OMG! My 7D has Vertical Banding!

[GPapa]

I’ve been following this 7D vertical banding discussion for a few months. It just won’t go away. For me it has either been convincingly shown to result from operator error or to be the result of deliberate, bad photographic process. I've heard references to individual instances where single body had exceptional banding but have not seen any convincing evidence. So I remain a skeptic.

Some vertical banding threads are perpetuated by brandX fanboys claiming vertical banding is a typical characteristic of 7D images and it is a fatal flaw in the latest Canon brand technology. That their brandX technology doesn’t suffer from this fatal flaw somehow proves brandX is superior. I'm admittedly speculating a bit here. But I find these threads offensive and I think it causes harm to people trying to learn about DSLR photographic equipment. Vertical banding isn’t a typical characteristic of a 7D image. Neither Canon nor brandX are superior. They are different and each have their advantages and weaknesses.

Other threads are perpetuated by a small band of loyal Canon zealots who have made it their goal in life to bad mouth Canon on the internet so it will hurry up and develop a sensor technology that allows them to shoot award winning photos without having to worry about proper exposure. These guys sound technologically reasonable and astute but are tilting at windmills and are also causing harm to people trying to learn about currently available DSLR photographic equipment.

I don’t recall anyone ever starting a thread sincerely asking for help to solve a problem they were having with vertical banding on a 7d; at least not one where the conversation wasn’t quickly dominated by fanboys and zealots.

I haven’t noticed vertical banding to be a problem with my photos. But then, I shoot mostly sports, in JPEG, at high ISOs and I learned a long time ago from the Sports Illustrated Photographer’s Guide, to “expose to the right” to avoid unpleasant noise. Apparently this combination is one of the least likely places for noise, including vertical banding, to show up.

So . . . OK, then, the next question for me is, what do I have to do to see if my camera has a vertical banding problem? I don’t want to discover “how” by accident at some critical shoot where I’m not shooting well exposed JPEGs at high ISOs.

I came up with a test based on what I understand from the ongoing vertical banding threads. The following test answers my question well enough for my purposes. Try it yourself and be your own judge. If you are still in the information gathering stage then all I can offer you are my results and my observations of the ongoing discussion.

The Test Subject:

I picked a subject that had some detail, but not so much that the detail would over power the vertical banding noise. The subject needed to have both good light and good shadows so the vertical banding would have a good place to appear. I chose a black document case with some texture to capture some light and shadow contrast and a black metal frame which offered some more contrasting detail and shadow areas. I placed it next to a north facing window in a room with white painted walls and fluorescent lights.

The Test Setup:

I used live view to focus on the edge of the case nearest to me with a 70-200 f/2.8 lens mounted on a 7D in full manual mode. The subject was about 6 feet away. I used ISO 100, which was said to have the most prominent vertical banding. Using the 2 second timer and with the camera resting on a very solid desk, I took a series of images at f/2.8, f/4, f/5.6, f/8, and f/11 without adjusting any other factors.

The first capture was my idea of a well exposed image. The last one, at f/11, was so under exposed I hardly got any detail out of the well lit panel facing the window.

I loaded each of the images into Adobe Camera Raw and took three screen shots.

1. One showing how the full image looked along with the histogram and settings “As Shot”.
2. One showing how the full image looked after pushing the exposure 4 stops.
3. And, one showing the area of the image with the worst case of vertical Banding at 100% after pushing the exposure 4 stops.

I also provided a link to the original screen capture for each image, not for pixel peeping, but for those who want to see one of the images with a bit more magnification.

I have exceeded the 6,000 character limit for a single post so The Test Results are in the post immediately following this one.

 

[GPapa]

The Results:

Below: ISO 100, f/2.8, 1/4s, As Shot
Original: http://longfellows.smugmug.com/photos/1093788386_JtwTx-O.jpg



Below: ISO 100, f/2.8, 1/4s, +4 stops
Original: http://longfellows.smugmug.com/photos/1093788355_V57Fs-O.jpg



Below: ISO 100, f/2.8, 1/4s, +4 stops, 100%
Original: http://longfellows.smugmug.com/photos/1093788310_k6Bau-O.jpg



Below: ISO 100, f/4, 1/4s, As Shot
Original: http://longfellows.smugmug.com/photos/1093788499_bguFj-O.jpg



Below: ISO 100, f/4, 1/4s, +4 stops
Original: http://longfellows.smugmug.com/photos/1093788482_thqmH-O.jpg



Below: ISO 100, f/4, 1/4s, +4 stops, 100%
Original: http://longfellows.smugmug.com/photos/1093788451_kvJVv-O.jpg



Below: ISO 100, f/5.6, 1/4s, As Shot
Original: http://longfellows.smugmug.com/photos/1093788612_n5gCR-O.jpg



Below: ISO 100, f/5.6, 1/4s, +4 stops
Original: http://longfellows.smugmug.com/photos/1093788595_hPKAA-O.jpg



Below: ISO 100, f/5.6, 1/4s, +4 stops, 100%
Original: http://longfellows.smugmug.com/photos/1093788553_hanyC-O.jpg



Below: ISO 100, f/8, 1/4s, As Shot
Original: http://longfellows.smugmug.com/photos/1093788713_p9XEA-O.jpg



Below: ISO 100, f/8, 1/4s, +4 stops
Original: http://longfellows.smugmug.com/photos/1093788685_vTy9n-O.jpg



Below: ISO 100, f/8, 1/4s, +4 stops, 100%
Original: http://longfellows.smugmug.com/photos/1093788645_5ytah-O.jpg



Below: ISO 100, f/11, 1/4s, As Shot
Original: http://longfellows.smugmug.com/photos/1093788824_VbiWH-O.jpg



Below: ISO 100, f/11, 1/4s, +4 stops, +25 fill
Original: http://longfellows.smugmug.com/photos/1093788805_nbri9-O.jpg



Below: ISO 100, f/11, 1/4s, +4 stops, +25 fill 100%
Original: http://longfellows.smugmug.com/photos/1093788760_hkmfE-O.jpg



My Conclusions:

1) Vertical Banding noise is a real possibility on a 7D.

2) It is not a typical characteristic of an image taken with reasonable technical skill by a person with reasonable expectations. (If you are looking for a point and shoot that will produce award winning photographs don’t buy a 7D)

I have to admit, I opened the f/11 file first, and when I applied the 4 stop exposure push and the 25 units of shadow fill there was a definite moment of alarm. The thought, “OMG! I’ve got one of the defective camera bodies!” did cross my mind. Vertical banding was definitely present. But after looking at what I had to do to generate it, I don’t see it as an unavoidable problem with my 7D camera body. Even when shooting candid portraits indoor, in low light, with no flash. It is not present in photos I took before I knew vertical banding even existed. Now that I know where it comes from, post processing under-exposed images, I am confident I won’t be accidentally dropping shadow detail I expect to recover later.

You can do a test similar to what I did with your own camera and judge for yourself if your camera and your technique put you at risk of having vertical banding noise ruin your photos.

For me, I’m willing to accept that when I miss an exposure by as much as it took to generate distracting amounts of vertical banding, generated in post processing without trying to fix it with noise reduction, it is not due to a fault of the Canon 7D technology.

Actually, part of the thrill of getting a really good capture is the fact that it actually does take a bit of skill on my part. Having good equipment that is capable of getting a really good capture is important and says something about the image quality I hope to achieve. But, if all it took to get a trophy Elk was to go out in the woods with an $1,800 rifle and point it in the general direction of an Elk, pretty much anyone could win a trophy. Not much of a challenge there. Not much potential for personal satisfaction in that effort. Kind of boring. Like, what if, no matter what you ate, or how you prepared your food, it all came out tasting like the same, well prepared chicken recipe? How great would that be? If that’s what photography was like I probably wouldn’t be participating. I’d be eating my meals from the McDonald’s dollar menu and every mother with a cell phone would be selling cover photos to Sports Illustrated.

As for the folks deliberately generating and circulating images with horrendous vertical banding, while claiming it is typical of what to expect from Canon sensor technology, I say this: It is difficult enough to learn how to choose and how to use the proper equipment for a given type of photography, without having someone intentionally leading folks down the wrong, or at best a limited, path with their bad information. Please, find something productive to do with your life.

PS: If anyone would like me to upload the raw images the answer is no. I have a 56k dial up connection.

 

[dsmcl77]

Thx GPapa,

I bookmark this thread to remind me later that if a picture is not of a good quality it is due to my own fault and that I should not blame the camera who's only the messenger of what the photographer is "able" to do.

Next exercise: trying to do some Vbanding with my 7D
--
Steph.
(dsmcl77)

 

[Jerry Canon]

I can see it now, some Yahoo's post: "Show Us Your Best 7D Banding Picture"!
--
Jerry Brendle

 

[DigiPops]

I bookmarked it also.

Makes me feel better already. Now, when my 7D comes back from Canon, I'll see how many shots I can ruin.

 

[Robert Krawitz]

I was also able to reproduce the banding, but it took rather extreme circumstances. For example, if I took a shot exposed -2 stops and pushed it 4, I got plenty of banding in some of the darker shadows. I wasn't worried, because it was part of a bracket sequence; when I fused the exposures, of course, enfuse didn't exactly try to use the shadows from the -2 exposure.

