[vincechu]

I've done a lot of testing at high ISO to see if I can get the banding to occur more by shooting different scenes, and found the following.

Funnily enough it seems present most present in the photos I uploaded the other day to the thread, subsequent shooting with all NR set to off, seems to be much better. So perhaps it is the NR causing the banding?

The banding, more specifically the noise, only seems to occur in some shadows, and in other cases the noise does seem to occur in patches and pretty random and in some cases the banding isn't even there (e.g. in one photo of the night sky i snapped). As other users have pointed out to me, its easily sorted out in LR3 using the colour noise reduction slider

EDIT: hmmm the more i test, the more I think its me making a bigger deal out of this than it is lol. On my k5's screen the banding looks quite obvious and horrendous lol, but on my pc the images (DNG) do look much better, at times to the point that banding isn't not visible! guess its just me afterall lol?

[rawr]

Banding is not one of the issues K5 users have ever reported.

Either way, as with all of these discussions, I think you need to test it out on a more controlled way. Just shoot a blank dull grey surface if you specifically want to search for banding. Put your camera on a tripod, then shoot the test scenes under poor, steadily decreasing light and set the ISO progressively higher all the way through the range you want to test. Also shoot at RAW+ so that you have out of camera JPG and RAW files for comparisons.

But what you have presented seems like a JPG processing/NR issue, as you say.

[WerTicus]

I'm not seeing any banding with my K-5 or your test pictures.

The K-5 is the most filmic digital sensor I've ever seen.

[vincechu]

Thanks guys for all the input :-)

I sent some raws to JohnBee to check out and like others have said my k5 seems fine and they don't see the banding too - I just think it was a problem caused by my poor noise reduction (inexperience), and me over examining my photos and wanting to find a problem

anyway for those interested, here's what JohnBee said:

I think you've likely encoutered that artifacts such as banding can and are often times dealt with by the RAW developer. However in other cases, its not so easy or possible and so you may notice banding appearing in certain images, and not in others at the same sensitivity.

On the other hand, the ISO51200 sample didn't exhibit any banding on my end using both RAW Therapee and Adobe Camera Raw(without NR). Though I still wouldn't recommend ISO51200 for anything but extreme measures given that the limited amount of detail available at this level.

Having said that, I did notice a significant magenta case in your table DNG. And at first glance(without looking) I'd say this was most likely due to underexposure. To which I'd add, one helpful tip I've discovered, is to lean to the right as far as possible without blowing highlights when shooting at higher sensitivities. Which helps reduce overall artifacts and more importantly... the type of problems such as the ones seen here.

Other than that, I'm happy to report that your sensor seems pretty much identical to my own in terms of high ISO performance. And so I'd say you are safe! and that there is no need to send out your unit as far as this goes.

FRT. here's some RAW Therapee/Topaz Denoise output from your ISO51200 sample. And while its not the greatest! - I'd say its good enough for emergency use if/when needed



PS. I used to shoot at max sensitivities with my K20's on a regular bassis. However, with the K-5, I've gotten into the habit of limiting my output to ISO2000. And that's not because I'm afraid of max ISO, but because I don't like the low level noise reduction that the K-5 applies to RAW files. Which kicks-in at ISO2200. And therefore, by keeping the meter bellow 2200, you can push process in post(RAW developer) and avoid the detail loss that comes with the K-5 prime engine. The downside on the other hand is that you have more noise to deal with(so its a give and take).

Anyways, based on what I've seen here, I'd say you're K-5 is in good hands! - Take care and thanks for participating.

Sincerely,
JohnBee

[WerTicus]

honestly I dont think you EVER need to use noise reduction with the K-5 because of its filmic nature, you never 'improve' the noise. At 6400 iso and up you might like to swap to black and white though :P