D800E -Truly Amazing Low-Light Performance

[NancyP]

Good grief. That's standard bamboo it is sitting on, right? Huuuuuge. Also very attractive.
And yes, I am duly impressed by the recovery of that sad histogram photo.

[bjanes]

However, SNR at 18% Grey diminishes more evenly as one reduces exposure. At base ISO it's 46dB. At ISO 6400 it's 28.1dB. 46-28.1=17.9dB. On the basis that 3dB is equivalent to a difference of one F stop in exposure, a drop in 6 stops of exposure should result in a drop of 18dB of SNR at 18%, which is approximately what one gets whether underexposing at base ISO or correctly exposing at ISO 6400.

These are the sorts of things one should easily be able to test for oneself using some practical real-world examples. I'll take a few underexposed shots tonight to see how they compare at ISO 100 and 6400.

Ray,

I think that the SNR relationship you document is due to photon counting statistics and has little to do with the sensor, provided that it has a linear response. AT 18% gray, the noise is almost all photon noise and the read noise contribution is minimal. Photon noise is present in the light before it hits the sensor. According to Poisson statistics, if one counts N photo-electrons, the noise is sqrt(N). The SNR is N/sqrt(N), which reduces to sqrt(N). With sensors, 6 dB equals one f/stop. Since dB is a log measure, one can determine the square root by dividing by 2. Increasing exposure by 1 f/stop (6 dB), will result in 3 dB increase in the SNR (and conversely). Perhaps one of the forum gearheads can comment further on this matter.

Regards,

Bill

[langier]

I've tried a few pix with my D800 at up to 25,000 and compared to the D700 at the same ISO, the D800 wins easily with no smear from the specular highlights that need lots of post. That said, I'm back from a month in Mallorca, Greece and Serbia shooting some stuff on the tripod and others using Hail Mary lighting and hand held.

My main body was the D800 with 24-120 f/4 and second the D700 with 17-35 f/2.8. Sometimes, I'd swap lenses when i knew I needed more res and a wider FOV.

I just looked at one of my Hail Mary photos shot in an 800-year old monastery with light streaming through the door and cranking up the D800 to just 6400. It was contrast city when I was shooting and I was blessed to be the fly on the wall. The image holds together well all the way from the sun-lit floor to the with aggressive NR in ACR (60 for luminance, 35 color), the image more than makes the grade!

For printing, I've got to reduce the size of the files which make the files even nicer.

In the overall scheme of things, once the ink hits the paper for me, I really can't tell the difference from one of my cameras when I'm printing smaller prints (less than 16x24), but looking at my edits from my time at this monastery, I almost wish I had a second D800 cranked up even a stop higher. I'm very sure that my images will work quite well at 20x30 and larger.

I tend to ignore the math and simply shoot the camera then deal with making an image. The math, though important to understanding the craft seldom comes across as an important element my final images. The final image is King in my book!

Attached is the final print (reduced) with a full-res crop of a section of the image from the same file.

As always, one's results may vary.

Bottom line for me is that the D800 makes the grade for high ISO shooting.

[Slobodan Blagojevic]

Larry, impressive captures!

On a side note, your title contains a combination of Latin (Svetog) and Cyrillic (Cabe) letters. Proper Latin spelling would be "Save."

[KevinA]

A bit confused by all the talk of the 800 high iso performance. I'm  gobsmacked by the recovery shown, I tried the same with my Canon X and got nowhere near that.
I also shot the D800 against the D4 and the X at high iso 6400, the D4 and X were much the same but the 800 was someway behind and struggled with the shadows. So as I say a bit confused at the 800 high iso experience I had. If I was shooting high iso the 800 is not the camera I would pick, so what happens between shooting it at 100iso and pushing in post between shooting at an original high iso. At the high iso I thought the DR was quite limiting, no denying the exceptional low iso DR performance. It must tail off pretty quickly somewhere up the iso scale?

Kevin.

[BernardLanguillier]

I just looked at one of my Hail Mary photos shot in an 800-year old monastery with light streaming through the door and cranking up the D800 to just 6400. It was contrast city when I was shooting and I was blessed to be the fly on the wall. The image holds together well all the way from the sun-lit floor to the with aggressive NR in ACR (60 for luminance, 35 color), the image more than makes the grade!

Very nice image!

Cheers,
Bernard

[Alistair]

A bit confused by all the talk of the 800 high iso performance. I'm  gobsmacked by the recovery shown, I tried the same with my Canon X and got nowhere near that.
I also shot the D800 against the D4 and the X at high iso 6400, the D4 and X were much the same but the 800 was someway behind and struggled with the shadows. So as I say a bit confused at the 800 high iso experience I had. If I was shooting high iso the 800 is not the camera I would pick, so what happens between shooting it at 100iso and pushing in post between shooting at an original high iso. At the high iso I thought the DR was quite limiting, no denying the exceptional low iso DR performance. It must tail off pretty quickly somewhere up the iso scale?

Kevin.

I agree. The 800 is very much a low ISO camera and the DR tails off quickly as the ISO is cranked up. A D4 or a 1dx/5d3 is a better High ISO camera I feel but I value the 800 for its low ISO supremacy.

[bjanes]

I agree. The 800 is very much a low ISO camera and the DR tails off quickly as the ISO is cranked up. A D4 or a 1dx/5d3 is a better High ISO camera I feel but I value the 800 for its low ISO supremacy.

Do you care to furnish some data to back up your assertions? The DXO data tell a different story. The SNR at 18% saturation is virtually the same for both cameras, which is not surprising since both cameras have the same sensor size and collect the same amount of light and have similar shot noise (which is the predominant noise contribution at this saturation level). For engineering DR using the print values to account for the differing pixel counts of the two cameras, the D4 has a slight advantage. Since the D800 has more pixels and each pixel contributes read noise it has slightly less engineering DR, since read noise is in the denominator for this determination. For photographic DR where read noise is less a factor, the values would be more comparable.

You might want to read this white paper by DXO.

[Ray]

I agree. The 800 is very much a low ISO camera and the DR tails off quickly as the ISO is cranked up. A D4 or a 1dx/5d3 is a better High ISO camera I feel but I value the 800 for its low ISO supremacy.

I began by expressing my surprise at how much detail existed in such a seriously underexposed shot at ISO 100. However, Bill made the point that the D800 is not quite an ISO-less camera to the extent that the D7000 is.

The DR comparisons at DXOMark show a slight flattening of the curve for the D800 between ISO 100 and 200, and the DR figure at ISO 6400 is almost a stop less than the 6 stops down it should be if the graph were completely linear. This result implies there should  be a slight DR advantage with the D800 when using a higher ISO setting rather than underexposing at base ISO, and some recent shots I took last night confirm this is indeed the case.

The noise or grain in the deepest shadows is noticeably coarser in the ISO 100 shots. I also notice a slight green cast in parts of the ISO 100 images, which would require more work to correct, but I haven't bothered. I suspect this green cast might be explained by the slight flattening of the Color Sensitivity graph as it approaches ISO 100. At high ISOs the difference in Color Sensitivity between one ISO and the next full stop of change is of the order of 1.5 bits, but between ISO 400 and 200 it's only 1.2 bits better, and between ISO 200 and 100, only 0.9 bits better.

The Nikon D4, according to the DXO comparisons graphs in attached image, does appear to have better DR at high ISO, particularly at ISO 1600 where it's 2/3rds of a stop better than the D800. However, the Canon 5D3 is hardly better than the D800 at any high ISO, to any degree that would be noticeable. The 1Dx matches the performance of the D4 at ISO 3200 and higher. However, at ISO 1600 it's no better than the D800, and at ISO 100, the DR of the 1Dx is a full 2 1/2 stops worse than the D800E.

[BernardLanguillier]

I agree. The 800 is very much a low ISO camera and the DR tails off quickly as the ISO is cranked up. A D4 or a 1dx/5d3 is a better High ISO camera I feel but I value the 800 for its low ISO supremacy.

The D800 is clearly superior at high ISO to the D3. The D3 is the camera that all sports photographers were calling incredibly (like in "I cannot believe my eyes") good at high ISO 4 years ago. The scenes we take pictures of have not changed in the mean time.

So I am not sure how the D800 can be called anything but an outstanding high ISO camera.

Cheers,
Bernard

[ErikKaffehr]

Hi,

I'm pretty sure Bill has a good point. Here is a plot tonal range plot of Nikon D800, Nikon D4 and Canon 1DX.

I guess that there are two points leading to the D800 called a low ISO camera:

1) If the image is viewed at actual pixels the D800 image is given a 40% disadvantage.

2) The Nikon at low ISO is very good but has more to loose with increasing ISO.

So I guess that D800 matches the "pro competition" up to 12000 ISO (or so) when correctly viewed.

Best regards
Erik

Do you care to furnish some data to back up your assertions? The DXO data tell a different story. The SNR at 18% saturation is virtually the same for both cameras, which is not surprising since both cameras have the same sensor size and collect the same amount of light and have similar shot noise (which is the predominant noise contribution at this saturation level). For engineering DR using the print values to account for the differing pixel counts of the two cameras, the D4 has a slight advantage. Since the D800 has more pixels and each pixel contributes read noise it has slightly less engineering DR, since read noise is in the denominator for this determination. For photographic DR where read noise is less a factor, the values would be more comparable.

You might want to read this white paper by DXO.

[thierrylegros396]

After reading such remarkable results, I decided to test my RX100.

Of course, results are not as good, but interresting when using LR4.2 for processing.

I've tried 0 to -7.3 fStop exposure in 1/3 sequence.

0 to -4 fStop give very good results up to A4, -5 Fstop is acceptable up to A6 (10x15), -6 is usable for web, and -7 as a memento

The biggest diference with your D800E is that colors are falling off pretty quickly after 5 fStop.

Interresting correlation with ISO results, ISO 80(100 & 125) to 800 deliver good results, 1600 OK for 10x15, 3200 usable for web, and 6400 as a memento.

So, now I know that may underexpose without too many drawbacks.

Have a Nice day.

Thierry

[Ray]

After reading such remarkable results, I decided to test my RX100.

Of course, results are not as good, but interresting when using LR4.2 for processing.

I've tried 0 to -7.3 fStop exposure in 1/3 sequence.

0 to -4 fStop give very good results up to A4, -5 Fstop is acceptable up to A6 (10x15), -6 is usable for web, and -7 as a memento

The biggest diference with your D800E is that colors are falling off pretty quickly after 5 fStop.

Interresting correlation with ISO results, ISO 80(100 & 125) to 800 deliver good results, 1600 OK for 10x15, 3200 usable for web, and 6400 as a memento.

So, now I know that may underexpose without too many drawbacks.

Have a Nice day.

Thierry

Hi Thierry,

That sure is an impressive pocket camera. I guess the reason I don't have one is due to the fact I would find the maximum wide angle of 28mm and the maximum telephoto of 100mm a bit limiting. It would be great if someone could design a camera like the RX-100 with a choice of two or more quickly-interchangeable zoom leness, say, from 16-48mm and 50-300mm, or if that's too difficult, provide different flavours of the same camera with different fixed-zoom lenses. The camera is so light one could easily carry two of them at the same time.

Checking the DXO graphs for this camera I'm amazed that the DR of the RX-100 is well over a stop better than that of my Canon 5D, at base ISO. ISO 400 seems to be the cross-over point where DR is equal for both cameras. At higher ISOs, the RX-100 gets left behind, which is not surprising considering that the sensor area is less than 1/7th that of the full-frame 5D.

SNR at 18% is another story though. As someone coming from the world of MFDB, I guess you wouldn't be too happy about that.

[ErikKaffehr]

Hi,

Having the RX100 I must say I'm quite impressed. Image quality at low ISO and at the wide end is pretty close to my full frame 24MP DSLRs. The long end is something else, it's still sharp in the middle bat sharpness drops rapidly to the corners, at least on mine.

It is also quite usable at high ISO.

Regarding SNR I think that DxO-s "tonal range" is probably the nicest and most telling. See the enclosed plot of RX100, D3200 (APS-C) and canon 5DII (full frame). Tonal range is about collecting photons and here bigger is mostly better.

I also added the tonality plot of the Canon 5DIII, Nikon D4 and the Phase One IQ180. Interestingly, the curves almost overlap! The Phase One  is capable of the highest "tonal range" but only at low ISO. What we see here is that the MFDB has advantage of sensor size but has much worse "quantum efficiency" compared to the Nikon D4 and the Canon 5DIII.

Quantum efficiency aside, I have a feeling that Canon, Nikon and also Sony have improved the full well capacity of the pixels, too.

Best regards
Erik

Hi Thierry,

That sure is an impressive pocket camera. I guess the reason I don't have one is due to the fact I would find the maximum wide angle of 28mm and the maximum telephoto of 100mm a bit limiting. It would be great if someone could design a camera like the RX-100 with a choice of two or more quickly-interchangeable zoom leness, say, from 16-48mm and 50-300mm, or if that's too difficult, provide different flavours of the same camera with different fixed-zoom lenses. The camera is so light one could easily carry two of them at the same time.

Checking the DXO graphs for this camera I'm amazed that the DR of the RX-100 is well over a stop better than that of my Canon 5D, at base ISO. ISO 400 seems to be the cross-over point where DR is equal for both cameras. At higher ISOs, the RX-100 gets left behind, which is not surprising considering that the sensor area is less than 1/7th that of the full-frame 5D.

SNR at 18% is another story though. As someone coming from the world of MFDB, I guess you wouldn't be too happy about that.

