K3 - K5 ISO1600 Noise comparison

Here are some ISO1600 shots from both the K3 and K5, nothing too scientific just handheld so don't read anything into sharpness just a noise test.
I've included 100% crops of unprocessed DNGs, processed DNGs with +25 in lumin and colour noise reduction in LR5 and jpegs straight from camera.




K3 Jpeg


K3 DNG no processing


K3 DNG +25 noise in LR5



K5 Jpeg


K5 DNG no processing


K5 DNG +25 noise in LR5

Its an interesting comparison, thanks. Unfortunately I think the colour balance discrepency messes things up a bit - the magenta cast on the K5 image somewhat disguises the (mainly) yellow colour noise, so it is a bit difficult to assess confidently.

Can you reshoot the K5 image and try and match up the colour balance? I think then we could see a more valid comparison.

But, it does appear, even with the colour cast issue, that the K5 DNG has less colour noise 'out the box'.

Here's the K5 DNG with no noise reduction but with WB set to same values as K3
Interesting that the colours from the K3 & K5 are different even though WB are set the same in LR5

I have never seen a clear explanation of exactly what camera fine tuning settings are actually interpreted and acted on by a 'third party' editor like LR. Even assuming (as I am) that your custom image settings on both cameras are identical that doesn't mean that the 'zero points' for each preset are not different on each model. There is a 'hue' adjustment slider within the custom image menu though, so its worth checking thats not different, and of course also the main WB fine tuner as well.

But, clearly, the K5 DNG shows less obvious and finer colour noise than the K3 one. This test at 1600 is contradicted by David Storms one, where he concluded that the K3 was better up to 3200, if I remember rightly?

My own view on all this, as someone who enjoys messing around with different noise reduction software, is that its not so much the starting point with noise thats important - its where you end up after processing that matters more, and the critical part of this is the detail recovery capacity of the file following noise reduction blurring. The additional pixel data with the K3 may well prove beneficial here.

I've had a play about with the images in LR5 and come to the conclusion I'm not worried about noise up to ISO1600, you can be more aggressive with the noise reduction in LR5 on K3 DNGs without losing detail, maybe because the noise is finer grain.
I'll have to try higher ISO settings when I've got time.

Is the noise 'finer grained' because of the extra pixels? If so, then my theory that greater detail recovery is possible could be true. If so, then you can 'get away' with that more aggressive NR (as you mentioned), so maybe there's not such a disadvantage?

Hi Nigel, no this is not quite correct. The K-3 in my tests was clearly better at low ISO's, a very tiny margin better to my eyes at 1600 ISO, but the K-5iis was a lot better than the K-3 at ISO 3200.

I did also notice the different type of noise on the K-3 - it does seem to be finer grained and I would expect it to respond very well to processing, although I haven't experimented much with this yet.

Regards
David

I was thinking about what all this means in the real world. One thing that is more obvious is that we now have cameras, at greater resolution, producing an ISO performance equivalence of at least one stop better than we did not so many years ago, eg K7, K-r era. Surely, given equal optical lens performance, this lowers the need to rush out and buy fast, heavy, expensive f2.8 telephotos for example?
Another interesting comment often used by myself and others is "I never tend to shoot above ISO X00 anyway" (insert as appropriate). While this is true I suspect it is the subconscious knowledge of knowing when noise starts to creep in and have an effect and this value quoted will become higher in the future!

You may very well be right.

Noise per se is not the problem though, any more than grain was with film. As long as the type of noise and the effect it produces is something we like, then it's fine.

Shooting JPEG as I do, I don't have a problem with noise and just whack that ISO up to whatever is necessary. It's much better than old ASA 1000 films ever were.

Incidentally, if noise does look off then sometimes converting to black and white helps.

i have got to say here i really dont know what all the fuss is about this shooting at iso 1600 / 3200 I know no one who shoots at the rate of iso ??? or I have used it that high? I also can say that I don’t know any photographers who do shot at 1600 iso or higher.
i would be more interested at looking at the focusing and speed then the iso ?
So my question is who truthfully shoots at 3200 or high

Sometimes, I do Richard. I don't worry about the ISO and use it as necessary at whatever speed is necessary.

All features are important. I haven't noticed anyone complimenting Pentax on changing the frame rates though. There were plenty of grumbles when it was 1.6fps at low speed!

I do.

As Smeggy says, whatever ISO it takes to get the shot.

The only reason anyone has an upper ISO limit is because they don't like the results, or never shoot in low light.

If there was no noise at high ISO, what possible reason would there be not to use it?

Edit:

12800 on the K-5 is perfectly usable where needed:

[IMG]http://i162.photobucket.com/albums/t244/chris5gd/Technical/IMGP0198-ff-br_zps567fbcb3.jpg[/IMG]

Resized to 800px, so of course much of the noise has gone. Even quite a bit bigger and it only takes a little NR to clean it up quite nicely.

Here's a 100% crop of a bit of it. A little chroma noise removed in ACR but no luminance NR at all:

[IMG]http://i162.photobucket.com/albums/t244/chris5gd/Technical/IMGP0198-100_zps117b1ff1.jpg[/IMG]

However, I'd have liked to go higher still and used a shorter exposure ideally.

I routinely go over 1600
In film days I chose high iso (asa?) precisely for the grain.
Since the K7 I have loved the film-like quality of Pentax noise. I very rarely try to suppress it, and turn the image to sludge.
My last (all day) event shoot at Borringdon (Tudor) Hall, was all shot at 1600 and above.
I did check with the client that she wasn't going to try and produce huge posters from the images tho It was DVD artwork and online promos so no problems at all.

I'll shoot at whatever ISO it needs too

We had a night shoot at my local photoclub earlier in the week. I turned up late as usual sans tripod. Still just for giggles I was showing off the K-5 + prime combo (FA31 in this instance) because I could

K-5, 31 @ f/1.8, 1/40s, ISO20000 Handheld.



95% of the other members are zoom users with various opposition models, D5100, D5200, D7000, D7100, D3X, 600D, 700D, 7D and i'm slowly but surely building the Pentax reputation as the night shooters camera and the discrete camera and the beautifully handling camera and last but not least as the camera with the most lens
If fact one of the new members borrowed one of my K-5's + a DA35 f/2.4 and is this evening becoming a fully fledged Pentaxian and that is because short of a FF camera costing €000s, Pentax own the high iso world.

I am holding off shelling out on the K-3 purely because I think the K-5II is better at >iso6400 and I personally spend a lot of time in the region of >1600 and