D7100 Banding Issue...

Started 7 months ago | Discussions thread
mosswings
Veteran MemberPosts: 6,004Gear list
Like?
Re: High ISO and no detail
In reply to Mako2011, 7 months ago

Mako2011 wrote:

swimswithtrout wrote:

Nice shot, but that's why I continually point back to my Astrophotos, that are pushed WELLLLL..... beyond 5 stops, from the depths of a black hole itself, using ISO 1600 to 3200 and using 2-3 MINUTE exposures ! Where's the banding ?

Again, at ISO 1600-3200, with zero detail in the shadows (dark sky), I would not expect to see banding. That's just not a situation it would normally show in.

Here they are again, huge crops, since they were taken with just the Nikon 70-300 and not a telescope. There's plenty of random noise, but where's the banding ??

Shouldn't see any. Try doing that at ISO 100 12-bit RAW in the shadows where there is some detail in the scenes dark areas being pushed. In cases like that, it starts to be seen with even minor increase in EV (1-2 EV) but you still have to zoom in to see it. 14-bit reduces it as does shooting JPEG. Just not a problem for most shooting situations, as you just demonstrated.

-- hide signature --

My opinions are my own and not those of DPR or its administration. They carry no 'special' value (except to me and Lacie of course)

Let me add a little more technical detail to Mako's correct observation.

Banding is caused by the residual mismatch in sensor readout circuit elements. Since sensor architectures group all sensel outputs within a row or column together, it the mismatch in the row or column buffers/converters that is creating this, and it's very, very small - within the last 2 bits of a 14 bit data word. The converters on the Toshiba (and all other contemporary sensors) are very very good; they can resolve those last two bits quite precisely, which is another way of saying that system noise at ISO 100 is extremely low. However, systematic mismatch in the readout circuitry is comparable to the noise, actually slightly higher, so it appears under conditions where the sensor's DR is maxed out - i.e., at base ISO. 12 bit operation can make the mismatch more prevalent because each bit in the readout represents a larger change in brightness, and if device mismatch+noise flips a bit at the bottom end, it's more visible.

When ISO is increased above base, analog gain is taken and noise will increase. This buries the device mismatch in the noise and so the banding.

Sony's EXMOR sensors are "bulletproof" because they use a sophisticated architecture that digitally corrects for a plethora of device mismatches. This complex architecture (which, by the way, also has its drawbacks in the form of higher heat generation and noise in long exposures - observed in the D800 because it uses two digital converter banks, one on each side of the sensor) is heavily patent protected.

The Toshiba sensor also shows some imperfections in its sensel isolation - where the readout of one pixel can affect the value of another. This is called "streaking", and looks like a very very VERY weak form of CCD "blooming". It's again due to another design compromise in the pursuit of power reduction, and again Sony's EXMOR technology addresses it more effectively than Toshiba's.

Finally, all later model Nikons until the D5300 exhibited some green shifting in pulled shadows. This has to do with a trick Nikon has used to extend dynamic range in single-exposure images (the vast bulk of images taken) - it clips the distribution of the black signal recovered from the "black reference" sensels at the edges of the sensor at their peak rather than preserve the entire distribution. Absolute black becomes a bit lower in the sensor range, but the shadows don't hold color as well. Other manufacturers preserve the entire distribution and so tend to maintain pulled shadow color consistency a bit better. This effect, again, is very small and seen only in heavily pulled shadows.

What this clipping trick does do, however, is make low-ISO multiple-exposure astrophotography problematic with Nikons as extremely dim stars disappear in the clipped lower tails of the black distribution and never get a chance to build up into visibility over many exposures. The D5300 is the first modern Nikon to facilitate this sort of photography.

 mosswings's gear list:mosswings's gear list
Olympus XZ-1 Nikon D90 Nikon D7100 Nikon AF-S DX Nikkor 18-105mm f/3.5-5.6G ED VR Nikon AF-S Nikkor 70-300mm f/4.5-5.6G VR +1 more
Reply   Reply with quote   Complain
Post (hide subjects)Posted by
TrueNew
THATNew
TrueNew
trueNew
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow