B-b-b-b-banding!!!

Started Mar 26, 2013 | Discussions thread
Gar Ber
Regular MemberPosts: 115
Like?
Re: B-b-b-b-banding!!!
In reply to AngryCorgi, Mar 28, 2013

I've spent way too much time investigating this D7100 banding issue, but I hope all of you guys investing your time into this actually have reasons to, because judging by several replies in this thread (and several other older banding threads) seem to indicate some of you just need to hear "it's not that bad" and "it's not noticeable under normal situations". I'm not trying to say this is wrong, but it doesn't carry much relevance if you don't actually know what it all means for your photography style and workflow. There are some averaging indicators that you can compute without including any user bias (that would in the end tell you banding on D7100 it has x % or x dB impact on the final picture), but user preferences have an overwhelmingly greater influence on the importance on this in my humble opinion.

For example, although Canon sensors have been dealing with banding (albeit a little different in manifestation than on D7100, but anyway) for a long time, but most users that I know either don't ever come across it or have learned by now to deal with it one way or another to obtain the picture they want.

I think for current and future D7100 users it'll be the same. Most will never even see it, but those that need to, will be able to work around it.

That being said, the most important thing for you, the photographer, that reads about banding is to decide if it actually matters to you or not. For me it does for photography I do at least occasionally - night urban scenes at low iso (preventing street lamps from blowing means dealing with really deep shadows in other areas).

I'm not really sure why it is starting to become accepted that banding becomes apparent with above 2EV shadow pushing, because doing a 2EV push is a rather absolute picture manipulation, whereas noticing banding visually is a rather relative observation that is furthermore highly dependent on picture content and the medium where the observation takes place (monitor / printed picture, overhead projector, ...TV, etc).

What follows is a resume of my findings about D7100, they are based on actually taking pictures with a production D7100 myself (not owning the camera, but testing it) and analyzing raw pictures obtained from others (mostly DPreview users that posted links to their raws). Measurements were done using a limited number of pictures (<50), so please consider the numerical values of informative nature at best.

For banding to become computationally measurable (and as a  consequence potentially user observable) the originating raw brightness data needs to be low. Data seems to indicate any underlying "banding" in areas with original luminance above 2.7% quickly become unnoticeable, no matter the amount of shadow pushing. For the photographer this means: if you don't have really dark areas , you will most likely not be able to tell apart any possible banding from the actual picture even if pushing really hard, or also in other words ETTR will help here.

There is some indication in data that "banding" has some periodicity and is row-based (in landscape orientation), however this depends on the demosaicing process used. Statistical averaging points to a period of 4 pixels. For the photographer this means: if you need to deal with banding, you might get very different results using Lightroom, DXO, Capture One, Silkypix, Noise Ninja, UFRaw, Rawtherapee or some other software to read your raws and then to remove the noise. Furthermore, the banding will only be observable, even if not dealt with at all, when the picture is viewed at large enough sizes, A4 and A3 prints will not exhibit any sort of banding even if untreated.

Data shows that high iso grain overcomes banding amplitude. I don't have nearly enough pictures of the same setting at various iso settings to see where the boundary is, but it seems to be somewhere at or between ISO400 and ISO800. To the viewer it will most likely be lower, so more towards ISO 400 than ISO 800, since it is unlikely any viewer will be able to visually isolate single lines from the others given the growing luminance of grain - this is just a personal observation. For the photographer this means: if you shoot at higher iso settings you'll be dealing more with grain noise than banding.

In the end please take my findings with caution: they are based on a limited amount of pictures and on a limited time test of D7100. The most important thing: if you really think about getting a D7100 and you're concerned about banding, think if this kind of banding would actually make a difference to you or not, because the impression I got is that otherwise D7100 performs quite well.

Cheers

Reply   Reply with quote   Complain
Post (hide subjects)Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow