E-3 banding problems

[OlyFlyer]

Hi,

currently there are a few threads on DPReview discussing E-3 banding problems, some say it is "huge" some call it "massive" and some say it is visible in every image, already at ISO 100. Some even say it is the worse of the Oly cameras. Unfortunatelly (?) I fail to find banding in my images and actually so far I have been asking for samples but could not really find any worth discussing. I did a few tests with my E-3, but I fail to see banding in my images.

What is your experience? I am mostly interested to read what other E-3 owners have to say. On DPR there are a lot of comments from non E-3 owners, and as usual, the problem seems to be bigger for those who only read reviews and have no experioence with the camera, or their experience is limited to a few test shots, no real life experience.

Here are two links:

http://forums.dpreview.com/forums/re...hread=28151051

http://forums.dpreview.com/forums/re...hread=28187271

The second one is started by myself and is illustrated with quite a few images. I hope you have the patience to read through, if not I am ready to post my images here as well for the sake of discussion.

[aryanto]

I am just wondering why not many people take photo of the sky or sunset photo where it is very easy to see the banding?
Or is that just almost all camera has this problem with sky / sunset?

[drakon09]

It is easy to see banding if:

1 - you're at above ISO 800
2 - and you seriously underexpose the shot;
3 - and then try to rescue the image by pulling back the exposure.

[OlyFlyer]

Originally Posted by aryanto I am just wondering why not many people take photo of the sky or sunset photo where it is very easy to see the banding?

Maybe E-3 shooters passed the need to photograph every sunset? Anyway, the sky is clearly visible in many of my images, but I still see no banding. I must admit, it is about 30 years ago I took the last sunset image, so maybe it is time to do it again, just to test banding.

Originally Posted by aryanto Or is that just almost all camera has this problem with sky / sunset?

I never knew there was a problem with sky / sunset images other than most of them can bore you to death. Anyway, in which way a sky / sunset image is different than any other images?

I don't want to be provocative, but I think your comment is just contributing to some of my statements where I say that most of the people who talk about banding don't even know how to reproduce or what banding actually is. Also, if PP:ed too hard, every camera with NMOS or CMOS sensor will give you banding.

[OlyFlyer]

Originally Posted by drakon09 It is easy to see banding if: 1 - you're at above ISO 800 2 - and you seriously underexpose the shot; 3 - and then try to rescue the image by pulling back the exposure.

OK, how seriously? If you look at my images, they are seriously underexposed and than I did some careful PP to rescue shadows but still see no banding. I tested with ISO 100, 400 and 1600.

[aryanto]

I actually never notice any banding in any of my pictures using E-3. That is probably because I did not edit them much, or I am just too oblivious and have low expectation. hahaha.

[drakon09]

Originally Posted by OlyFlyer OK, how seriously? If you look at my images, they are seriously underexposed and than I did some careful PP to rescue shadows but still see no banding. I tested with ISO 100, 400 and 1600.

Well, I managed to squeeze out some banding in a textureless surface after pulling up about three stops; though in my book, pulling up three stops means you really really messed up the shot in the first place.

As you said, all MOS-based images eventually will show patterned noise, if you torture the image hard enough!

[cjtune]

OK, so you have a good copy of an E-3... that doesn't make all of the other E-3's better. But maybe it's just in our heads, which means you have to try playing psychologist then.

I still remember when I was among the first to bring up the banding issue starting with the E-330, that fellow Raist3D in the DPReview forum was a vehement denier/skeptic. Today, his tune is completely different... hmmm.

[Mikefellh]

I've have always wondered if the banding issue is really user error, the way they resize the image for viewing on a monitor (and wonder if the monitor, CRT vs. LCD makes a difference).

Does the banding also show up when you print?

Having experience over decades with scanning, sometimes you get moire effects and sometimes you don't depending on how YOU resize the image.

Does the banding appear when vieing at 100%, or only when you've resized the image (either resampled, or reduced merely for viewing)?

[cjtune]

Originally Posted by Mikefellh Does the banding also show up when you print?

First time I brought this up to the local Oly support, I printed an example pix on A4.

They blamed my printer instead.

[Mikefellh]

Originally Posted by cjtune First time I brought this up to the local Oly support, I printed an example pix on A4.

OK, but that doesn't stop it from being a resizing issue since you didn't print an image at 100% (assuming you didn't crop the image to A4 size).