The common situation where I see this as most likely to be an issue is photographing sunsets where there's also a foreground with desirable detail. Something like this:



Of course, there too I used exposure bracketing, to get good foreground detail without blowing out the sky. This of course is not full size (it's about 1/4 size), but the +2 exposure of this set had all the foreground detail without having to fiddle with the curves.

The issue is not confined to ISO 100; if anything, it's worse at higher ISO, where I found a shot I took that showed banding pushed even 1 stop. It really only looks like an issue in very dense shadows. At least from what I've tried thus far, it appears to be something where proper technique pretty much makes it a non (or at worst minimal) issue under normal circumstances.

 

[Andreas Helke]

I am happy that I learned about the 7D banding problem before I had to discover it in my own camera.





This is a type of photo I often would like to take. And it is taken at ISO 100. Normally I don´t do such post processing because I know I will end up with unacceptable image quality if I try.

I own a 350D and 5D both of which have very poor image quality in the shadows.

When I did see the good image quality that the 7D produces at high ISO I had hoped that Canon finally had a camera that does a reasonable job for providing good quality photos if you need to tweak your post processing. Unfortunately i found out that it still does a very poor job at low ISO. Fortunately I found that out before I actually bought the camera thanks to people like John Sheehy.

I wonder why the camera reviewers forget to mention such problems. I think its unlikely that they don´t discover them. More likely is that they don´t want to mention them prominently because they don´t matter for the majority of photos that don´t need extreme post processing adjustments.





This is the look of my example above with a raw conversion that only does minor contrast tweaks.

 

[Ecoh]

"At least from what I've tried thus far, it appears to be something where proper technique pretty much makes it a non (or at worst minimal) issue under normal circumstances."

I think you make a good point. Proper technique ! Bracketing and graduated filters are still useful techniques, even with digital.

In some ways, we are witnessing the "dumbing down " of photography. Film photographers worked harder to get the photo right, with digital, some photographers expect the camera and photoshop to work miracles.

 

[Alastair Norcross]

We've heard it from you a million times already, Andreas. We feel your pain. Now, perhaps, you could join a 12-step program, and bore them all to death.
--
Alastair
http://anorcross.smugmug.com
Equipment in profile

 

[rwbaron]

underexpose the image by 4 stops and then apply +4 stops of exposure compensation in post to see the problem. Gee, why hadn't I thought of that? I'm only applying +4 stops of EC in post to PROPERLY EXPOSED images and haven't been able to see any significant problem yet.

Here's my test

7D top left, 40D top right, 50D bottom

ISO 100 RAW, 4 seconds at f5.6, tungsten lighting, processed in DPP at standard pic styles with all sharpening and NR to zero and ALO off.

As shot (100% crops)



With +4 EC in post



I can see some pattern noise in the 40D shot and it's quite obvious in the 50D but IMO there's nothing in it on the 7D file. For me this was post processing far past anything I would do to an image so I've concluded it's a complete non-issue for me. I've been shooting with the 50D for two years now and after thousands of RAW frames I've never experienced the pattern noise in one shot that I considered a keeper and post processed. The fact that the 7D is so much better is just nice insurance.

I find it interesting that there are so many people on this forum with their underware in a bundle over this issue. Maybe I don't understand their photography needs but monitors and prints have limited DR that can't show what a DSLR from 6 or 7 years ago could produce. DR isn't the end all be all just like high ISO noise isn't either. Before digital virtually all the serious landscape and wildlife photographers shot slide film with it's much more limited DR compared to negatives because they preferred the punch, sharpness and detail in addition to having an original positive for color and exposure reference. Sure, they used grad filters to hold back the sky but even then they were working with maybe an additional 2 or 3 stops over the 4 to 5 of the film.

I also believe some of this angst is based in a lack of understanding of basic photography limitations from capture to final output. You can't just point the camera at any scene without regard for the light angle and quality and expect to get a quality capture that will render well on screen or in print. I laughed out loud when one forum member (who's one of the most vocal on this issue) used the example of shooting a street scene with one side in direct sun and the other in deep shadow as support for more usable DR in a camera. IME you can expose for the highlights and then lift those shadows in post to a point where they show detail on screen but it's going to look like bad HDR so why would you.

Sure, it's nice to have 14 stops to work with but you have to choose your subject and lighting angle at capture because no matter how much DR your camera provides you still have to compress it into what a monitor or your print will render. Lifting shadows and recovering highlights excessively results in some pretty lifeless and flat looking images. Will I take more usable DR in my next camera? Certainly, but not at the expense of resolution or color accuracy or sharpness at the pixel level. It all has to balance.

