useful" is the key word here. The last stop in the shadows is a tossup and its usability depends on the particular photo and how much detail/color retention is needed. On a luminance-only basis I would say the DR measured in this lighting setup is 10.5 - 11 stops.

-- hide signature --

Renato.
http://www.flickr.com/photos/rhlpedrosa/
OnExposure member
http://www.onexposure.net/

Good shooting and good luck
(after Ed Murrow)

 rhlpetrus's gear list:rhlpetrus's gear list
Leica D-Lux (Typ 109) Nikon D7000 Nikon 1 V1 Nikon AF-S DX Nikkor 12-24mm f/4G ED-IF Nikon AF Nikkor 35mm f/2D +4 more
unlimited New Member • Posts: 6
Re: Testing DxO DR numbers with an experiment

Hello,

Do you mind to share RAW files for the last 2 shots?

Thanks,
Xdrew

Stujomo
Stujomo Senior Member • Posts: 1,629
Re: 10.5 - 11 stops on a luminance basis

I would redo the test but use a plain white towel with plenty of texture. At the moment you have very high contrast test chart which may or may not represent what you may have in the shadows of a reall world scene. Give it another go with a towel to compare the results and then try the test again with a small group of objects. I would be interesting to compare the results.
--
Stuart...

http://www.sjmphotography.co.nr/

http://sjmphotography.wordpress.com/

 Stujomo's gear list:Stujomo's gear list
Nikon D1H Nikon D700 Nikon D800 Sony a7 Nikon AF Nikkor 50mm f/1.8D +6 more
OP Horshack Veteran Member • Posts: 8,544
Download links for last two raws
OP Horshack Veteran Member • Posts: 8,544
More precision regarding clipping point

For the 100% crop compared (the "9 |||| 10" lettering), the average ADU of the whitespace in photo #3 is 2371, as measured in Rawanalyze. This means photo #2, which I list as the first saturation photo to use for DR, is actually 15% beyond clipping.
--
Kodak Instant Camera
Kyocera 1MP Camera phone (pre-paid phone plan)
http://horshack.smugmug.com/

OP Horshack Veteran Member • Posts: 8,544
Re: 10.5 - 11 stops on a luminance basis

Stujomo wrote:

I would redo the test but use a plain white towel with plenty of texture. At the moment you have very high contrast test chart which may or may not represent what you may have in the shadows of a reall world scene. Give it another go with a towel to compare the results and then try the test again with a small group of objects. I would be interesting to compare the results.

The only plain white towel I can think of with any contrast would be a dirty one

Ideally I'd like to get R/G/B copies of this ISO chart and add those to the test.

Stujomo
Stujomo Senior Member • Posts: 1,629
Re: 10.5 - 11 stops on a luminance basis

Thats the whole point of using the towel. Not everything you photograph will be a high contrast test chart. The lightest and darkest images which retain texture of the towel would give you a good indication of the useful dynamic range of the camera. You can of coure use highlight and shadow recovery techniques to extent the range. The texture of the towel will easily get lost in noise. Try it and see what the results are and see how much difference they would be from using a high contrast chart. It's worth doing just for the experince.

Horshack wrote:

Stujomo wrote:

I would redo the test but use a plain white towel with plenty of texture. At the moment you have very high contrast test chart which may or may not represent what you may have in the shadows of a reall world scene. Give it another go with a towel to compare the results and then try the test again with a small group of objects. I would be interesting to compare the results.

The only plain white towel I can think of with any contrast would be a dirty one

Ideally I'd like to get R/G/B copies of this ISO chart and add those to the test.

 Stujomo's gear list:Stujomo's gear list
Nikon D1H Nikon D700 Nikon D800 Sony a7 Nikon AF Nikkor 50mm f/1.8D +6 more
rhlpetrus Forum Pro • Posts: 26,131
Re: More precision regarding clipping point

Thanks, good job!