[hjulenissen]

D800 (and rx100) users: do you feel that 18% SNR figures are relevant to your photography?

-h

[BernardLanguillier]

D800 (and rx100) users: do you feel that 18% SNR figures are relevant to your photography?

In fact... yes.

As long as we use cameras, the technicalities thereof are relevant.

Most drivers don't feel that a torque curve/efficiency curves are important to their driving/wallet, but in fact... they are.

Cheers,
Bernard

[mac_paolo]

D800 (and rx100) users: do you feel that 18% SNR figures are relevant to your photography?

-h

I really do.

[Ray]

Hi,

Having the RX100 I must say I'm quite impressed. Image quality at low ISO and at the wide end is pretty close to my full frame 24MP DSLRs. The long end is something else, it's still sharp in the middle bat sharpness drops rapidly to the corners, at least on mine.

It is also quite usable at high ISO.

Regarding SNR I think that DxO-s "tonal range" is probably the nicest and most telling. See the enclosed plot of RX100, D3200 (APS-C) and canon 5DII (full frame). Tonal range is about collecting photons and here bigger is mostly better.

I also added the tonality plot of the Canon 5DIII, Nikon D4 and the Phase One IQ180. Interestingly, the curves almost overlap! The Phase One  is capable of the highest "tonal range" but only at low ISO. What we see here is that the MFDB has advantage of sensor size but has much worse "quantum efficiency" compared to the Nikon D4 and the Canon 5DIII.

Quantum efficiency aside, I have a feeling that Canon, Nikon and also Sony have improved the full well capacity of the pixels, too.

Best regards
Erik

Hi Erik,
I think at this stage we need to specify how differences in the measurements as displayed on the graphs are significant in practice. The units used vary, depending on the specification. With regard to DR we have units of EV (exposure value) which are equivalent to a one stop difference in exposure. I think DXOMark have stated somewhere that anything less than a 0.5EV difference in DR is of little consequence, and probably unnoticeable.

Likewise, with regard to Color Sensitivity, which is measured in bits, anything less than 1 bit is of little concern.

I wonder if the same applies to Tonal Range, which is also measured in bits. A difference in excess of one bit may be worth something. A difference of less than one bit can be ignored, for most practical purposes.

A difference of 3dB in SNR appears to be the threshold which is noticeable. Differences less than 3dB are of little consequence,

These are the standards I use. If camera A has a 1/2 stop better DR than camera B, whether at high ISO or low ISO, I'm not particularly fussed. A one-stop increase in DR seems to be worth something. A two-stop increase in DR is spectacular, just as a 6dB increase in SNR is, and maybe a 2 bit increase in tonal Range.

[Alistair]

Do you care to furnish some data to back up your assertions? The DXO data tell a different story. The SNR at 18% saturation is virtually the same for both cameras, which is not surprising since both cameras have the same sensor size and collect the same amount of light and have similar shot noise (which is the predominant noise contribution at this saturation level). For engineering DR using the print values to account for the differing pixel counts of the two cameras, the D4 has a slight advantage. Since the D800 has more pixels and each pixel contributes read noise it has slightly less engineering DR, since read noise is in the denominator for this determination. For photographic DR where read noise is less a factor, the values would be more comparable.

You might want to read this white paper by DXO.

NO, I have neither data nor graphs and I have no interest whatsoever in reading any whitepapers on the subject. It is merely my personal observation that the DR advantage of my D800e disappears quite quickly as ISO is increased. Once that advantage disappears it is my personal view that there are better handling and better featured cameras out there (e.g. D4, 5dIII, D1X, OMD, etc. etc.).

A quick look at the plethora of graphs that indignantly sprung up, yours included, at my first utterance of anything approaching criticism of the D800 seem to support this view (at least the first part of it), but what do I know.

[Ray]

NO, I have neither data nor graphs and I have no interest whatsoever in reading any whitepapers on the subject. It is merely my personal observation that the DR advantage of my D800e disappears quite quickly as ISO is increased. Once that advantage disappears it is my personal view that there are better handling and better featured cameras out there (e.g. D4, 5dIII, D1X, OMD, etc. etc.).

A quick look at the plethora of graphs that indignantly sprung up, yours included, at my first utterance of anything approaching criticism of the D800 seem to support this view (at least the first part of it), but what do I know.

It's now clear to me that the D800E is not an ISO-less camera, and that shots at ISO 6400, for example, compared with the same exposure at ISO 100,  have more detail in the deep shadows, and more accurate color rendition.

The 5D3 may have better specs in some respects, than the D800E, but DR and SNR are not amongst them.