Nikon D7000 Dynamic Range

[Ray]

We probably all know by now that the DR of the D7000 at base ISO (together with that of the Pentax K-5) trumps every other digital camera on the market, whatever the format.

The P65+ is a laggard and under-achiever compared to the tiny format D7000, as regards DR.

But before you go purple in the face with rising blood pressure, and release yourself into an explosion of expletives directed against DXO, let's consider for a moment what the DXOMark rankings really mean.

I think many of you may have bravely ventured beyond the single-figure rankings, and dared to examine the individual graphs comparing one sensor with another of your choice.

If you have, you will likely have noticed on the top left corner immediately above the graphs, there's a (screen-print) option.

When you toggle between these two options, you should notice a substantial change in the results, on the graphs.

With the option 'screen', you are comparing the qualities of one pixel with another (for example, a D7000 pixel with a P65+ pixel).

What the DXO results tell us, is that the P65+ pixel is pretty much the same as the D7000 pixel with regard to SNR at 18% grey, tonal range and color sensitivity, at base ISO. Above base ISO, all bets are off. We sink into the abyss and the P65+ flounders.

However, most images are composed of all, or most, of the pixels on the sensor, depending on the degree of cropping. If we want to know what the comparative image quality would be like, if we were to downsample the P65+ image to the same file size as the D7000, for example, then hit the 'print' option.

You should see a significant jump in the image quality parameters for the P65+. The D7000 still retains its DR supremacy, but takes a dive in all other qualities (SNR, tonal range, color sensitivity).

Now, at this point, I think some further explanation is required about the 'normalisation' size in 'print' mode. It's specified at only 8mp which represents a print size of 8x12 at 300 dpi.

That seems a bit on the small side. I don't imagine that P65+ owners will often make prints that small. Is that size relevant?

Well, yes it is. DXO have chosen that small size in order not to exclude too many camera models. Any image consisting of fewer than 8mp would have to be interpolated.

Here's what DXO have to say on the matter.

Original measurements are useful to help gauge the actual image quality when viewed at 100%, but they do not predict how printed images will compare. In order to give a better prediction of how prints compare, a normalized version is more reliable and is provided based on 8Mpix

And this is what they have to say about the 'single-figure' rankings.

Sensor Overall Score is normalized for a defined printing scenario—8Mpix printed on 8”x12” (20cmx30cm) at 300dpi resolution. Any other normalization, even with higher resolution, would lead to the same ranking, given that any camera that could not deliver the chosen resolution would be eliminated from the comparison.

In other words, whether a P65+ image is downsampled to the 8mp of the Canon 20D, the 24mp of the D3X, or the 16mp of the D7000, the ranking is unchanged.

So we should bear in mind that DXO do not offer rankings for small format cameras uprezzed to the higher pixel-count size of larger formats. This seems to me to be at the hub of the confusion.

I think perhaps DXOMark should place a big sign above their single-figure rankings, such as:

These rankings apply only to images that have been downsized to the native print size at 300 dpi, whatever that may be according the pixel count of the smallest camera in the comparison.

Okay! Now to the issue at hand. It seems that some cameras are extremely good at high-ISO performance, but average at low-ISO performance, such as the Nikon D3s.

The D7000 seems to be very good at low-ISO performance, DR-wise, but pretty average at high-ISO. It seems we can't have everything, although we may want everything.

With regard to every parameter that DXO test, except DR, the P65+ is streets ahead. No question.

So what is the significance and the benefit of such high DR, at base ISO, I ask myself? How does it translate to improved image quality?

I found it surprisingly difficult to find any comparisons on the internet that address this issue. But after some searching I did come across a few images that might provide some indication of this extraordinary DR of the D7000.

I'm assuming here that for editorial purposes I am not infringing any copyright. If I am, Michael will probably send me a PM.

The images are from http://robertbromfield.com/nikon-d7000-review-and-impressions/

The image I've selected is a shot at F8 with the Nikkor 14-24/2.8 at 14mm, which becomes 21mm on the DX format.

Robert complains that the image seems less sharp than the same lens on the D700. Does he not know that the 14-24/2.8, at 14mm, is sharper at F2.8 than at F8, as well as being sharper at F4 and F5.6, than at F8?

To compare the D7000 with the D700 using this lens, one should be comparing the D7000 at F8 with the D700 at F12.

Okay! To the nitty gritty. If the D7000 really does have such fantastic DR, then the shadows in Robert's night shot should be relatively clean.

We don't have comparisons with the P65+, or any other camera, so this exercise fails at that level. Can't be helped!

But at least I hope I have successfully defined the issues. The D7000 shot at night without flash, really does seem to show a remarkable amount of detail in the dark shadows, which is precisely what a high DR means.

Here are the images. First, the full jpeg unaltered; second the same image lightened in Photoshop using shadows/highlight tool; third a crop of the bottom left corner after lightening the shadows; fourth, a 100% crop showing the shadow cut-off point where all values are the same.

I'm not a particularly technical sort of guy, compared with others on this forum, such as BJL, but I've read that Nikon have a habit of clipping black levels at a certain point to obliterate objectionable banding and noise, unlike Canon who reveal such banding. Astronomers prefer Canon for this reason. Sometimes, very degraded detail is better than no detail, but probably not for most practicing photographers.

If the next iteration of the D7000 gives us 16 EV of DR, then those grey/black patches should reveal some detail. 

[stamper]

I am sitting wondering what all this guff has to do with photography? Ray do you ever spend time taking photographs?

[Ray]

I am sitting wondering what all this guff has to do with photography? Ray do you ever spend time taking photographs?

Of course I do, stamper. Since the advent of digital photography I've accummulated enough photos, mostly in RAW format, to keep me busy for 3 lifetimes, processing and reprocessing the images according to my changing taste and the progressive developments in software.

Quite a few of those images would have been of high contrast scenes in rainforests, for example, where a D7000 would have been ideal.

[Graystar]

Okay! Now to the issue at hand. It seems that some cameras are extremely good at high-ISO performance, but average at low-ISO performance, such as the Nikon D3s.

The D7000 seems to be very good at low-ISO performance, DR-wise, but pretty average at high-ISO. It seems we can't have everything, although we may want everything.

Low ISO DR performance depends on how much light the pixels can suck up (called saturation.)  High ISO performance depends on sensor area and control of read noise.  Cameras of the last couple years have very low read noise.  So it makes sense that sensors of the same size have similar performance at high ISO.  I’ve compared ISO 3200 images from the D90 and D7000 and they pretty much look the same.  The D90 actually has lower read noise than the D7000 at ISOs 1600-6400, which is why its DR matches or surpasses the D7000 at those sensitivities.

So what is the significance and the benefit of such high DR, at base ISO, I ask myself? How does it translate to improved image quality?

I found it surprisingly difficult to find any comparisons on the internet that address this issue. But after some searching I did come across a few images that might provide some indication of this extraordinary DR of the D7000.

I’ll bet that you don’t own a recent Nikon.  From a dark area such as your crop, a Nikon owner would expect to be able to bring out that much detail.  The D7000’s DR boost comes from having a higher pixel saturation point, as well as lower read noise at low ISO.  To see the expanded DR in action you need a scene with a large DR, such as an indoor shot of a window on a sunny day.

[Ray]

Low ISO DR performance depends on how much light the pixels can suck up (called saturation.)  High ISO performance depends on sensor area and control of read noise.  Cameras of the last couple years have very low read noise.  So it makes sense that sensors of the same size have similar performance at high ISO.  I’ve compared ISO 3200 images from the D90 and D7000 and they pretty much look the same.  The D90 actually has lower read noise than the D7000 at ISOs 1600-6400, which is why its DR matches or surpasses the D7000 at those sensitivities.

I’ll bet that you don’t own a recent Nikon.  From a dark area such as your crop, a Nikon owner would expect to be able to bring out that much detail.  The D7000’s DR boost comes from having a higher pixel saturation point, as well as lower read noise at low ISO.  To see the expanded DR in action you need a scene with a large DR, such as an indoor shot of a window on a sunny day.

I have a D700. Does that count as a recent Nikon? How much did you bet?

Whatever the technical reasons that might explain why a pixel on one particular sensor can 'suck up' more light (as you put it) than a larger pixel on another sensor, the fact remains, according to DXO, that the smaller D7000 pixel appears to have a higher DR than the larger D700 pixel, at base ISO. The fact that it doesn't have a higher DR than the D700 at high ISOs would seem to me to be a matter of the sophistication and power of in-camera processors and analog-gain transistors.

As I understand, a digital camera has only one 'real' ISO, known as the base ISO. All higher ISOs than base are, in reality, underexposures that are subject to analog gain prior to A/D conversion, and further processing along the chain before the data is written to the memory card.

Now I agree that the scene I've shown may not necessarily be ideal for the purposes of demonstrating the DR potential of the D7000, but it was the best I could get after trawling the internet. A RAW image would be ideal.

If you have a better image demonstrating the DR capability of the D7000, please show it.

[Graystar]

I have a D700. Does that count as a recent Nikon?

Sure.  Your D700 will pull up detail like that better than the D7000. 

Whatever the technical reasons that might explain why a pixel on one particular sensor can 'suck up' more light (as you put it) than a larger pixel on another sensor, the fact remains, according to DXO, that the smaller D7000 pixel appears to have a higher DR than the larger D700 pixel, at base ISO. The fact that it doesn't have a higher DR than the D700 at high ISOs would seem to me to be a matter of the sophistication and power of in-camera processors and analog-gain transistors.

It has to do with read noise.  At ISO 200 the read noise on the D700 is high.  At higher ISO levels the read noise drops by more than half.  The D7000 has a read noise level that starts low and stays low.  And it's more efficient as well.

The D7000 is a great camera with a very efficient sensor.  I’d still rather have a D700...though.

If you have a better image demonstrating the DR capability of the D7000, please show it.

I haven’t look extensively for one...I’m sure there must be some better images somewhere, but I found this over at DPReview...
http://forums.dpreview.com/forums/read.asp?forum=1034&message=36843187

[RFPhotography]

Bloody hell.  My head hurts. 

[Ray]

Sure.  Your D700 will pull up detail like that better than the D7000.  

It has to do with read noise.  At ISO 200 the read noise on the D700 is high.  At higher ISO levels the read noise drops by more than half.  The D7000 has a read noise level that starts low and stays low.  And it's more efficient as well.

The D7000 is a great camera with a very efficient sensor.  I’d still rather have a D700...though.

I haven’t look extensively for one...I’m sure there must be some better images somewhere, but I found this over at DPReview...
http://forums.dpreview.com/forums/read.asp?forum=1034&message=36843187

Of course it has to do with read noise, as well as may other factors, but that doesn't explain why one pixel has significantly lower read noise, as a proportion of total signal, than another pixel, like a D700 pixel which is larger. It's not the absolute size of the read noise that's important, but it's relative size compared with the strength of the signal.

The landscape shot at dpreview is a nice shot, clear and sharp, but it doesn't demonstrate the limits of the D7000 dynamic range. In order to do that you need a subject with a very high brightness range which not only includes bright clouds but dense shadows in the undergrowth, not just normal shade. I get a sense that any camera could have taken that shot at dpreview, including the first DSLR I ever bought, the 6mp Canon D60 about 7 years ago.

You know, were not talking about a miniscule, pixel-peeping DR difference here, at base ISO, like 1/4th or 1/3rd of a stop, but a whacking 1 & 1/2 stops.

The D700 has a lower pixel count than the D7000, so there's no downsizing advantage for the D700 regarding noise and DR, but there is for the D7000.

At normalised print sizes, the largest size you would make a D700 print without interpolation, the D7000 has a 1 & 2/3rd stop DR advantage. That really is significant. If I had a D7000, I'd demonstrate it, or prove DXO wrong.

[ErikKaffehr]

Ray,

Just for your information, Phase One has a couple of comparable images for download, which I may consider good for comparison. The download link is here:

http://www.phaseone.com/en/Downloads/Sample-images.aspx


These images are in some Capture One format, so you need Capture One to read them. I downloaded Capture One demo, and exported the files as DNGs so I could process them in Lightroom. Phase One had three images, Architecture/Landscape low ISO, outdoor portrait and high ISO indoor portrait. In my view the outdoor portrait comparison was essentially worthless, but the other two pretty good. The comparison on the Architecture/Landscape image was shot on Canon 1DIIIs the others on Nikon D3.

That P65+ architecture image was incredibly good, having much better DR than the Canon image, AT THE PIXEL LEVEL.

Preprocessing the images in Capture One may give the P65+ an unfair advantage, but may have been necessary.

My article on these images is here: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/41-phase-one-images-for-download

I didn't show the images, because I did not ask Phase One to permit publication of their images.

Best regards
Erik

[ErikKaffehr]

Hi,

A generic comment on DR. I really think that the discussion around DR is overblown. Much of the discussion at all is overblown.

My experience is that DR is essentially plentiful when shooting with present days DSLRs at base ISO correctly exposing to the right. In real life we always have lens flare that reduces the achievable DR anyway. If we shoot in a dark cathedral and want to reproduce shadow detail and window mosaics at the same time we may need to use HDR (combining a bracketed exposure in a single image) or similar techniques. But, in my experience a correctly HTTR exposed image is as good as HDR in most cases.

One issue that is overlooked is MTF. MTF measures how much contrast an optical system can transfer for a certain feature size. The feature size is described as cycles, lp/mm, frequency etc. Anyway, MTF diminishes as feature size is reduced.  In an MTF versus frequency plot the MTF value drops almost exactly linearly for a diffraction limited (that is perfect) lens at a given aperture.

Therefore, doubling image size would also double the contrast transferred for a certain size of feature. A larger format would reproduce an object larger on sensor would the angle of view kept constant. So significantly more detail would be visible in the darks, simply because shadow detail would have more contrast.

The reasoning above is of course based on the assumption that MTF vs. resolution is similar on both systems. This may or may no be the case. Ray used to argue that larger formats need to be stopped down more than smaller format to keep DOF constant. This is indeed true, but maximum DOF is not always desirable. MF lenses used to have lower MTF at given lp/mm than 135, but we have a new generation of MF lenses like some of the Phase One lenses and the new Leica lenses for the S2 system that are brilliant performers, and so are the new Schneider and Rodenstock digital lenses. IF YOU GET A DECENT COPY!

http://www.josephholmes.com/news-medformatprecision.html (why not read this?! ;-)

My opinion is that we don't need to argue about things that we don't have tested. I would suggest that we learn to use what we have optimally. It may be that we perceive a need for better quality, and MF-digital may the way to achieve it. That's fine.

But while discuss the fine point before setting up the camera on a decent tripod, use mirror lockup and a cable release (or self timer) or testing and adjusting auto focus?

Best regards
Erik

[Ray]

Hi,

A generic comment on DR. I really think that the discussion around DR is overblown. Much of the discussion at all is overblown.

My experience is that DR is essentially plentiful when shooting with present days DSLRs at base ISO correctly exposing to the right. In real life we always have lens flare that reduces the achievable DR anyway. If we shoot in a dark cathedral and want to reproduce shadow detail and window mosaics at the same time we may need to use HDR (combining a bracketed exposure in a single image) or similar techniques. But, in my experience a correctly HTTR exposed image is as good as HDR in most cases.

One issue that is overlooked is MTF. MTF measures how much contrast an optical system can transfer for a certain feature size. The feature size is described as cycles, lp/mm, frequency etc. Anyway, MTF diminishes as feature size is reduced.  In an MTF versus frequency plot the MTF value drops almost exactly linearly for a diffraction limited (that is perfect) lens at a given aperture.

Therefore, doubling image size would also double the contrast transferred for a certain size of feature. A larger format would reproduce an object larger on sensor would the angle of view kept constant. So significantly more detail would be visible in the darks, simply because shadow detail would have more contrast.

The reasoning above is of course based on the assumption that MTF vs. resolution is similar on both systems. This may or may no be the case. Ray used to argue that larger formats need to be stopped down more than smaller format to keep DOF constant. This is indeed true, but maximum DOF is not always desirable. MF lenses used to have lower MTF at given lp/mm than 135, but we have a new generation of MF lenses like some of the Phase One lenses and the new Leica lenses for the S2 system that are brilliant performers, and so are the new Schneider and Rodenstock digital lenses. IF YOU GET A DECENT COPY!

http://www.josephholmes.com/news-medformatprecision.html (why not read this?! ;-)

My opinion is that we don't need to argue about things that we don't have tested. I would suggest that we learn to use what we have optimally. It may be that we perceive a need for better quality, and MF-digital may the way to achieve it. That's fine.

But while discuss the fine point before setting up the camera on a decent tripod, use mirror lockup and a cable release (or self timer) or testing and adjusting auto focus?

Best regards
Erik

Thanks Erik. I did see that link to the Phase comparison images before, and attempted to download them and have a look. But something went wrong, or I did something wrong and my trial version of C1 didn't seem to recognise the files. It became un unnecessary hassle, so I gave up.

The fact is, I don't need to be convinced that P65+ images, at base ISO, are better in all respects than Canon 1Ds3 images, including DR. The DXO results indicate that this is so. I've no reason to disbelieve them.

What amazes me is how Nikon have overtaken Canon with regard to these image quality parameters, particularly DR.

If you want the best low light performance, get the D3s. If you want the best DR performance at base ISO get the D3X or the D7000.

It's truly remarkable that this cropped-format D7000 has the DR performance of the D3x all the way from ISO 100 to ISO 6400. At every ISO they are about the same. The differences are negligible.

However, they are not the same for all of the other parameters, such as SNR and color sensitivity. The D3X is slightly better than the D7000 at each ISO setting, although one might wonder how significant such differences are in practice. For example, DXOMark comment that a difference of one bit in color sensitivity is barely noticeable. The differences in color sensitivity between the D3X and D7000, at all ISOs, is in the order of 1.1 to 1.3 bits, according to the graphs.

Of course, the differences in total sensor resolution are significant. An increase of 50% in pixel count is not to be sneezed at, and as you've pointed out before, the larger format can often access the same image detail from the lens at a higher MTF and a lower lpp/mm.

For those who are always searching for the shallowest of DoF, the larger format has an advatage. A lens at F2.8 will tend to produce sharper images on FX than the same quality of lens at F2.8 on the DX format, as well as producing a shallower DoF.

If the DoF is equalized, FX format at F4.5 (for example) will also usually be sharper than DX at F2.8, but may not be sharper at F9 than DX at F5.6.

What might be interesting to consider is what happens if we need to equalize both DoF and shutter speed (because the subject is moving). We might then be comparing say a D700 at F9 and ISO 400, with a D7000 at F5.6 and ISO 160.

In this situation, the DXO graphs are telling me that the D7000 should produce approximately a full stop better DR than the D700 at all ISOs that differ by this 'DoF/ISO/shutter speed' factor of approximately 1 & 1/3rd stops.

Furthermore, the FX D700 in this situation, loses all advantages with regard to SNR, tonal range and color sensitivity. It's slightly worse in fact, but let's not quibble.

Do you agree with this analysis?

Cheers!

[stamper]

Quote

Furthermore, the FX D700 in this situation, loses all advantages with regard to SNR, tonal range and color sensitivity. It's slightly worse in fact, but let's not quibble.

Unquote

Ray that is why you started the thread ..... to quibble! You could start a fight in an empty house. Bet you are happy to have two more to shadow box with?

