Hey all -
It looks like we may have an easily-reproducible banding issue with the 20D and a somewhat uncommon method of using the Internal Flash with long exposures. It would be great to determine if this is just a quirk with the 20D, or a problem with individual cameras.

Please try the following:

1) Engage internal flash by pressing the Flash button on the side of the lens mount.
2) Set Custom Function 15 to "1:2nd-curtain sync."
3) Set camera to manual.
4) Set ISO to 100
5) Set shutter speed to 30 seconds
6) Set aperture so that the EC meter reads about 1 stop down - you may need to find a relatively dark room to make this work.
7) Take a photo (will probably need a tripod or set on a table, etc.)

Look at the photo full-rez and tell us if you see any banding similar to this:



Thanks!

Sam

I've replied in the other thread - but Yes (unfortunately) I do get banding.

Jan

JVasek wrote:
Sam

I've replied in the other thread - but Yes (unfortunately) I do get banding.

Jan

Jan - Were you sure to set the Custom Function to fire the Flash on 2nd curtain? Thanks for checking it out!

Didn't even have to look at it at 100% to see the banding.

Posted here: http://www.pbase.com/olgaj/image/36470871

Olga

Yes 2nd curtain sync enabled as per your instructions.

Here is a 100% crop of the effect I am seeing.

Jan

Yes, it's there.
Very similar to your banding example with change in 'frequency' of the banding somewhere halfway in the picture.
Banding is horizontal.

This is a setup I would probably not have thought of trying in real live, but nevertheless it looks like we have a (minor) issue here. Won't stop me shooting though.

ekrz wrote:
This is a setup I would probably not have thought of trying in real live, but nevertheless it looks like we have a (minor) issue here. Won't stop me shooting though.


Thanks checking! Yes, I agree it's not a problem many should run into. However, it's indicative of a larger problem which some suspect has to do with power regulation in the camera. This same problem can be experienced in much more common situations - I see it when use my 50MM f/1.4 with AI Servo on tracking shots. In that situation, about 75% of my shots were effected, although the banding was more isolated. Luckily it doesn't seem to be an issue with my other lenses, so I can work around it.

My hope is that the identification of this very reproduceable problem will help Canon fix the larger issue. However, I suspect they already know about the issue and have shrugged it off due to cost constraints. Is worth more to them to fix it or to deal with a handful of people complaining about it? I think we know the answer.

I get a completely different result from using the 580EX in lieu of the built in flash. No banding.

Sam,

Getting these banding effects by using AI Servo plus a specific lens is what I would call a more serious problem. I will now definitely start checking this with my lenses.

Is there an increased banding effect using specific settings (like high ISO values) ?
Can you share an example of the effect with AI Servo / 50 1.4 ?

Thanks for starting the topic.

ekrz wrote:
Is there an increased banding effect using specific settings (like high ISO values) ?
Can you share an example of the effect with AI Servo / 50 1.4 ?


The banding I see with AI Servo is usually limited to one area of the image - I usually only see it in the first or last 1/4 of an image on the top or bottom, although sometime it will cover more or less. Typically I see it with AI Servo where the lens is hunting - low contrast areas on static images, or tracking shots where the subject is moving towards or away from you. It doesn't appear to be limited to any ISO - I've seen it on images down to at least ISO 400, but it is more noticeable on higher ISO images.

Here's a controlled shot that's had curves adjusted to make the problem more noticeable:

"Normal" banding:


"Interference" Banding:


Here's a more practical example. After shooting this event, I noticed that aboput 75% of my images had the banding in them - some with banding covering about 1/4 of the frame to up to almost 1/2 the frame. Most of these were shot at ISO 1600:

No banding:


Exact same settings, shot seconds before, with "interference" banding:


This is the problem I'm most worried about. Luckily this problem only seems to occur with my 50MM f/1.4 and only with AI Servo - it is easy enough for me to use my 85MM f/1.8 with this sort of event instead. Unfortunately a user on DPR reported this same problem with the 135MM f/2L on a 10D and I've just ordered one to use with my 20D on this sort of event, which will require AI Servo - I'm keeping my fingers crossed.

Have you tried it with the BG-E2? If it's caused by power drain, maybe the extra capacity will help.

Access wrote:
Have you tried it with the BG-E2? If it's caused by power drain, maybe the extra capacity will help.

I haven't - I don't have one. I would also like to try using a power adapter instead of a battery as well, for the same reason. That's one of the reasons I'm trying to get others involved.

Zero banding for me. I zoomed in as far as I could possibly go and saw nothing but clearness.

I'm seeing banding as well, but under different circumstances. I'm thinking it's related, at least in my case, to the long exposure noise reduction that the 20D does for exposures of 1 sec or longer. I want to experiment some more...turn it off and try a long exposure. Here are two crops from a sunset session last week.

Both were taken with the 17-40 at ISO 400. The first is a 0.5 second exposure...no noise reduction and no banding.

I am seeing it in mine as well, and its even visible in the tiny LCD when zoomed to the max. I thought nothing of it when I first saw it and thought it was only my eyes playing tricks on me. But it should not be a problem unless one wants to do some really large blow-ups, right?

Cheers!

Can you guys do some long exposures with lens cap on and post some pictures without using flash?

Does this problem only occur when you use flash, internal or external?

ok guys, just did these tests with my two week old 20d. shot with no flash and with flash. 5 second exposures or longer.

no banding


shot with lens cap on, found a red pixel for 30 seconds, but no banding

shot with lens cap on and with flash, same red pixel (@#$!!) but no banding


take/send your cameras back they are defective.

edit: damn spoke too soon. just took another one 13 seconds with internal flash on - banding.

ah well.i never use internal flash anyway and i think i've done a faster than 5 second shot with digital cameras 6 times total in as many years.

but for those that enjoy long exposures this sucks.

Yeah, I have never seen banding if I kept the lens cap on. I've only seen banding in areas of the picture that were not completely black. Most of the banding shows up in the darker regions, but I've never seen it in a completely black area before.

Also, I wonder if everyone experiences banding at ISO 3200 as well but just doesn't realize because they never use it, or have not taken an iso3200 picture with dark areas in it.

I would never have noticed banding at 3200 if I had only taken a few shots. But after taking about 50 around the house last night, I got banding in close to half of them.

I'd like to send my camera in if it's something that can be fixed, but I get the feeling all 20d's show this characteristic.

jdee wrote:

Also, I wonder if everyone experiences banding at ISO 3200 as well but just doesn't realize because they never use it, or have not taken an iso3200 picture with dark areas in it.


I shot nothing but ISO 3200 at a rodeo one night and didn't see any banding, but my shots were at 1/640s. No long exposures and no internal flash.

Again, there is normal noise banding - which pretty much anyone will see on any dSLR when shooting high ISO, pushing exposures in RAW or doing intense curve adjustments - and then there is the problem we're talking about which I'm going to continue referring to as "interference banding" (unless someone's got a better term) which is very abnormal and only seems to occur with certain lenses (usually with difficult AI Servo shots) or when using the internal Flash with long exposures. Interference banding is independent of ISO, although it will be more noticeable with higher ISOs.