Horshack wrote:

For the 100% crop compared (the "9 |||| 10" lettering), the average ADU of the whitespace in photo #3 is 2371, as measured in Rawanalyze. This means photo #2, which I list as the first saturation photo to use for DR, is actually 15% beyond clipping.
--
Kodak Instant Camera
Kyocera 1MP Camera phone (pre-paid phone plan)
http://horshack.smugmug.com/

-- hide signature --

Renato.
http://www.flickr.com/photos/rhlpedrosa/
OnExposure member
http://www.onexposure.net/

Good shooting and good luck
(after Ed Murrow)

 rhlpetrus's gear list:rhlpetrus's gear list
Leica D-Lux (Typ 109) Nikon D7000 Nikon 1 V1 Nikon AF-S DX Nikkor 12-24mm f/4G ED-IF Nikon AF Nikkor 35mm f/2D +4 more
photoneptune Forum Member • Posts: 61
Re: 10.5 - 11 stops on a luminance basis

Hey, here's an idea:

Why don't you do it?

Stujomo wrote:

Thats the whole point of using the towel. Not everything you photograph will be a high contrast test chart. The lightest and darkest images which retain texture of the towel would give you a good indication of the useful dynamic range of the camera. You can of coure use highlight and shadow recovery techniques to extent the range. The texture of the towel will easily get lost in noise. Try it and see what the results are and see how much difference they would be from using a high contrast chart. It's worth doing just for the experince.

Horshack wrote:

Stujomo wrote:

I would redo the test but use a plain white towel with plenty of texture. At the moment you have very high contrast test chart which may or may not represent what you may have in the shadows of a reall world scene. Give it another go with a towel to compare the results and then try the test again with a small group of objects. I would be interesting to compare the results.

The only plain white towel I can think of with any contrast would be a dirty one

Ideally I'd like to get R/G/B copies of this ISO chart and add those to the test.

_sem_ Veteran Member • Posts: 5,033
Re: 10.5 - 11 stops on a luminance basis

Horshack wrote:

Ideally I'd like to get R/G/B copies of this ISO chart and add those to the test.

For a practical evaluation you may print a suitable test chart. Colour accuracy may not be perfect, but certainly good enough for relative comparison.
http://www.google.com/images?q=printer+test+chart

You may take an ordinary colour chart (in fact any scene with vivid colours would do). If you insist on resolution, you may take a BW resolution chart and paint it in PS.

Btw, two more practical DR tests (shadows lifting with harsh light) may be found here:
http://forums.dpreview.com/forums/readflat.asp?forum=1039&message=36621309
http://forums.dpreview.com/forums/readflat.asp?forum=1034&message=36110058

Stujomo
Stujomo Senior Member • Posts: 1,629
Re: 10.5 - 11 stops on a luminance basis

I've done it with both a D80 and a D1h and about 10 years ago with B&W films. Its interesting to do thats why I suggested it. I will post the extemes from the d80 RAW files.

The whites was left white and the shadows where left very dark as it would be in a real image.

photoneptune wrote:
Hey, here's an idea:

Why don't you do it?

Stujomo wrote:

Thats the whole point of using the towel. Not everything you photograph will be a high contrast test chart. The lightest and darkest images which retain texture of the towel would give you a good indication of the useful dynamic range of the camera. You can of coure use highlight and shadow recovery techniques to extent the range. The texture of the towel will easily get lost in noise. Try it and see what the results are and see how much difference they would be from using a high contrast chart. It's worth doing just for the experince.

Horshack wrote:

Stujomo wrote:

I would redo the test but use a plain white towel with plenty of texture. At the moment you have very high contrast test chart which may or may not represent what you may have in the shadows of a reall world scene. Give it another go with a towel to compare the results and then try the test again with a small group of objects. I would be interesting to compare the results.

The only plain white towel I can think of with any contrast would be a dirty one

Ideally I'd like to get R/G/B copies of this ISO chart and add those to the test.

 Stujomo's gear list:Stujomo's gear list
Nikon D1H Nikon D700 Nikon D800 Sony a7 Nikon AF Nikkor 50mm f/1.8D +6 more
Keyboard shortcuts:
FForum MMy threads