K5 image quality concerns - thoughts please

Mike, I don't know if this is relevant to the problem, but the exif data for your files (revealed by the exif reader FileTagSleuth) contain references to the Pentax-FA 43mm Limited lens?? E.g. -



Philip

It's only a 1/3 of a stop though. I know that's not insignificant, but the obvious issue with the image in question is too much Noise Reduction, rather than too much noise.

Look at the Original size image in flickr, it's steeped in noise reduction that severalty smears the image.

Something isn't sitting right here.

If you're going to complain about an image with too much noise then show us the image with that noise rather than showing us the image with an over the top amount of noise reduction.

This image was described as "usual minimal tweaking in lightroom 3". There's no way on earth that such a huge amount of Noise reduction could be classed as "minimal tweaking"

Even at -0.3EV and @ ISO3200 there's absolutely no rational reason to apply that much NR in PP for a K-5 shot.

Ken you obviously remember my Rasta Alien shot at ISO12800. I used far less NR on that to end up with a clean image.

I'll take you up on that if I may smeggy - I'll take alook tonight. I have played with it some more and it does now look cleaner.. though not sure if it might look a little less "crsip" or razor sharp.. have another look via the link folks - just the first one - and see what you now think


thanks


Mike

that's odd - I defo selected sigma 18-125mm from lens profile drop down

ah the NR that you now see was done after my original post and after anyone who looked at it early doors.

I'll take it back to where it was and re -export to flickr

OK it's back to how it was imported into LR on flickr now

The level of noise in your first image does seem a bit higher than I would expect from ISO 3200, but no more so than would be explained by underexposing and then raising the exposure in Lightroom later. When you raise the exposure in PP any noise present is amplified as a result.

As Smeggy says, -1/3EV is not that much but the issue is more how much you raised the exposure by in PP than how much exposure compensation you used when shooting the image. I might be wrong but there is quite a large area in this image covered by white tiles which may have caused the camera to underexpose the image in the first place, forcing you to use more than +1/3 EV to bring it back. As I say I could be wrong but it wouldn't hurt to take a look in Lightroom and see how much you raised the exposure (or brightness, or fill light - any tool that makes the image lighter has the same effect).

The K-5 rightly has a reputation for high ISO / low noise performance because it excels in this area in comparison to its peers. That doesn't mean it's somehow magically noise free at higher ISO levels. Having said that, effective use of the noise reduction tool in Lightroom can give pretty clean images at ISOs up to 8000 or so, albeit with a little loss of detail.

Have a look at the results here for an idea of what level of noise you should expect at different ISOs.

That is really under exposed and by far more than a 1/3 of a stop.


My take on tweaking your JPG ...

Hi MIke

I tend towards the under exposed argument as well - I recently did a set of landscapes on the K5 and over did the filters and under-exposed - noisiest set of images I have ever produced.

That said - LR4 still managed to recover a high percentage. I have just run your compressed flickr version of your image through LR4 and eveb using that version - there is a marked improvement

Carl

It is not that badly under-exposed - the jpeg is only about 1.2 EV under, there should be no problem recovering that from the K-5 RAW. There is a bit of out-of-gamut clipping on the hair, otherwise a perfectly workable image.

Throwing the jpeg into SilkyPix Pro gives this :

smeggy have sent you the raw .. be interested to see what you can do with it

I would say the jpeg is indeed underexposed, and has probably been boosted in exposure already? Which would make the effective ISO much higher than indicated. In any event there are really bad noise shadows on the forarms which are hard to remove - the rest of it cleans up pretty easily - even with my tools that are RAW orientated and sometimes less than outstanding on jpeg.

My take using darktable: