Sunday, March 18, 2007

VPN and HPN / Bandings of the K10D

Here is an interesting recent experiment carried out by a Pentax K10D user who attaches the vertical battery grip D-BG2 on his K10D:-

It is found that in addition to the "usual" Vertical Pattern Noise (VPN) , there is also Horizontal Pattern Noise (HPN), which the author has verified in his experiment and that it is confirmed that the HPN is simply caused by the battery grip when it is turned on.

Whilst people recently like to use the term VPN when they referred to the problem, previously the term "banding" was more commonly used when this was referred. Now I would refer both the VPN and HPN collectively as bandings, no matter they are vertical or honizontal ones.

In this article, I shall briefly look into the technical reasons behind for why there are the bandings, and, could firmware cures this problem or not, fully or partially.

Before going on, it would be good to see one more recent illustration of the VPN and HPN problem:-

VPN has a nasty friend!

In January, Pentax released a new firmware version of 1.10 which they claimed that "In rare case, vertical stripe at long time exposure." as one of the two corrections contained in the firmware. Pentax did acknowledge the problem and they did take the measure, which is a good thing, IMHO. However, the attainable correction amount purely by software (firmware) and the effectiveness of this solution is yet unknown, and actually is questionable, I would say.

To know for a *brief* explanation of such banding, here is the offical answer from Nikon, for the cause of vertical banding seen in some of their D200 units. But as we can see, Nikon offers free adjustments for users who found the problem and they are required to take back their D200 back to Nikon for servicing. Thus, this would imply it is a hardware adjustment or at least a low level firmware parameter setting by some kinds of "factory" calibration which is something impossible to be correctable at the user level, including a simple firmware update.

For comparatively more technical explanation, here is what Ken Rockwell says, which I think is clear and correct. To supplement what he says more in depth technically, we can have a look on how the CCD imager in our "CCD DSLRs" gets the image data and how it handles the signals for the image, at low level. Let's read back my earlier article about this:-

The Secrets of the K10D (Part 1 of 3) - The Heart: The CCD Imager

Unlike CMOS, any CCD needs to clock out chains, further by lines, of image data by channels (same as chains). So, each "clocked" line and *hardware* channel will have individual differences, as it is impossible they can be made identically. Particularly, actually the ADC need to determine the black level too for each clocked line but this black level value determinated is a variable. Especially when in extreme cases, the black level obtained can frustrate largely and it cannot be determined in an accurate way. There are a number of good technical reasons for how comes the errors, one of them is the "Smearing" effect, which you can find a good explanation in the following lengthy NuCore presentation, along with quite some other explanation on the causes and their cures of other digital image artifacts (Warning: Very Large PDF File!):-

So, banding is actually caused by the inherit limitation of how CCD works. To deal with this problem, different ADCs can have different *hardware* measure(s), depending on different design, which is then could or could not be catered with by the firmware. But undoubtedly, precise and accurate calibration is required so as to eliminate the effect as far as possible. Also, if the hardware and software are intelligent enough, the undesirable effect can be eliminated further.

Do note signal noise, no matter it is from inside of the CCD and/or from the ADC, or, from external interferences by other hardware, is the main culprit for causing *any* banding. So, it could be easily explainable for why when the user, whose experiment I have quoted in the beginning of this article, found that horizontal banding is caused by the switching-on of his D-BG2 battery grip. It is simply because the battery grip just cause some kinds of interference to the CCD/ADC circutiry which picks up the noise and finally shows up as horizontal banding.

To be fairly say about the comparison for CCD Vs CMOS, whilst CMOS imagers do not suffer from the banding problem, as the image data are picked out per pixel, they do suffer from the random noise which caused by the individual hardware difference per pixel! Yes, the difference is by pixel, not even by channel or by line! But if this could be eliminated and minimised by proper filtering and image processing means, which Canon succeeded to deal with this hardware limitation with their DIGIC processors and associated correction circuitries.

Also, just for the sake of a more balanced view and as a final remark of mine, I guess those bandings problem would be a non-issue for most of practical shooting cases, as it happens only when very high ISOs are chosen, say, 800 or 1600, and then when the user (or the camera) underexposes the scene, *and then* the user chose to push up the exposure by at least 2 stops, or more. THEN, you can see the bandings, no matter they are vertical and/or horizontal ones.

So, next time when you see minor bandings when you shoot under extreme situations and post processing in an extreme way, you should regard this as normal and should be acceptable, especially for a CCD camera. But when you see obvious bandings in quite normal shooting situations when you just use your camera normally at low light, maybe with just higher ISO without underexposing and/or push processing, I am afraid you have a poorly calibrated camera, no matter for the hardware and/or the hidden firmware parameters, then in such case, you should get your camera back to Pentax for a proper calibration and re-adjustment.

I hope my this short conclusion and recommendation can help all those users who frequently asked about the VPN, HPN, VB or HB questions from day to day, no matter which CCD DSLR model(s) you own, a Pentax one or not! :-)

Related Posts

Creative Commons License
RiceHigh's Pentax Blog by RiceHigh is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.