[ErikKaffehr]

Hi,

I have both DX and FX, Sony Alpha 700 and Sony Alpha 900, in my case. The feeling I have is that I have a tendency to take advantage of what I have. The Alpha 700 with it's 16-80/3.5-4.5 lens is more nimble and definitively good enough, most of the time. With the Alpha 900 I want to go for perfection. It's like shooting MF film and 135 Film. When shooting MF I used to adjust to MF, very little DOF, camera on tripod and so on. If I was shooting MF the 135 way the pictures were like 135.

A2 prints from the Alpha 700 can be impressive. I got the response at the photo club: Ah, one of your stitched full frames?! No, just a cropped APS-C!

The technique we have now is just incredibly good!

Best regards
Erik

Thanks Erik. I did see that link to the Phase comparison images before, and attempted to download them and have a look. But something went wrong, or I did something wrong and my trial version of C1 didn't seem to recognise the files. It became un unnecessary hassle, so I gave up.

The fact is, I don't need to be convinced that P65+ images, at base ISO, are better in all respects than Canon 1Ds3 images, including DR. The DXO results indicate that this is so. I've no reason to disbelieve them.

What amazes me is how Nikon have overtaken Canon with regard to these image quality parameters, particularly DR.

If you want the best low light performance, get the D3s. If you want the best DR performance at base ISO get the D3X or the D7000.

It's truly remarkable that this cropped-format D7000 has the DR performance of the D3x all the way from ISO 100 to ISO 6400. At every ISO they are about the same. The differences are negligible.

However, they are not the same for all of the other parameters, such as SNR and color sensitivity. The D3X is slightly better than the D7000 at each ISO setting, although one might wonder how significant such differences are in practice. For example, DXOMark comment that a difference of one bit in color sensitivity is barely noticeable. The differences in color sensitivity between the D3X and D7000, at all ISOs, is in the order of 1.1 to 1.3 bits, according to the graphs.

Of course, the differences in total sensor resolution are significant. An increase of 50% in pixel count is not to be sneezed at, and as you've pointed out before, the larger format can often access the same image detail from the lens at a higher MTF and a lower lpp/mm.

For those who are always searching for the shallowest of DoF, the larger format has an advatage. A lens at F2.8 will tend to produce sharper images on FX than the same quality of lens at F2.8 on the DX format, as well as producing a shallower DoF.

If the DoF is equalized, FX format at F4.5 (for example) will also usually be sharper than DX at F2.8, but may not be sharper at F9 than DX at F5.6.

What might be interesting to consider is what happens if we need to equalize both DoF and shutter speed (because the subject is moving). We might then be comparing say a D700 at F9 and ISO 400, with a D7000 at F5.6 and ISO 160.

In this situation, the DXO graphs are telling me that the D7000 should produce approximately a full stop better DR than the D700 at all ISOs that differ by this 'DoF/ISO/shutter speed' factor of approximately 1 & 1/3rd stops.

Furthermore, the FX D700 in this situation, loses all advantages with regard to SNR, tonal range and color sensitivity. It's slightly worse in fact, but let's not quibble.

Do you agree with this analysis?

Cheers!

[Ray]

Hi,

I have both DX and FX, Sony Alpha 700 and Sony Alpha 900, in my case. The feeling I have is that I have a tendency to take advantage of what I have. The Alpha 700 with it's 16-80/3.5-4.5 lens is more nimble and definitively good enough, most of the time. With the Alpha 900 I want to go for perfection. It's like shooting MF film and 135 Film. When shooting MF I used to adjust to MF, very little DOF, camera on tripod and so on. If I was shooting MF the 135 way the pictures were like 135.

A2 prints from the Alpha 700 can be impressive. I got the response at the photo club: Ah, one of your stitched full frames?! No, just a cropped APS-C!

The technique we have now is just incredibly good!

Best regards
Erik

Erik,
I know what you mean. When I got my first full frame DSLR, the Canon 5D about 5 years ago, I was very pleased with that extra image quality that resulted from a better SNR, smoother tonal range and better color sensitivity. It was certainly very apparent when processing images and examining detail at 100% on the monitor, but I can't say that I spent much time comparing large prints of identical scenes taken with the 5D and other cropped formats, even when I later acquired a APS-C format with higher pixel count than the 5D, the 15mp 50D.

