Nikon D3 - Official image samples online

[NikosR]

http://nikonimaging.com/global/products/di...r/d3/sample.htm

Looks good.

[neverfinder]

http://nikonimaging.com/global/products/di...r/d3/sample.htm

Looks good.
[a href=\"index.php?act=findpost&pid=138850\"][{POST_SNAPBACK}][/a]

Wow have you seen the noise at 6400 bam thats a slap in the face for canon.

[BJL]

Wow have you seen the noise at 6400 bam thats a slap in the face for canon.
[a href=\"index.php?act=findpost&pid=138916\"][{POST_SNAPBACK}][/a]

The D3 at 3200 and 6400  does seem if anything even better than the 5D, with only slightly larger photosites. Well, two year's progress and all that.

Rather than being a slap for Canon, I would say that this is simply good news for everyone who cares about extreme low light shooting: it is clear now that there is no monopoly on the ability to reduce noise levels.

In fact, potentially three or four DSLR sensor makers are making big improvement in low light performance, perhaps now getting fairly close to the fundamental physical limits of photon shot noise. Canon as usual, Nikon for the D3 sensor, Sony for the D300/A700 sensor (guessing from it having half the maximum stated ISO limit of the D3 [3200 vs 6400] with half the sensor and photosite area), and maybe even Panasonic, which has significantly reduced high ISO noise in 4/3" sensors at the same time as increasing pixel count and reducing pixel spacing to 4.75 microns.

[Morgan_Moore]

the 6400 shot would be exposable at a 20th at f2 with 400 (being 160th at 5.6)

(yes that would be hard to get sharp hand held)

But reveals that the situation is not actually that dark

It would be interesting to see a sample that really pushes the sensor

I bet the quality drops off exponentially with diminishing light

Yes it will be useful for bands etc but I am not sure it will revolutionise what is possible

most cameras given a bit of light to chew on look OK even at high ISO

Can it photograph a black cat in a coal hole - I doubt it

Yes I have ordered one - my first proper upgrade since the SLRn

S

[BJL]

the 6400 shot would be exposable at a 20th at f2 with 400 (being 160th at 5.6)

(yes that would be hard to get sharp hand held)

But reveals that the situation is not actually that dark
[a href=\"index.php?act=findpost&pid=138964\"][{POST_SNAPBACK}][/a]

That doesn't matter much: "reciprocity" applies here. The sensor received very little light, as little as if it has been ISO 6400 at 1/20 and f/2 with four EV less illumination, and it produced a quite sharp, low noise image. The only difference in a darker environment would be if a far longer exposure produced additional dark current noise, and dark current seems to be totally insignificant at "hand-holdable" shutter speeds.

In general, and particularly with the D3, the main role for very high exposure index (ISO) seems to be getting shutter speeds high enough to freeze action, and maybe to freeze camera motion, as you mention, realms where dark current noise irrelevant and "reciprocity" reigns.

I bet the quality drops off exponentially with diminishing light
[a href=\"index.php?act=findpost&pid=138964\"][{POST_SNAPBACK}][/a]

With "on meter" exposures, all EI 6400 exposures receive a roughly equal amount of light.

[NikosR]

the 6400 shot would be exposable at a 20th at f2 with 400 (being 160th at 5.6)

(yes that would be hard to get sharp hand held)

But reveals that the situation is not actually that dark

It would be interesting to see a sample that really pushes the sensor

I bet the quality drops off exponentially with diminishing light

Yes it will be useful for bands etc but I am not sure it will revolutionise what is possible

most cameras given a bit of light to chew on look OK even at high ISO

Can it photograph a black cat in a coal hole - I doubt it

Yes I have ordered one - my first proper upgrade since the SLRn

S
[a href=\"index.php?act=findpost&pid=138964\"][{POST_SNAPBACK}][/a]

Digital Era Myth

There seems to be a persistent myth circulating in internet forums, that to show off the so called 'High ISO performance' one needs to shoot under very dark conditions (i.e Low EV).

I believe this is not true and I can find no rational justification for it. The only thing that is required is for the subject to exhibt a good mix of shadows and highlights and the ISO setting of the sensor to be set at a high value.

The sensor does not 'know' if a picture was taken under light conditions but with such aperture and speed that allowed very little light to fall on it, or in darker conditions with larger aperture and/or slower speed which allowed the same amount of light to reach the sensor.

The colour temperature of the ambient light and the white balance compensation applied also play a significant role.

Also, I'm not sure how a camera 'chews on' light

[markhout]

I appreciate the progress made, but I'm still struggling to understand if and how this 'high ISO performance' would directly or indirectly benefit my landscape images (taken at 100 or perhaps 200 ISO).

Would it help with the dynamic range ('the mix of shadows and highlights' referred to in the above posting)?

Thanks.

[John Sheehy]

Wow have you seen the noise at 6400 bam thats a slap in the face for canon.
[a href=\"index.php?act=findpost&pid=138916\"][{POST_SNAPBACK}][/a]

Look again.  Lots of noise reduction - no color in the shadows, even at IS0 200.

The images suggest lower-than-average shot noise, perhaps, but read noise doesn't look like anything revolutionary.  You really need RAW files (and test files, not real photographs) to tell what has really been done, hardware-wise, in the D3.

[John Sheehy]

I appreciate the progress made, but I'm still struggling to understand if and how this 'high ISO performance' would directly or indirectly benefit my landscape images (taken at 100 or perhaps 200 ISO).

Would it help with the dynamic range ('the mix of shadows and highlights' referred to in the above posting)?
[a href=\"index.php?act=findpost&pid=138987\"][{POST_SNAPBACK}][/a]

You're not going to tell from these JPEGs.  They have destroyed the deep shadow areas with blurring and desaturation.

[BernardLanguillier]

Digital Era Myth

There seems to be a persistent myth circulating in internet forums, that to show off the so called 'High ISO performance' one needs to shoot under very dark conditions (i.e Low EV).

I believe this is not true and I can find no rational justification for it. The only thing that is required is for the subject to exhibt a good mix of shadows and highlights and the ISO setting of the sensor to be set at a high value.
[a href=\"index.php?act=findpost&pid=138978\"][{POST_SNAPBACK}][/a]

True, but very dark situations often have large black areas where the noise shows a lot, as well as some tonal transitions from dark to mid-dark where the noise shows a lot.

Regards,
Bernard

[Morgan_Moore]

The sensor does not 'know' if a picture was taken under light conditions but with such aperture and speed that allowed very little light to fall on it, or in darker conditions with larger aperture and/or slower speed which allowed the same amount of light to reach the sensor.[a href=\"index.php?act=findpost&pid=138978\"][{POST_SNAPBACK}][/a]

That would appear to be a completely rational argument.

However I still adhere to the opinion that my digital cameras, used in good light with stupid fast shutter speeds anda high ISO setting exhibit less noise than when I use the same cameras in a situation that actually demands a high ISO

Maybe because longer exposures tend to be involved in low light conditions

I am sure the tech boys can forward an opinion about why longer exposures = more noise - heat whatever - I have no idea.

I would never choose 160 at 5.6 in a dark environ much go for a wider aperture and slower ISO or wide ap and faster shut if the movement/lens lenght  demanded it

Hopefully I would do BETTER under the same conditions having got away with 800ISO at 2 or 2.8

That image is therefore not likely to be relevant to me at least

I rekon the D3 will be good for 800 in real situations

This is exciting becuase I consider all other nikons not really usable beyond base ISO - 100

S

[Kevin W Smith]

True, but very dark situations often have large black areas where the noise shows a lot, as well as some tonal transitions from dark to mid-dark where the noise shows a lot.

Regards,
Bernard
[a href=\"index.php?act=findpost&pid=139007\"][{POST_SNAPBACK}][/a]

Longer exposure times increases noise, even if the light level hitting the sensor is the same on a shorter exposure.  IOW, if that trumpet player were a still life, and shot at the equivalent EI of f22 @ 1/10th, it might be a tad noisier.

For my main line of work, which often involves 15-30 second exposures, I think we can predict that since the sensor has better noise performance at high ISO's it will also have better image quality at low ISO's with long shutter speeds, because of less need for post-exposure NR reduction.

Anyway, as far as we can tell from these in-camera jpeg's, the high ISO noise performance is a -dramatic- leap forward for Nikon! I rarely need it (when I do I borrow my studio mate's Canon) but if I tried that shot with my D200 (or a D2x) it wouldn't be usable. Nikon's get funky at 400, by 1600 they're ugly, and 3200 pure junk. That kind of quality at 6400 is, subjectively, superb.

