K20D vertical banding

I did a series of long exposures today. In total 16 shots, 4 of those show the banding again.
The exposure times were 3, 6 and 15 s when it showed (one shot each at 3 and 6 s, 2 at 15 s). All clean shots showed different exposure times (between 2 and 30 s).
Does the exposure time play a role? I don't know and one argument against it is that in the starting post of this thread I had two pics taken with the same settings, one showing banding and the other not...

Maybe time I contact Pentax Germany...

Prieni

Might be a good idea they could have some info about it.

Just crossed my mind does the Samsung GX20 suffer from this banding:

I don't think it does. I've worked out that over 20% of my shots have vertical bandings. And from a quick look I can't find one with an exposure time more than 1/15sec. i have had it in speeds of 250 as well.
Like you I have also taken 2 shots, same settings within 20 sec of each other and 1 was ok & the other had bandings.Tis very strange.

I have never seen banding on my K20D.
I have reviewed a lot of pictures, but I couldn see any banding.

It seems to be effecting only some, I like you have checked mine and found nothing.



Edit Spelling

Did you review the pics at 400%? In the first example (i.e. first post of this thread) it was noticeable at 100% (if only just).

For the last series it is different: If I look at 100% I cannot see it, even if I know it's there. At 200% I can see it, but only if I know it's there. It's only at 400% that I see it.
(I'm using the 400% magnification to correct for CA in SilkyPix, not to specifically screen for the banding).

This fact, of course, does also put things in proportion. Yes, the banding shouldn't be there in the first place. But even if it does occurr it doesn't usually ruin your shot.

Prieni

P.S. Since there are many Brits in this forum you should state the fuel consumption in MPG as well...

Edit: Grammar

Some of my worse examples can be seen at about 100%


This is the original

Steve,

I don't contest that. In fact the first post of this thread does show an example where I could see it at 100%.

Not only does the banding occurr irregular (so far we weren't able to tie certain settings or conditions to it) it also does occurr with varying intensity...

Prieni

I've seen the same. Banding definitely there, but only really noticeable at 300/400% view. Some shots have it all over at 100% though. To some extent it seems to be how much contrast is in the shot. The worst I have so far is a bright shot over water. It was underexposed and after adjusting in the RAW converter it had really marked banding all over visible at 100%.
I wonder how many people have the problem but haven't actually noticed it. Not everyone looks at a lot of their shots at 100% plus!

Shutter speed doesn't seem to affect it. I shoot a lot under halogen lights with relatively long exposures and it's no more prevalent then as far as I can tell. This may be because the contrast is generally lower in these type of shots though. Like others I have similar shots taken a few seconds apart one of which is affected and the other not. It usually seems to be the second or third shot.
It's very strange!

I wonder if some of these cameras have components which are close to the edge of their tolerances. Several components like this could compound the error and a tiny change in temperature or voltage could cause problems. Just my guess though.

I'm holding fire on sending mine back just for the moment until Pothers hears back from Pentax. I expect it'll have to go back eventually, but if they genuinely don't know about the problem then it may be as simple as a firmware fix. We can but hope!

It's just a hardware problem and there is nothing you can do against it.
Banding noise is most visible at high ISO speeds and in the shadows, or when an image has been excessively brightened. Banding noise can also increase for certain white balances, depending on camera model.
So check the white balance and see if anything has to do with it.

To soften the pain all the other manufacturers cameras also suffer.
The problems with the Nikon D300 were fixed with and firmware upgrade so maybe there might come a solution in the form of that

Canon also had the banding with the 1D, 1Ds and Canon used a firmware upgrade to solve it.

Since my initial comments a week or so ago I have been conducting tests with my Samsung GX-20 and now have a method where I can repeatedly create images showing the various faults: fine vertical banding, horizontal/vertical regional banding and posterisation.

1. Cool the camera, preferrably to 10C or less.
2. Shoot a test image that helps you see image problems, lit with tungsten light.
3. Expose for >60 seconds (I ended up at 75s).
4. Allow time between exposures to allow for the dark frame capture and to allow the electronics some time to cool down (I allowed about 7 minutes).

Since the Samsung has no PC remote control software, I built an intervalometer to get the camera to snap test pictures through the night at a regular exposure/interval.

With the above settings, the majority of pictures shot below 10C have some sort of issue, and then the probability/severity of failure decreases as the camera temperature increases. However I have had some faulty images shot above 20C. Note the PhotoME camera temperature could be different to that of the faulty circuit, and there will be some camera electronics self heating going on.

With exposures at 30 seconds, failures are much rarer. I think in some there is a hint of vertical banding but its magnitude is much lower and not troublesome in most images. However I did spot colour inconsistances at this exposure which may indicate the underlying fault is still at work.

Below is an early test image I snapped and developed by Lightroom using its defaults, using 17C and 120s exposure.



You can clearly see the horizontal regional banding across the lower third of the image. Look closely and you can also see at the bottom right a vertical region band interacting with the horizontal band.

And a 100% crop:



There you can also see the break up of detail in the shadow.

I have not had time yet to post more general images and conclusions from my test. I have a few hundred failure images now to sort through. However I do know that:

- Lower camera temperature increases fault probability.
- Long exposures will commonly result in faults.
- Sometime the fault is minor, but can result in unuseable pictures in real conditions (see my original samples images on Flikr).
- More likely to be spotted in darker exposure and tungsten lighting when the pixel's RGB data values are small in some colour channels.

Paul
Samples: link

Is the fault repeatable with noise reduction switched off?

Good work, mera!
The example does look very bad indeed and goes to show that the problem is not purely academic.
Your example looks a bit like the first time I encountered this oddity. (See this thread)

John, that's a very good question. mera, more work for you (give us the little finger...)

Prieni

I don't think you can disable noise reduction on a K20D/GX-20. Dark frame subtraction is always employed in longer exposures. Note High Noise reduction was off in all tests. I also did not see any evidence that SR on/off impacts the results.

I hope these tests don't worry the paranoid, I am sure this is a relatively rare fault in an otherwise superb camera. But if you suspect the camera then you may be able to try the tests to confirm if the camera is ok or not, and if sending it back hopefully help Pentax reproduce the fault and fix it. I also suggest reading the vertical banding thread on pentaxforums.com - that is where I learned that long exposures can trigger the fault repeatedly.

Any further tests will have to wait until after the weekend - I have a date with Scottish weather. It will the the DS I will be taking on that trip - it loves the snow.