D7100 'streaking'

Started Apr 7, 2013 | Discussions
PepsiCan Contributing Member • Posts: 739
Re: UWB

Jack Hogan wrote:

PepsiCan wrote: Expose correctly in the camera for best results.

Undisputable.  Now define 'correctly'

Let me give that a try.

Using Unitary White Balance, expose to the right.

Hmm...we may need a separate thread for this

 PepsiCan's gear list:PepsiCan's gear list
Canon PowerShot S95 Adobe Photoshop Lightroom 4 Nikon D7100 Nikon AF-S DX Nikkor 16-85mm f/3.5-5.6G ED VR Nikon AF-S DX Nikkor 35mm F1.8G +5 more
fotolopithecus Senior Member • Posts: 1,699
Re: What it boils down to...

krikman wrote:

There is a resource on the web http://www.focus-numerique.com/.

They reviewed photo and have similar shots made by d7000, d5200, d7100:

d5200:

d7100:

d7100

d7000:

These shots are not precisely the same, but it does appear that the D7000 is not exhibiting any banding, although it's hard to say if it would if the shots were identical. I think we could safely say there's less though.

 fotolopithecus's gear list:fotolopithecus's gear list
Nikon D610 Tamron SP AF 90mm F/2.8 Di Macro Nikon AF-S Nikkor 50mm f/1.8G
krikman Regular Member • Posts: 419
Re: What it boils down to...

These shots are not precisely the same, but it does appear that the D7000 is not exhibiting any banding, although it's hard to say if it would if the shots were identical. I think we could safely say there's less though.

They all shot with same lighting at 1/4 f/5.6

D7000   24 f/1.4

other 50 f/1.4

All had +3 exposure in ALR and +100 shadows. No detail ajustments (sharpening and NR)

All other setting are same.

fotolopithecus Senior Member • Posts: 1,699
Re: What it boils down to...

krikman wrote:


These shots are not precisely the same, but it does appear that the D7000 is not exhibiting any banding, although it's hard to say if it would if the shots were identical. I think we could safely say there's less though.

They all shot with same lighting at 1/4 f/5.6

D7000   24 f/1.4

other 50 f/1.4

All had +3 exposure in ALR and +100 shadows. No detail ajustments (sharpening and NR)

All other setting are same.

I was referring more to the framing, and the dark area of the cameras where the banding is more visible. The dark area on the D7000 is smaller, and so might conceal some banding, or not. I'm not sure what if any effect the differing focal lengths might have on this, but there might be some. Anyway, as I mentioned I think it could be said that if the D7000 has banding that it is quite a bit less.

 fotolopithecus's gear list:fotolopithecus's gear list
Nikon D610 Tamron SP AF 90mm F/2.8 Di Macro Nikon AF-S Nikkor 50mm f/1.8G
krikman Regular Member • Posts: 419
Re: What it boils down to...

I was referring more to the framing, and the dark area of the cameras where the banding is more visible. The dark area on the D7000 is smaller, and so might conceal some banding, or not. I'm not sure what if any effect the differing focal lengths might have on this, but there might be some. Anyway, as I mentioned I think it could be said that if the D7000 has banding that it is quite a bit less.

Agree. And I somewhat tired by our speculations about D7100 internal problems while waiting delivery of this perfect, i said excellent body.

Mark

_sem_ Veteran Member • Posts: 5,032
Re: What it boils down to...

krikman wrote:

There is a resource on the web http://www.focus-numerique.com/.

They reviewed photo and have similar shots made by d7000, d5200, d7100: ...

I think this test is not useful because it is not clear if there is streaking involved. The D7100/D5200 surely have banding but that's much less of a problem than the cases of streaking seen in this thread. Streaking might, following the D300s sample, affect the D7000 too...

Jack Hogan Veteran Member • Posts: 6,982
Re: What it boils down to...

krikman wrote:

There is a resource on the web http://www.focus-numerique.com/.

They reviewed photo and have similar shots made by d7000, d5200, d7100:

Amazing how much sharper is the D7100.

mosswings Veteran Member • Posts: 9,132
Re: Discussion closed.

fotolopithecus wrote:

krikman wrote:

Don't worry guys.
I still have no D7100 at hand, but I have (A) window, (B) screen, (C) Nikon D300s.

Nikon D300s proven for absulutely no visuble artifacts.

ISO400. Exposition by left part. +5EV +recovered shadows. Bands is here.

The reason we found it in D7100 is because nobody pushed shadows that much in D300 because of noise.

Bravo Nikon! Just too low noise in shadow area!

