D7100 noise & banding vs other cameras

chary zp

Senior Member
Messages
1,489
Solutions
1
Reaction score
407
Reading this thread about "12 x 14 bit bla bla bla"


I did my own test. Rather underexposing a scene at both

14 and 12 bit, lossy and lossless and iso 100 and iso 800

and I have barely stopped gasping. The setting is in the name:

67b64df18439445ca1c72bd1b5677c14.jpg

WB is some strange issue. It is shot with a fixed preset and then picked to get the wall of the lit building "gray" using the marquee sample gray point. I cannot explain why is the WB so off between some shots. Maybe the marquee sampling is actually relative and not absolute?

This is how much I recovered:

4d7a37d7b43e441ca0cb98b6ffdf26b8.jpg

I know about the D7100 banding issue, but this is a bit too much to my liking, especially the amount of blue noise increasing towards the bottom of the frame.

Yes, it is pulled up very much, but anyway. The question is, is this the issue of all 24mpx chips or is for example D5300 chip much cleaner in this aspect? Is there a chance that the new D7200 will be a much more tolerant to huge recovering?
 
This is unique to the D7100 sensor, which is a Toshiba. Sony sensors don't band, but have some purple-blue shifting when pulled. All sensors shift when pulled to some extent. The D7100 can shift slightly green. You might want to wait for the D7200 or look into the 5300/5500.
 
Yes, this is what the D7100 looks like. Unfortunately the problem often comes up even without pulling up the shadows if shooting at very high ISOs (Hi1, Hi2 settings).

The D5300 is supposed to be free of this problem, but I don't have one to confirm this ...
 
Yes, this is what the D7100 looks like. Unfortunately the problem often comes up even without pulling up the shadows if shooting at very high ISOs (Hi1, Hi2 settings).
True, but if you use 3rd party Noise reduction/De-bandding, it can easily be fixed. And still not quite as bad as the issue was in some other bodies. The color shift is a bigger issue, IMO

One of My favorite examples:

Original photo - minus 5 stops exposure compensation
Original photo - minus 5 stops exposure compensation

Lightroom edited - 5 stops plus exposure
Lightroom edited - 5 stops plus exposure

JPEG - Moderate + De-banding in Topaz Denoise
JPEG - Moderate + De-banding in Topaz Denoise

The D5300 is supposed to be free of this problem, but I don't have one to confirm this ...
Seems to be a bit better.

--
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)
 
Last edited:
It might be an easy fix with expensive software. Is there anything free available?

Also, the discoloration comes in bands too (like at the bottom in the OP's example). Does Topaz fix that? I often get ugly darker/lighter bands on the second picture when shooting two photos in quick succession at very high ISOs. The problem with these is that they're visible even when reducing the size of the images (again like in the OP's thumbnails).
 
Last edited:
My old V8 motor use to valve bounce when I pushed it past the rev limits recommended. Does that mean I had a faulty motor?

regards
 
My old V8 motor use to valve bounce when I pushed it past the rev limits recommended. Does that mean I had a faulty motor?
This is a flawed analogy.

Considering that:
  • neither the previous (D7000/D5100) nor the following (D5300) generation of cameras has this issue,
  • that it was not at all easy to find out about the problem before purchase (e.g. Nikon sent a review sample to dpreview that doesn't have banding, check the dpreview samples),
yes, it is clearly a problem.

Did Nikon state that it is not recommended to use the D7100 at as high an ISO setting as the previous generation of cameras? Did they put a redline on it? If not then why wouldn't I expect it to perform at least as well as older ones?
 
Last edited:
My old V8 motor use to valve bounce when I pushed it past the rev limits recommended. Does that mean I had a faulty motor?

regards
I would say you had a poorly designed motor that either had design limits that were unrealistic (Ie it is normal for people to push it past the rev limit) and/or not enough robustness was accounted for in the design of the motor... how did you get over the limit, did your rev limiter break?
 
My old V8 motor use to valve bounce when I pushed it past the rev limits recommended. Does that mean I had a faulty motor?
This is a flawed analogy.

Considering that: neither the previous (D7000/D5100) nor the following (D5300) generation of cameras has this issue,
Also flawed considering that:
neither the previous (D7000/D5100) nor the following (D5300) generation of cameras has this issue,
The D7000 and D5100 are kind of outliers when it comes to aggressive shadow recovery. D5300 doesn't completely match how the D7K/D5100 could be pushed beyond what was considered "Normal"
...that it was not at all easy to find out about the problem before purchase (e.g. Nikon sent a review sample to dpreview that doesn't have banding, check the dpreview samples),
DPR never pushed the review sample hard. The Noise/bandding characteristics of the D7100 were widely documented as soon as the camera was released to the public. It's been a known for a very long time.
yes, it is clearly a problem.
For some and not for others. And as shown, can easily be "fixed/avoided" if really needed. Not unlike 5D3. Trade off's when you go 24mp in APS'c for the first time.
Did Nikon state that it is not recommended to use the D7100 at as high an ISO setting as the previous generation of cameras?
No reason you can't...I showed you how.
Did they put a redline on it?
Does anyone? Did Nikon every say the D7000/D5100 was designed to push the limits of shadow recovery capability? I've never seen them say that in any documentation or publication.
If not then why wouldn't I expect it to perform at least as well as older ones?
Sometimes (very often) expectations don't match reality. Camera tech reality is a dynamic thing. Are you upset the flash sync speed of the D7000 doesn't match the D70's?
 