[OlyFlyer]

Originally Posted by drakon09 Well, I managed to squeeze out some banding in a textureless surface after pulling up about three stops; though in my book, pulling up three stops means you really really messed up the shot in the first place. As you said, all MOS-based images eventually will show patterned noise, if you torture the image hard enough!

Yes, if I torture an image, like taking it deliberatly underexposed by -2 or more and than try to restore to a good level than I also get all kind of pattern, but I can hardly call that a big issue or a generic problem. Also, if the image is correctly exposed for the largest part, careful PP can fix the darker parts without problem.

[cjtune]

Originally Posted by Mikefellh OK, but that doesn't stop it from being a resizing issue since you didn't print an image at 100% (assuming you didn't crop the image to A4 size).

If it's a resizing issue, why only affect higher ISO images?
Nah, it's not. I think there were drawn out arguments somewhere about this.

[OlyFlyer]

Originally Posted by cjtune OK, so you have a good copy of an E-3... that doesn't make all of the other E-3's better.

But the opposite is also true, and is more probable. Since very few have problems it is possible that their cameras are not working properly which means it is not a generic problem, just a problem with some camera individuals, just like the AF issue some of us had. Just because I had the problem I would never have said Oly had a generic AF problem. The funny thing is those on DPR who claim the E-3 having a generic banding problem also said the E-3 had a generic AF problem in EVERY camera. I can certainly verify my first camera had AF problem, but my second has not, so it is not a generic problem, since the majority of E-3 owners don't have the problems. Anyway, why I started my own tests was because some people said it shows up in normally exposed ISO 100 images, not only at ISO higher than 1600 in extreme situations. That is certainly not true. Of course, if it shows up at ISO 3200 in underexposed images, well, I am sorry, but so what? It is a well known fact that Oly is not the most optimal choise for the highest ISO, the ISO values above 1600 even start flashing on/off as if the camera is asking "Are you absolutely sure about your ISO selection?" Anyway, I would welcome some sample images with E-3 banding to learn something, but so far I have seen only images from one person who took OOC JPEG the first day he got the camera just to test the ISO3200. I took some real images up to ISO 1600 and there I can not see any banding.

Like the one below, ISO 400



The original



Shadow / highlight corrected



And a 100% crop of part of the corrected dark areas. The banding I see is just seen by a pixel peepers eyes and I am sure it is not visible in any prints or large screen display.

Not even my first E-3, the faulty one, demonstrated any banding at low ISO, here is one taken at ISO 100...



...and here is the same image with +1 EV and heavily lifted dark background using the shadow / highlight adjustments in PS.

Originally Posted by cjtune But maybe it's just in our heads, which means you have to try playing psychologist then.

I can't say what is in your head, but I am sure that there is no banding at low ISO unless there is a user or a camera error at some stage. I don't know how over ISO 1600 images are generally, but the ones I have taken for test show also no banding but look quite ugly due to ISO noise. Even those can IMO be fixed quite well with careful processing in PS, but they need more work.

Originally Posted by cjtune I still remember when I was among the first to bring up the banding issue starting with the E-330, that fellow Raist3D in the DPReview forum was a vehement denier/skeptic. Today, his tune is completely different... hmmm.

Well, maybe I will also change my mind, but first I have to know how to get banding. I don't usually take underexposed images of featureless walls or roof fans, and I don't regard those as representative test images to proof a problem. Even so, I actually tried to take some images of my roof as well but still failed.

In other words, I am still waiting for some sample images to show me when banding occures to be able to properly repeat and judge how important the "problem" is.

[OlyFlyer]

Originally Posted by Mikefellh I've have always wondered if the banding issue is really user error, the way they resize the image for viewing on a monitor (and wonder if the monitor, CRT vs. LCD makes a difference). Does the banding also show up when you print? Having experience over decades with scanning, sometimes you get moire effects and sometimes you don't depending on how YOU resize the image. Does the banding appear when vieing at 100%, or only when you've resized the image (either resampled, or reduced merely for viewing)?

Actually, it may be so that there are several things involved maybe that's why I can't really make it out. One parameter is that the image must be underexposed, but that is not the case for me. I do see some banding if I undreexpose an image and then try to restore, but why would one do that on purpose? Even so, it is normal that if an mage is underexposed some problems may turn up, like in the following examples from two different images.

ISO 1600, -2 EV underexposed, original resized image, noise not handled.



As the 100% crop below shows, there are all kind of patterns and even a very slight banding.



But again, this is pixel peeping bad photography.

Than I tried to "save" using Studio increased the exposure by +2 EV.



As the 100% crop below shows, there are all kind of patterns, terrible noise and even more banding, but apart from the noise which must be handled, I believe printing the image in full size won't be a problem.



According to me, in that case it is better to increase ISO to 3200 and expose right, as in the unprocessed image crop below.



There is less noise and to my eyes the image looks better than the ISO 1600 pushed up.

A better skilled person can probably do an even better job improving the image through processing, but even my very fast effort gives improvements to the above image.

My conclusion is still that the banding I can produce this way is acceptable and through proper processing the ISO noise can be reduced to an acceptable level for full size printing as well.

I don't know how serious this problem is, as those who talk about as something big don't have the E-3 or refuse to show samples or clear explanations.

[cjtune]

Originally Posted by OlyFlyer I can't say what is in your head, but I am sure that there is no banding at low ISO unless there is a user or a camera error at some stage. I don't know how over ISO 1600 images are generally, but the ones I have taken for test show also no banding but look quite ugly due to ISO noise. Even those can IMO be fixed quite well with careful processing in PS, but they need more work. Well, maybe I will also change my mind, but first I have to know how to get banding. I don't usually take underexposed images of featureless walls or roof fans, and I don't regard those as representative test images to proof a problem. Even so, I actually tried to take some images of my roof as well but still failed. In other words, I am still waiting for some sample images to show me when banding occures to be able to properly repeat and judge how important the "problem" is.

I think it doesn't really matter if people strive to prove to you or not banding exists, and it may be due too entirely to personal preferences as to what is an acceptable threshold - it's doesn't help affected E-3's get any better.

[OlyFlyer]

Originally Posted by cjtune I think it doesn't really matter if people strive to prove to you or not banding exists, and it may be due too entirely to personal preferences as to what is an acceptable threshold

Banding is more than personal preference. It is there or it is not there, to me it is very technical indeed. How disturbing it is is on the other hand totally personal.

Originally Posted by cjtune - it's doesn't help affected E-3's get any better.

It does indeed help. All that negative attitude on DPR must be balanced somehow. I came to ask here to find a more balanced answer and I believe if there was a REAL big problem it would be known to quite a few E-3 owners. You mean it is OK to proof something is bad by posting some images like a fellow who posted a real bad E-420 failure and say it is the same as from the E-3, but it is not OK to try to proof he is wrong?

[cjtune]

I think whether right or wrong, truth or untruth or personal preference is not really the matter to me or even the more important utility coming out of this thread or mine or any other on banding or related image defects.

It is the awareness about this issue that I am glad that has been finally brought up to the point that there are many people willing to acknowledge discuss this openly and provide test data to show to all, for or against it. Once awareness gains enough momentum, potential buyers will be in the know to test for it and see it for themselves, and even more importantly, is that Oly realises it is something they must look into too.

In my day job, once we have even 1 out of 16 units fail out of a sampling plan, alarm bells go on. This means we fail to achieve industry standards. When you talk about 100000 units per month going into the market, that's about 6000 unhappy or angry customers and will, supposedly according to Marketing, actively try to discourage at least 10 more potential customers from buying the same product or brand. Minorities will pretty soon be majority. So trying to show exhaustively how the 15 others didn't fail doesn't help unless the R&D team gets their hands on the failed units and dissect them thoroughly. I wish you luck to be able to find a worser-off E-3 or some other E-xxx that bands as I believe your Electronics qualification would go a long way into helping Oly resolve or improve the issue for all future users.

[hellfire8888]

i would say e3 got banding..yes got banding..

[OlyFlyer]

Originally Posted by hellfire8888 i would say e3 got banding..yes got banding..

OK, but how much? In which situations? Using which settings? I agree, I managed to produce some at ISO3200 underexposed than pushed up to normal but that is not the normal way of taking images, is it? So just by saying "e3 got banding..yes got banding" you add no information at all, just add to the negativity. We need facts not a negative attitude. AFAIK, every CMOS / NMOS sensor got banding, one more than the other but all has. There are threads concerning the D3 as well with banding issues, so the statement "e3 got banding" is about the same as saying "the sun will rise tomorrow". Quite obvious.