Wasn't the OP's shot at 200iso, with +2 ? His D90 didn't seem to exhibit anything under the same conditions.

Yes, but the D90 was shot in the only recording mode it has, 12 bit lossy.  Krikman's D300 shots were recorded in 14 bit lossless mode.  On the D90/D300 generation sensors, this generates mostly just noise in those lower two bits.  What this image says is that the streaking effect is very small and mostly getting lost in the D300/D90 noise, which the D90 then just truncates (actually, the A/D conversion stops at the 12 bit level, but on the D300 it continues until the conversion -theoretically- settles out to 14 bits).  The D7000 and D7100/D5200 record normally in 14 bit mode and have significantly lower noise floor which mandates 14 bit (otherwise you'd get noise floor posterization).

One of the things I notice in this image is the lack of tonal shifts in the streaks.  By contrast, there is a definite red shift in the D7100 shots streaks - or perhaps I should say a green shift in the dark sections.

I've shot a couple of test images myself but haven't been able to induce streaking in my D90.  That doesn't say I did the test right, though.

The more interesting tests are Horshack's.  They reveal a noticeably worse streaking in the D7100 sensor than in the D7000, and almost no streaking in the full frame sensors - even the Canon, which has strong banding.  So there seems to be some correlation between sensor size and streaking robustness, but EXMOR sensors are not completely free of the effect.  It may be that the tonal shifts in the streaks in the D7100 are making the effect more obvious than in the D7000. And this might circle back to a consequence of the choices made in per-channel black clipping in each.

I'm not sure if it was Jack Hogan or Jim Pearce that said this, but whoever it was characterized the Toshiba sensor as "edgier"; that just like the D300 you needed to get exposure right or it could bite you.  The D7000 sensor is a remarkably forgiving sensor - it limits gracefully under pretty much all bizarre conditions that we can think up, which has conditioned us to expect the ability to do photographic contortions with our cameras on a whim.  The D7100 is more of a photographic gymnast - capable of fantastic things, but requiring more skill and control from the photographer, and often operating on the edge of disaster.

A bit florid, perhaps, but I enjoyed the metaphor. Or is it analogy?

 mosswings's gear list:mosswings's gear list
Olympus XZ-1 Olympus Stylus 1 Nikon D90 Nikon D7100 Sony a6400 +7 more
fotolopithecus Senior Member • Posts: 1,699
Re: Discussion closed.

mosswings wrote:

fotolopithecus wrote:

krikman wrote:

Don't worry guys.
I still have no D7100 at hand, but I have (A) window, (B) screen, (C) Nikon D300s.

Nikon D300s proven for absulutely no visuble artifacts.

ISO400. Exposition by left part. +5EV +recovered shadows. Bands is here.

The reason we found it in D7100 is because nobody pushed shadows that much in D300 because of noise.

Bravo Nikon! Just too low noise in shadow area!

Wasn't the OP's shot at 200iso, with +2 ? His D90 didn't seem to exhibit anything under the same conditions.

Yes, but the D90 was shot in the only recording mode it has, 12 bit lossy.  Krikman's D300 shots were recorded in 14 bit lossless mode.  On the D90/D300 generation sensors, this generates mostly just noise in those lower two bits.  What this image says is that the streaking effect is very small and mostly getting lost in the D300/D90 noise, which the D90 then just truncates (actually, the A/D conversion stops at the 12 bit level, but on the D300 it continues until the conversion -theoretically- settles out to 14 bits).  The D7000 and D7100/D5200 record normally in 14 bit mode and have significantly lower noise floor which mandates 14 bit (otherwise you'd get noise floor posterization).

One of the things I notice in this image is the lack of tonal shifts in the streaks.  By contrast, there is a definite red shift in the D7100 shots streaks - or perhaps I should say a green shift in the dark sections.

I've shot a couple of test images myself but haven't been able to induce streaking in my D90.  That doesn't say I did the test right, though.

The more interesting tests are Horshack's.  They reveal a noticeably worse streaking in the D7100 sensor than in the D7000, and almost no streaking in the full frame sensors - even the Canon, which has strong banding.  So there seems to be some correlation between sensor size and streaking robustness, but EXMOR sensors are not completely free of the effect.  It may be that the tonal shifts in the streaks in the D7100 are making the effect more obvious than in the D7000. And this might circle back to a consequence of the choices made in per-channel black clipping in each.

I'm not sure if it was Jack Hogan or Jim Pearce that said this, but whoever it was characterized the Toshiba sensor as "edgier"; that just like the D300 you needed to get exposure right or it could bite you.  The D7000 sensor is a remarkably forgiving sensor - it limits gracefully under pretty much all bizarre conditions that we can think up, which has conditioned us to expect the ability to do photographic contortions with our cameras on a whim.  The D7100 is more of a photographic gymnast - capable of fantastic things, but requiring more skill and control from the photographer, and often operating on the edge of disaster.

A bit florid, perhaps, but I enjoyed the metaphor. Or is it analogy?

Are you saying mossy, that the D7000 is more like a tube amp, and the D7100 more like a solid state amp when it comes to clipping the music?

 fotolopithecus's gear list:fotolopithecus's gear list
Nikon D610 Tamron SP AF 90mm F/2.8 Di Macro Nikon AF-S Nikkor 50mm f/1.8G
mosswings Veteran Member • Posts: 9,132
Re: Discussion closed.
1

fotolopithecus wrote:

mosswings wrote:

I'm not sure if it was Jack Hogan or Jim Pearce that said this, but whoever it was characterized the Toshiba sensor as "edgier"; that just like the D300 you needed to get exposure right or it could bite you.  The D7000 sensor is a remarkably forgiving sensor - it limits gracefully under pretty much all bizarre conditions that we can think up, which has conditioned us to expect the ability to do photographic contortions with our cameras on a whim.  The D7100 is more of a photographic gymnast - capable of fantastic things, but requiring more skill and control from the photographer, and often operating on the edge of disaster.

A bit florid, perhaps, but I enjoyed the metaphor. Or is it analogy?

Are you saying mossy, that the D7000 is more like a tube amp, and the D7100 more like a solid state amp when it comes to clipping the music?

Oooo, audiophile references, I love it!  Perhaps.  Tube amps generate more even harmonics and gentler rollloffs into clipping, which sounds more musical and warm.  But solid state amps generate more odd harmonics, which are harsher, and will clip strongly.  Both cover the music with subtle veils of their own unique character.  Which viel you choose is up to you.  However, if either one started generating a Madras shirt when reproducing certain types of music I'd get worried.

Now...where are my Tiptoes amplifier feet to reduce the effect of microphonics on my solid state amplifier?  You can hear their benefit, you know...but only with 0000 gauge transmission line Litz wired oxygen free cables which any truly serious audiophile would use as a matter of course.

Next, let's talk wine.

 mosswings's gear list:mosswings's gear list
Olympus XZ-1 Olympus Stylus 1 Nikon D90 Nikon D7100 Sony a6400 +7 more
krikman Regular Member • Posts: 419
Re: Discussion closed.

I'm not sure if it was Jack Hogan or Jim Pearce that said this, but whoever it was characterized the Toshiba sensor as "edgier"; that just like the D300 you needed to get exposure right or it could bite you.  The D7000 sensor is a remarkably forgiving sensor - it limits gracefully under pretty much all bizarre conditions that we can think up, which has conditioned us to expect the ability to do photographic contortions with our cameras on a whim.  The D7100 is more of a photographic gymnast - capable of fantastic things, but requiring more skill and control from the photographer, and often operating on the edge of disaster.

A bit florid, perhaps, but I enjoyed the metaphor. Or is it analogy?

Nobody can rob us of enjoing photography.

krikman Regular Member • Posts: 419
Re: Discussion closed.
Now...where are my Tiptoes amplifier feet to reduce the effect of microphonics on my solid state amplifier?  You can hear their benefit, you know...but only with 0000 gauge transmission line Litz wired oxygen free cables which any truly serious audiophile would use as a matter of course.

Next, let's talk wine.

Don't drink too much.

fotolopithecus Senior Member • Posts: 1,699
Re: Discussion closed.

mosswings wrote:

fotolopithecus wrote:

mosswings wrote:

I'm not sure if it was Jack Hogan or Jim Pearce that said this, but whoever it was characterized the Toshiba sensor as "edgier"; that just like the D300 you needed to get exposure right or it could bite you.  The D7000 sensor is a remarkably forgiving sensor - it limits gracefully under pretty much all bizarre conditions that we can think up, which has conditioned us to expect the ability to do photographic contortions with our cameras on a whim.  The D7100 is more of a photographic gymnast - capable of fantastic things, but requiring more skill and control from the photographer, and often operating on the edge of disaster.

A bit florid, perhaps, but I enjoyed the metaphor. Or is it analogy?

Are you saying mossy, that the D7000 is more like a tube amp, and the D7100 more like a solid state amp when it comes to clipping the music?

Oooo, audiophile references, I love it!  Perhaps.  Tube amps generate more even harmonics and gentler rollloffs into clipping, which sounds more musical and warm.  But solid state amps generate more odd harmonics, which are harsher, and will clip strongly.  Both cover the music with subtle veils of their own unique character.  Which viel you choose is up to you.  However, if either one started generating a Madras shirt when reproducing certain types of music I'd get worried.

Now...where are my Tiptoes amplifier feet to reduce the effect of microphonics on my solid state amplifier?  You can hear their benefit, you know...but only with 0000 gauge transmission line Litz wired oxygen free cables which any truly serious audiophile would use as a matter of course.

Next, let's talk wine.

Lol, yes exactly right mossy, but to me the D7000 sounds more like music.

 fotolopithecus's gear list:fotolopithecus's gear list
Nikon D610 Tamron SP AF 90mm F/2.8 Di Macro Nikon AF-S Nikkor 50mm f/1.8G
mosswings Veteran Member • Posts: 9,132
Re: Discussion closed.

fotolopithecus wrote:

Lol, yes exactly right mossy, but to me the D7000 sounds more like music.

Have to say that you might be right at least in one sense...that the D7000 may inspire more confidence in us geeky types whose technique is not always up to snuff.  The D7100 can probably pull out more if you know what you're doing.  It seems we may have reached a peak in user friendliness with the D7000 iimaging chain.

That makes for a tougher buy decision.  Again, we are in danger of losing perspective here, because these behaviors may have near zero relevance to shot quality in most scenes.  Helmets and airbags have given some a heightened feeling of safety and have increased their risky behavior, diminishing the safety benefits somewhat.  But at the end of the day it's proper technique and prudent decision making that keep you the safest.

It's tough when it sinks in that we have already reached the point of diminishing returns with current Bayer sensor technology and that we have to return to film-era technique.  But look to where the bar has been moved...3-4 stops better high ISO of quality...insane levels of resolution and postprocessing potential...superb AF and remarkable lens price/performance.  I always have to sit myself down with a glass of wine and remember that.

 mosswings's gear list:mosswings's gear list
Olympus XZ-1 Olympus Stylus 1 Nikon D90 Nikon D7100 Sony a6400 +7 more
mosswings Veteran Member • Posts: 9,132
Streaking is one of several image-dependent sensor artifact mechanisms

Unfortunately, it is rather ill-understood or part of the most closely held technological know how of the sensor manufacturers, as it is one of the key differentiating features of their products.  I was able to find a few general articles on the subject, but the pictures in this otherwise useless blog article from a machine-vision camera manufacturer are very interesting as they look a lot like what we're seeing:

http://info.adimec.com/blogposts/bid/88624/CCD-vs-CMOS-Image-Artifacts-to-Consider-with-CMOS-Image-Sensors

Column noise, line noise (streaking), and black level shift stem from imperfections in the dynamic bias and cell reset circuitry, it seems.

This article from the 2009 Intl. Image Sensor Society Conference discuss other such dynamic pattern noise mechanisms. In particular, what the authors call "smearing" can be traced to a cell biasing voltage that is slow to settle:

http://www.imagesensors.org/Past%20Workshops/2009%20Workshop/2009%20Papers/073_paper_purcell_invited.pdf

The next article is pretty fabrication technology geeky, but shows just what incredibly complex dynamic adjustments are going on at the pixel level to yield the ultra low noise levels we are enjoying.

http://cdn.intechopen.com/pdfs/17221/InTech-Image_artifacts_by_charge_pocket_in_floating_diffusion_region_on_cmos_image_sensors.pdf

imagesensors.org, the website of the IISS, has a wealth of papers on low level electrical and fabrication design and measurement.  It rather boggles my mind what imaging engineers are doing these days.

However, what this suggests to me is that the effect we're seeing is happening at the lowest levels of the sensor circuitry...and because its geometry dependent, it is not amenable to postprocessing tricks.

 mosswings's gear list:mosswings's gear list
Olympus XZ-1 Olympus Stylus 1 Nikon D90 Nikon D7100 Sony a6400 +7 more
Iliah Borg Forum Pro • Posts: 25,646
Re: Streaking = Smart IQ
1

From the discussion here it looks more like small problem in analogue domain. Maybe some samples of NEFs 14-bit lossless can help to understand better - the links to those posted earlier already expired.

_sem_ Veteran Member • Posts: 5,032
Re: Streaking = Smart IQ

Iliah Borg wrote:

From the discussion here it looks more like small problem in analogue domain. Maybe some samples of NEFs 14-bit lossless can help to understand better - the links to those posted earlier already expired.

I think one D5200 sample is still available here (_DSC0066.NEF):

http://www.dpreview.com/forums/post/50800689

while the best Horshack's samples were not NEFs:

http://www.dpreview.com/forums/post/51260513

But considering the D300s sample:

http://www.dpreview.com/forums/post/51269663

question is whether this is in fact special to the D7100/D5200, or occurs with other wide-DR sensors too, if one looks hard enough or finds the right test conditions.

RudyPohl Veteran Member • Posts: 5,562
Re: Discussion closed.

mosswings wrote:

fotolopithecus wrote:

Lol, yes exactly right mossy, but to me the D7000 sounds more like music.

Have to say that you might be right at least in one sense...that the D7000 may inspire more confidence in us geeky types whose technique is not always up to snuff.  The D7100 can probably pull out more if you know what you're doing.  It seems we may have reached a peak in user friendliness with the D7000 iimaging chain.

That makes for a tougher buy decision.  Again, we are in danger of losing perspective here, because these behaviors may have near zero relevance to shot quality in most scenes.  Helmets and airbags have given some a heightened feeling of safety and have increased their risky behavior, diminishing the safety benefits somewhat.  But at the end of the day it's proper technique and prudent decision making that keep you the safest.

It's tough when it sinks in that we have already reached the point of diminishing returns with current Bayer sensor technology and that we have to return to film-era technique.  But look to where the bar has been moved...3-4 stops better high ISO of quality...insane levels of resolution and postprocessing potential...superb AF and remarkable lens price/performance.  I always have to sit myself down with a glass of wine and remember that.

Excellent summation... thanks.

Rudy

Whalligeo Regular Member • Posts: 245
Re: Discussion closed.

mosswings wrote:

fotolopithecus wrote:

Lol, yes exactly right mossy, but to me the D7000 sounds more like music.

Have to say that you might be right at least in one sense...that the D7000 may inspire more confidence in us geeky types whose technique is not always up to snuff.  The D7100 can probably pull out more if you know what you're doing.  It seems we may have reached a peak in user friendliness with the D7000 iimaging chain.

That makes for a tougher buy decision.  Again, we are in danger of losing perspective here, because these behaviors may have near zero relevance to shot quality in most scenes.  Helmets and airbags have given some a heightened feeling of safety and have increased their risky behavior, diminishing the safety benefits somewhat.  But at the end of the day it's proper technique and prudent decision making that keep you the safest.

It's tough when it sinks in that we have already reached the point of diminishing returns with current Bayer sensor technology and that we have to return to film-era technique.  But look to where the bar has been moved...3-4 stops better high ISO of quality...insane levels of resolution and postprocessing potential...superb AF and remarkable lens price/performance.  I always have to sit myself down with a glass of wine and remember that.

Quite agree, and after the glass of wine?.......

Lets start nagging about having changeable camera backs, specialized for high iso, slow iso, high res, low res, warm colours, cool colours, something that looks like Kodachrome 25 or 64, Velvia or Ektachrome. How about mono, now that would be something....We could kick it all off again, but this time, upgrades are just a swap of the camera back (sensor).

-- hide signature --

They said it couldn't be done, so I encouraged my peers not to bother.

mosswings Veteran Member • Posts: 9,132
Re: Streaking = Smart IQ

_sem_ wrote:

Iliah Borg wrote:

From the discussion here it looks more like small problem in analogue domain. Maybe some samples of NEFs 14-bit lossless can help to understand better - the links to those posted earlier already expired.

I think one D5200 sample is still available here (_DSC0066.NEF):

http://www.dpreview.com/forums/post/50800689

while the best Horshack's samples were not NEFs:

http://www.dpreview.com/forums/post/51260513

But considering the D300s sample:

http://www.dpreview.com/forums/post/51269663

question is whether this is in fact special to the D7100/D5200, or occurs with other wide-DR sensors too, if one looks hard enough or finds the right test conditions.

Iliah, thanks for joining in.

_sem_, Andrea Dawn's sunset rock shot (dsc0066.nef) is probably not a good candidate for this discussion.  There is a sun-aligned lightening of the shadowed rocks on the left, but it appears to be part of a constellation of sun rays and possibly a lens flare. I count at least 13 rays at all angles. The line artifacts seen there appear to be nothing more than regular deep shadow banding.

 mosswings's gear list:mosswings's gear list
Olympus XZ-1 Olympus Stylus 1 Nikon D90 Nikon D7100 Sony a6400 +7 more
Keyboard shortcuts:
FForum MMy threads