My old V8 motor use to valve bounce when I pushed it past the rev limits recommended. Does that mean I had a faulty motor?

regards
I would say you had a poorly designed motor that either had design limits that were unrealistic (Ie it is normal for people to push it past the rev limit) and/or not enough robustness was accounted for in the design of the motor... how did you get over the limit, did your rev limiter break?
You simply don't over rev it if you don't want to have a little valve bounce (which can be "Fixed" if you really want to)

Same with the D7100 ( If your really don't want bandding and color shift when pushing the limits of shadow detail recovery...simply don't "over rev" it. Or go ahead and fix it with NR. Not a big deal for many. Pros's and Cons. That old V8 got better with later iterations...so will DSLR's.
 
It might be an easy fix with expensive software. Is there anything free available?
Topaz DeNoise is actually pretty cheap. Often lots of specials if you watch to. They gave me the Plugin for free with another promotion.
Also, the discoloration comes in bands too (like at the bottom in the OP's example). Does Topaz fix that?
From my experience, yes.
I often get ugly darker/lighter bands on the second picture when shooting two photos in quick succession at very high ISOs. The problem with these is that they're visible even when reducing the size of the images (again like in the OP's thumbnails).
That sounds like something unrelated to the traditional Noise/banding characteristics. Burst rate should have no affect, as it seems to be the case in what your describing.
 
My example of the V8 was simply to point out that there are limits to which a 'reasonable' person would go to before issues started to arise. We accept that noise comes with high ISO. So why is it that 'banding' is regarded as being unacceptable yet noise is?

I think that any man made article can be pushed beyond what is 'reasonable'. By exceeding those 'reasonable' limits we can reasonably expect that issue will arise.

At the end of the day any banding issue will only be an issue for a very very very small proportions of users. And if and when banding does occur then there are ways of addressing the issue.

We need to keep everything in perspective.

regards
 
At the end of the day any banding issue will only be an issue for a very very very small proportions of users. And if and when banding does occur then there are ways of addressing the issue.

We need to keep everything in perspective.

regards
I have 2 D7100s and never have had an issue with banding.
 
My example of the V8 was simply to point out that there are limits to which a 'reasonable' person would go to before issues started to arise. We accept that noise comes with high ISO. So why is it that 'banding' is regarded as being unacceptable yet noise is?

I think that any man made article can be pushed beyond what is 'reasonable'. By exceeding those 'reasonable' limits we can reasonably expect that issue will arise.

At the end of the day any banding issue will only be an issue for a very very very small proportions of users. And if and when banding does occur then there are ways of addressing the issue.

We need to keep everything in perspective.
I never disagreed with that and think it a good way to look at it. Nice though that every now and then a car comes along were the limits are pretty darn grey and going beyond the mark on tach isn't a big deal. D7000 was kind of like that :)
 
I think the times when you would probably be capable of producing the banding would be times the image would already look so bad that the banding would be least of worries. I've technically seen it in my d7100, but only very rarely and only when truly annihilating an image with shadow lifts etc. this is much different from canon banding which is actually visible in prints and in my opinion significant. I wouldn't worry about it!

Base ISO files from the d7100 are very close to my d610, so close that the difference is imperceptible in prints. Don't sweat the banding it's not going to ruin any shots (that weren't ruined already by exposure etc). The dynamic range and acuity is very impressive
 
My example of the V8 was simply to point out that there are limits to which a 'reasonable' person would go to before issues started to arise. We accept that noise comes with high ISO. So why is it that 'banding' is regarded as being unacceptable yet noise is?
Because, when lifting shadows, banding shows up nastily earlier than noise ;) It is also nasty that DR figures in various reviews don't account for banding and one can only reach the measured DR range in practice using special debanding software. And we were spoilt by earlier models, and FX Nikons never had this issue... Wide DR at base ISO used to be a traditional advantage over Canon sensors.
I think that any man made article can be pushed beyond what is 'reasonable'. By exceeding those 'reasonable' limits we can reasonably expect that issue will arise.

At the end of the day any banding issue will only be an issue for a very very very small proportions of users. And if and when banding does occur then there are ways of addressing the issue.
Not that everyone would use it, particularly considering the state of the art of the camera processing engine which is even much more restrictive than the sensor. Though, if all of a sudden one company would come up with good in-camera automatic HDR processing that could handle the output of today's better sensors comparably to what we can currently do using LR in half a minute, I believe this would be relevant to a much wider audience. Especially to casual photographers who are less careful in finding "good light".
 
Especially to casual photographers who are less careful in finding "good light".
I disagree.

All examples I have seen of banding are from photographers being a lot of careful to find "poor" light.

Actually most of them seem to be more "technicians" than "photographers".

That's of course my opinion only.

BirgerH.
 
I did my own test. Rather underexposing a scene at both
67b64df18439445ca1c72bd1b5677c14
4d7a37d7b43e441ca0cb98b6ffdf26b8

I know about the D7100 banding issue, but this is a bit too much to my liking, especially the amount of blue noise increasing towards the bottom of the frame.

Yes, it is pulled up very much, but anyway.
It's a matter of knowing when 'very much' is simply 'too much'.



689a46fabbce4f9dad8210eb3155d894.jpg

And this is simply said too much. Regarding the glow, hot sensor, light leak or something in-between. For me straight +/- 2EV is often plenty, I don't touch curves at that point. If the shot is not close to what I want, I just repeat leaving post for minor tweaks. Don't stress too much and -yes- the D7100 sensor is not ISO-less.

cheers
 

Keyboard shortcuts

Back
Top