But I do remember quite vividly, when I took the first shots with my 5D, being very disturbed by the noise and banding in the deep shadows when, for example, shooting the scene through the window of my living room, exposing for the bright clouds outside.

In fact, I immediately returned the camera and tried another copy from a different series of serial numbers, implying the camera was from a different batch.

It was slightly better regarding banding, so I kept it. Now I understand if a professional photographer were assigned the task of photographing someone's living room whilst simultaneously showing off the view through the window, he might bring in some heavy lighting equipment so the interior were correctly exposed without the black shadows one tends to get when using a single flash source. Or he might even go to the the trouble of sticking a gel over the windows to act like a neutral density filter, if he were paid enough.

But I'm just an amateur, and I prefer the easier option if one is available. If I find myself in the position of wanting to photograph the interior of the humble abode of a Nepalese farmer and his family, at an altitude of 3,000 metres against a magnificant backdrop of the Himalayas, I'd like to prepare myself with a camera capable of the job.

In such circumstances, the 5D without appropriate internal lighting, would display severe and unacceptable shadow noise. The D7000 probably wouldn't.

Checking the DXOMark graphs again, I see that at ISO 100, the D7000 has 2 & 2/3rds stops greater DR than the 5D (or 2.74 stops to be precise).  Even Stamper would agree that's one enormous nit.

To put this in perspective, the Canon 5D has only 1.1 stops higher DR, at base ISO, than the P&S Canon G10. Does that seem incredible? May I suggest not, if you consider the other parameters, such as tonal range and color sensitivity. There, the Canon 5D is streets ahead of the G10, as well as regards SNR at 18% grey.

But it's not ahead of the D7000. SNR, tonal range, color sensitivity on the D7000 are at least as good as those of the 5D, sometimes slightly better, sometimes slightly worse depending on ISO.

Since the D7000 has a higher pixel count than the 5D, I would not expect any resolution advantage from the 5D, except at very wide apertures.

[RFPhotography]

HEADLINE:  Technology improves!

Shock of shocks.  Now go out and take some pictures.

[bjanes]

We probably all know by now that the DR of the D7000 at base ISO (together with that of the Pentax K-5) trumps every other digital camera on the market, whatever the format.

The P65+ is a laggard and under-achiever compared to the tiny format D7000, as regards DR.

But before you go purple in the face with rising blood pressure, and release yourself into an explosion of expletives directed against DXO, let's consider for a moment what the DXOMark rankings really mean.

The best discussion of the DXO DR ratings that I know of is the thread started by Emil Martinec. The DXO DR is closely related to the engineering definition of DR. DXO DR uses a SNR of 1 in the denominator, whereas the engineering definition uses the read noise in the denominator, which is essentially the noise at zero exposure (for cameras like the Canons which use an exposure offset and do not clip the signal).

The D7000 has a very high DR according to the DXO criterion, because it has low read noise. However, noise over most of the useful photographic range is determined by shot noise, and larger sensor sizes will have a better SNR over this useful range. A SNR of 1 is not acceptable for most photographic purposes. Full frame 35 mm has double the sensor size of the cropped frame D7000 and the P65+ has twice the sensor area over a full frame 35 mm dSLR.

Regards,

Bill

[Ray]

The best discussion of the DXO DR ratings that I know of is the thread started by Emil Martinec. The DXO DR is closely related to the engineering definition of DR. DXO DR uses a SNR of 1 in the denominator, whereas the engineering definition uses the read noise in the denominator, which is essentially the noise at zero exposure (for cameras like the Canons which use an exposure offset and do not clip the signal).

The D7000 has a very high DR according to the DXO criterion, because it has low read noise. However, noise over most of the useful photographic range is determined by shot noise, and larger sensor sizes will have a better SNR over this useful range. A SNR of 1 is not acceptable for most photographic purposes. Full frame 35 mm has double the sensor size of the cropped frame D7000 and the P65+ has twice the sensor area over a full frame 35 mm dSLR.

Regards,

Bill

It looks as though I read that thread, Bill, because I contributed a post on page 3, but thanks for the link.

The issue of whether the full range of DR as described by DXO can be used for acceptable photographic results seems a separate issue to me. I've always been willing to accept that the photographic detail in those last couple of stops, detail that one might try to bring out by raising the shadows in an image of a scene with a high brightness range, is likely to be of unacceptable quality for artistic purposes, but possibly still useful for certain other purposes, such as identifying a car number plate in a photo taken at night.

For me, the issue is one of relativity. If the D7000 is described as having 2.74 stops higher DR than the 5D (Mk1), does that mean that at base ISO I could expect similar quality in the shadows from a D7000 image which had been underexposed 2 & 2/3rds stops, to what I would get from an ETTR exposure on the 5D, in the same shadows?

I understand the principle that a sensor which is double the area of another sensor will be exposed to double the amount of light when taking the the same scene at the same T-stop and shutter speed. But what happens to that light after it has reached the sensor is another matter. Some sensors have a higher quantum efficiency, for example, or more sophisticated methods of processing the analog signal, or voltage.

If there's a case to be made that photon shot noise, and/or normal lens flare will nullify those extra stops of DR that DXOMark claim the D7000 has, then I'd like to see the evidence.

I don't want to buy a D7000 largely on the basis it has terrific DR, only to find out later that in practice it has no greater DR than the Canon 60D as a result of factors which DXOMark did not take into consideration.

So far, the results of every test on my own equipment I've made, corresponds very closely with DXOMark's results.

I'm reminded of a test I carried out in Chiang Mai a few years ago. I rented an apartment for a couple of months and had free (or cheap) broadband access. The subjective nature of DR came up on an LL thread, and Jonathan Wienke devised a DR test chart containing various sizes of letters and numbers which, as a result of some discussion amongst us, would distinguish between acceptable quality of DR and the absolute limits, limits which would presumably be similar to the engineering definition of DR.

I downloaded Jonathan's image, printed it out on my base-model Epson printer (cheap as chips in Thailand), and proceeded to photograph the chart in constant lighting, starting with an ETTR, then reducing exposure by one stop with each shot.

I took more than 12 shots. On examining the images, I found there was still some detail to be observed in the 11th shot. The largest letters or numerals were still legible, but only just.

Of course, an argument then ensued, mainly between Jonathan, myself and John Sheehey, as I recall, as to whether or not that barely legible detail in the 11th stop deserved to be part of the specified dynamic range of the 5D.

I'm now rather surprised to see in the DXOMark graphs that the Canon 5D has a DR at base ISO of 11.13 EV. Close enough!

Cheers!

[RFPhotography]

OK, let me see if I can get this straight.

You're in a terrific setting like Thailand and you're wasting time farting around with dynamic range tests and screwing around on the internet?   

This is exactly why the LL fora are known as the place for the measurebators to hang out.   

[kers]

It is interesting to see how digital photography gets better and better. I just used a Nikon d3s and was amazed how much colour it sees in almost total darkness using a 1.4 lens wide open.
My eyes are less sensitive! and see for sure less colour.
( and i am not blind at all)

[Ray]

OK, let me see if I can get this straight.

You're in a terrific setting like Thailand and you're wasting time farting around with dynamic range tests and screwing around on the internet?   

This is exactly why the LL fora are known as the place for the measurebators to hang out.   

Not only farting around with dynamic range tests, but farting around reading books, and farting around photographing temples at night with only street lighting for illumination, and being concerned about unacceptable noise in the shadows; and farting around getting to know the limitations of my equipment, sometimes processing images in the evening on my laptop to see how the day's shooting went and to see what mistakes I made, in the hope of not repeating them.

A holiday which allows for no time to fart around is no holiday in my opinion. My last trip was an organised cruise along rivers in Europe and Russia. Everyday there were organised tours from morning till dusk, then yet more tours in the evenings to see shows, Russian folk dancing, ballets, concerts etc. By the time I got back to Australia, I felt so exhausted I needed a holiday.

Get my point?