Someday photography will provide the range of the human eye from capture to presentation but we're a long way off at this time. In the meantime people need to learn to work with the quality and angle of lighting which is what photography is all about.

Bob
--
http://www.pbase.com/rwbaron

 

[rwbaron]

In some ways, we are witnessing the "dumbing down " of photography. Film photographers worked harder to get the photo right, with digital, some photographers expect the camera and photoshop to work miracles.

Can we make the paragraph above a "sticky" at the top of this forum? It says it all.

Bob
--
http://www.pbase.com/rwbaron

 

[birdbrain]

--
Phil

A bear, however hard he tries, grows tubby without exercise.
PB

Did you ever stop to think, and forget to start again?
PB

 

[Clint Dunn]

So...you overexpose the photo by 4 stops in post and you see banding???

Here's a thought...expose your photo properly in camera. Here's another thought, take multiple exposures and do blending if you don't want/can't use split neutral density filters.

Retarded post.

 

[Pix538]

LOL you made my day with your comment

 

[dsmcl77]

Yep! my 7D definitly got banding I might just go for Nikon...unless CANON CAN at least make my shirt appear Ironed :0



I know,bad joke. There are some days like that where i should probably stay in bed.
--
Steph.
(dsmcl77)

 

[Jerry-astro]

I think you missed his entire point, which was that the whole banding thing is overplayed. Yes, it happens when you ridiculously push an image. However, he clearly stated that he didn't see it as an issue and that its importance is seem mostly by rabid pixel peepers rather than anyone doing real photography.

That's been my contention all along, coupled with the fact that any banding that might show up in shadow areas is easily controlled with a good NR plug-in like DeNoise.

His post isn't "retarded"... I think it simply makes your point. Perhaps you might consider giving it a complete read before posting.

So...you overexpose the photo by 4 stops in post and you see banding???

Here's a thought...expose your photo properly in camera. Here's another thought, take multiple exposures and do blending if you don't want/can't use split neutral density filters.

Retarded post.

 

[Robert Krawitz]

What's unfortunate about that is that with exposure bracketing I think you could easily have gotten three or four shots that could have been merged into that final shot with a lot less effort (and certainly without angst over camera limitations).

 

[KariP]

I took a typical "problem" image - bright light in the middle and different shadows in the background - partially totally black

I can find some vertical banding = pattern noise. From the totally deep shadow - when i use Aperture totally wrong . Exposure slider to the right and shadows slider to the right . And really! there is vertical pattern noise hidden in the deepest blackness.

The same ISO 100 RAW file treated with DPP 3.9.2 does NOT show any banding - or perhaps a hint.

Pushing a deeply underexposured area some +5 steps really produces visible pattern noise with Aperture 3 and almost zero with DPP.

Anyone can repeat this easily at home with DPP 3.9 and some other software at hand .

It IS NOT the sensor - it is the software + new sensor combination.

I checked my old 40D RAW files with ISO 3200 and some banding - with the latest DPP the banding is gone. I really have the same original RAW files on my HD . Where did the noise go?

So - please - gentlemen start your cameras and check your software !

--
Kari
SLR photography started in 1968, 40D since 2007, and now 7D !
60.21 N 24.86 E

 

[glanglois]

--
Phil

A bear, however hard he tries, grows tubby without exercise.
PB

Did you ever stop to think, and forget to start again?
PB

Yeah, but my K-5 has prettier radial banding than your radial banding. So there!

Oh, right. I don't have a K-5. Just got carried away. Never mind .......

 

[Robert Krawitz]

I took a typical "problem" image - bright light in the middle and different shadows in the background - partially totally black

I can find some vertical banding = pattern noise. From the totally deep shadow - when i use Aperture totally wrong . Exposure slider to the right and shadows slider to the right . And really! there is vertical pattern noise hidden in the deepest blackness.

The same ISO 100 RAW file treated with DPP 3.9.2 does NOT show any banding - or perhaps a hint.

Pushing a deeply underexposured area some +5 steps really produces visible pattern noise with Aperture 3 and almost zero with DPP.

Anyone can repeat this easily at home with DPP 3.9 and some other software at hand .

It IS NOT the sensor - it is the software + new sensor combination.

I would argue that it's a sensor problem that's being corrected for in some way by DPP. I can reproduce it using ufraw (on Linux), which is based on the dcraw code base; it doesn't have any special code to handle banding that I know of. I don't have either Windows or a Mac handy; I guess I could try installing DPP under Wine, but I don't know that it's worth the effort.

The other thing you might want to check is whether you're losing any image detail from noise reduction with DPP. That might help settle the question of whether DPP is aware of the effect and takes special steps to deal with it, or whether it's simply using stronger (but generic) noise reduction.