The low ISO model shots look real good too. Lovely skin tone, and smooth tonal gradations (for a jpeg).

Can't wait to see a 14-bit RAW file but I'm expecting good things. Like usual with Nikon you'll probably get better conversions at first with their software, but Adobe tends to tweak ACR's performance with pro cameras every revision - the first version that read my D200 files was okay but not great, but now they render quite nicely.

BTW, anybody wanna buy two D200's?  

[NikosR]

That would appear to be a completely rational argument.

However I still adhere to the opinion that my digital cameras, used in good light with stupid fast shutter speeds anda high ISO setting exhibit less noise than when I use the same cameras in a situation that actually demands a high ISO

Maybe because longer exposures tend to be involved in low light conditions

[a href=\"index.php?act=findpost&pid=139022\"][{POST_SNAPBACK}][/a]

Well, I believe that the explanation for the myth does not lie in long exposure noise (you generally need quite long speeds to show up the amp noise, this would not show up as different in a 1/15s vs 1/60s picture), but it lies in the fact that large areas of shadows or 'semi-shadows' so to speak, do tend to occur more often in dark lighting conditions. But this has to do with the subject and not with the average available light intensity. A scene with large shadow parts can also occur (just not so often) in lighter lighting conditions.

[Morgan_Moore]

Well, I believe that the explanation for the myth does not lie in long exposure noise (you generally need quite long speeds to show up the amp noise, this would not show up as different in a 1/15s vs 1/60s picture), but it lies in the fact that large areas of shadows or 'semi-shadows' so to speak, do tend to occur more often in dark lighting conditions. But this has to do with the subject and not with the average available light intensity. A scene with large shadow parts can also occur (just not so often) in lighter lighting conditions.
[{POST_SNAPBACK}][/a]

Quite possibly correct again. (not often I back down on this board!)

Anyway the bottom line is that my D200 and D80 are sh*t off of the base ISO

So bad that I actually have got better results exposing my 25ISO eyelike at 400 than using the D200

It looks like the camera will be far better - hence my order

A recent situation where I have desired ISO..

[a href=\"http://tinyurl.com/2opveu]winehouse[/url] (note the desaturation - I certainly had to burn the background 'back to black!')

SMM

[Christopher]

Hm to a lot of people go back to the 1DMk3 images and you will be suprised. 6400 there is just a total mess. I also thought it's not sooo special but after seeing images again from both cameras...

[Jonathan Wienke]

Digital Era Myth

There seems to be a persistent myth circulating in internet forums, that to show off the so called 'High ISO performance' one needs to shoot under very dark conditions (i.e Low EV).

I believe this is not true and I can find no rational justification for it. The only thing that is required is for the subject to exhibt a good mix of shadows and highlights and the ISO setting of the sensor to be set at a high value.

You obviously haven't done much high-ISO shooting then. Shoot a Color Checker at ISO 1600 outside, aperture se to get an exposure of 1/1000 or so, then shoot the same Color Checker inside in low light at ISO 1600 with aperture set to get a shutter of 1/2 or so. Your goal is to get both shots exposed so that the lightest neutral patch is about level 245-250 with exposure set to 0 in the RAW converter. Then compare noise levels in the darkest patches.

There are several kinds of noise in digital sensors. The most important are read noise, which is constant at a given ISO regardless of shutter speed, and there is another kind of noise (the name escapes me now) that is directly proportional  to exposure time, and is caused by electrons leaking out of the photosites on the chip. This noise tends to have a fixed-pattern component (hence the "long exposure noise reduction" mode on most DSLRs where a dark frame is taken after the main exposure and subtracted from it), but it has a random element as well. So the slower your shutter speed, the noisier the image will be at a given ISO, even if overall exposure is constant. The difference may not be noticeable with only a 1-stop difference in shutter speed, but that doesn't mean it isn't there.

[NikosR]

You obviously haven't done much high-ISO shooting then. Shoot a Color Checker at ISO 1600 outside, aperture se to get an exposure of 1/1000 or so, then shoot the same Color Checker inside in low light at ISO 1600 with aperture set to get a shutter of 1/2 or so. Your goal is to get both shots exposed so that the lightest neutral patch is about level 245-250 with exposure set to 0 in the RAW converter. Then compare noise levels in the darkest patches.

There are several kinds of noise in digital sensors. The most important are read noise, which is constant at a given ISO regardless of shutter speed, and there is another kind of noise (the name escapes me now) that is directly proportional  to exposure time, and is caused by electrons leaking out of the photosites on the chip. This noise tends to have a fixed-pattern component (hence the "long exposure noise reduction" mode on most DSLRs where a dark frame is taken after the main exposure and subtracted from it), but it has a random element as well. So the slower your shutter speed, the noisier the image will be at a given ISO, even if overall exposure is constant. The difference may not be noticeable with only a 1-stop difference in shutter speed, but that doesn't mean it isn't there.
[a href=\"index.php?act=findpost&pid=139114\"][{POST_SNAPBACK}][/a]

I would be happy to be shown images were long exposure noise is visible at the shutter speeds we have been talking about (that is handholding speeds). Please look at the posts in the beginning of this thread which argued about the validity regarding noise of the Nikon sample image.

Of course the dark-current noise, amp noise etc are visible in speeds approaching or exceeding the second. That's why most cameras introduce a dark frame subtraction noise reduction for long shutter speeds. But that is not what this discussion has been all about. I mentioned long exposure noise in a previous post which you failed to quote.

So please substantiate your argument with a suitable example at speeds relevant to this discussion. Any such test should also guarrantee that the colour temperature in both pictures remains the same and the white balance correction is the same also. Otherwise the difference in perceived noise introduced by the different channels and the different digital amplification required for correcting the white balance will make the test invalid.

[Sfleming]

How long berfore Michael pops up with a D3?  I'm sure he's arguing with himself about it.  But I'm equally sure its calling ... c a l l i n g.... to him.  What if Nikon really HAS leap-fogged Canon on image quality?  We need Michael Riechmann in there to find out for us!  

Sure,  he's at this point telling himself he'll wait for the high resolution version surely coming out within six months or so.  Yeah I tried that with myself.  But the D3 whispered that I could easilly fleabay it for full price soons the 'heavy hitter pixel monster' came out.

Resistence is futile MR.  You NEED a D3.  Buncha lenses too.

I'm 7th on the list in Austin, TX. for the cam and 24-70.  Probably get my unit in '08 or so.  %^/

(Now where did I stash that extra $7k I had lying around?)

[Jonathan Wienke]

So please substantiate your argument with a suitable example at speeds relevant to this discussion. Any such test should also guarrantee that the colour temperature in both pictures remains the same and the white balance correction is the same also.

I just did some tests, using the lighting in my room, with a high ISO setting and aperture priority mode, and varying the aperture from f/2 to f/32, shooting 9 frames at 1-stop intervals. The results weren't quite what I expected. My 1Ds does have somewhat higher noise at longer shutter speeds (range tested was 1/2 to 1/400), but the 1D-II actually goes the opposite way, with slightly lower noise at longer shutter speeds. I would imagine that it is possible there are other cameras that split the difference and stay about the same.

But in both cases, small exposure errors were much more significant in their effect on noise than the actual shutter speed. Proper exposure (exposing as far to the right as possible without blowing important highlights) is the most effective way to minimize noise at high ISO. Shutter speed (all else being equal) has a small effect on noise levels, but it is pretty insignificant with the cameras I've tested so far in the 1/2 - 1/400 range.

[BJL]

1. To Jonathan and Morgan:

I indirectly addressed this issue in my first response.

Yes, "noise reciprocity" fails at sufficiently long exposures, like 1/2 second, because then dark current noise can increase to significant levels. But as far as I know, this is irrelevant at hand-holdable shutter speeds, say 1/30s and faster.

Jonathan, what did your tests show at 1/30s and faster?


2. To Morgan: you seem to be saying that the main use for very high exposure index (IE, or 'ISO') is with long exposure times, and thus in what is over all extremely low light. I doubt it. High EI seems to be most often used to get moderate to fast shutter speeds for action photography, and I would think that most times where a long exposure like 1/2s is needed, one is dealing with a tripod and stationary subject matter, and so might as well go to an even longer exposure time and use base ISO speed.

3. To Bernard: you have a good point, which is that the harshest test of high EI settings is scenes of high subject brightness range, like ones with deep shadows, particularly if there is something to attract the viewer's curiosity in those shadows. Nikon's 6400 example seems to fit that bill.