[D1N0]

They haven't stopped their view that the noise reduction chip is an evil detail consuming abomination naming it a con. I find the level of incompetence of some of their staff disconcerting. They are doing real damage to the Pentax brand.

[clickclick]

And lets not forget that this kicks in at ISO 640, if I remember correctly (straighten me out if wrong on that). I pretty much make a point of staying below ISO 800 for the types of shots I want fine detail in - 200 is often my max for these scenarios. I know this might not fit for astrophotography, but if you're shooting flower bokeh, you shouldn't be up in the range where the accelerator chip kicks in anyway. The one combo where I use the higher ISO and want detail is birding, and I get some wanting as much detail as possible, but there's a point where it gets down to getting the shot, and I'm sensing the mark II gives you that headroom to get the shutter speed up for those shots - I'm typically shooting in TAv and keep it at 3200 max on my mark I in these scenarios. Upping that to 6400 would be pretty nice, and I sense that might be realistic with the mark II. And again, is 6400 on the mark I with time spent in post doing noise reduction really going to be better than what the mark II gives me to start with?

[leekil]

That in itself is quite interesting. I know we're discussing the K-1II, but given the accuracy of AF with Sony's mirrorless cameras, you have to wonder how it mis-focused (if that is indeed the case). Did the Sony show that it is, in fact, fallible too - or was this another error on DPR's part? With on-sensor PDAF points, it can't be an issue with AF fine adjustment...
.

It looks like it could be an error -- compare to the A7R II -- but for those who are more prone to detect a bias, the A7R III review identified the III as sharper than the II, which is not the case in the high-detail areas.

A7R III image comparison

[normhead]

And yes, it's possible with a bunch of post processing, one could do the same job at getting rid of a lot of noise, and maybe not lose the detail.

I'd be really interested in seeing even one example of that being possible.

[biz-engineer]

i'm amazed to see how much discussion was going on about the details vs noise discussion of the Pentax K1 II, given that the K1 and K1 II largely outperform most cameras and smarphones seen in the field, the only next step in image quality beyond Pentax K1/D850/5DSr being in medium format territory which isn't so common.

[MatKus]

Normhead: It's not impossible for sure. You could do this even in Paint if you have enough free months to do this. Just who have time?

I don't care much about noise reduction in camera, if it is done right, but still I'd like to have choice to turn it off in those rare situations when pixel peeping details are so important.

[Mikesul]

I stopped using the DPreview comparison site years ago. The pictures they posted for most of the cameras I was interested in just did not seem right and when I go the cameras I confirmed they they were almost always better than the dpreview samples..

[BigMackCam]

I'd be really interested in seeing even one example of that being possible.

Well, I've been playing around with the ISO 6400 shots in Darktable, and in all honesty, the K-1 cleans up slightly better in post-processing than the K-1II straight out of camera.

With all noise reduction disabled for the K-1II image, I processed the K-1 image to reach approximately the same level of luminance noise reduction, and actually slightly better colour noise reduction. There's not a huge amount in it, but for me, the processed K-1 result is better so far as detail is concerned. Look at the body and arm of the magenta coat - this is where it's most noticeable - and to a lesser extent the red stripes on the shirt and trousers. The K-1II shows slightly blotchy / lost detail here, whereas the K-1 image with luminance and colour noise reduction from Darktable's Equalizer module (using custom settings - not the presets) retains the detail a little better. It's minimal, but it's there.

I think the K-1II is doing a remarkable job for in-camera noise reduction, and if that were on/off switchable or adjustable, it'd be awesome. For those who are prepared to spend time in post-processing, though, I'd say it's possible to squeeze just the tiniest bit more detail out of the original K-1's high ISO files...

The first image below is from the K-1II, and the second is from the K-1 with my processing applied.

[house]

I was critical of the mark II based on the initial shoot but now the differences are so marginal I can't even make my mind up which is better. The big deal is, as always, how incredible well Pentax compares to the top of the line from the other brands. When price and pixelshift is taken into account it gets silly. Sony pixelshift looks pretty poor in comparison to Pentax.

[ffking]

What's weird in the samples leekil posted above is that the two Pentaxes render the scene very similarly but the Sont, and especially the Nikon seem t bring out very different details- not more or less necessarily, just different

[mee]

It's more noticeable at higher ISOs... Here, at ISO 6400, you can see a bit of detail loss in the magenta and red woven material (especially the arm and body of the magenta coat). But this is a pixel-peeping 100% comparison, and the screenshot is enlarged for reproduction here...

'a bit' of detail loss? It looks totally mushy on the K-1 II shot to me. To the point I'd go with the original K-1 between the two. I can always run NR on the desktop. In those kind of situations I'd apply heavier NR to background or solid elements and much less on high detail objects (such as the patch). I'm not sure the K-1 II handles that elegantly...



They could resolve this entirely if they had a way to disable the baked in NR somehow.. I really hope the K-3 II successor has that since I'm pretty sure it will have some flavor of this chip built in.

[monochrome]

I was critical of the mark II based on the initial shoot but now the differences are so marginal I can't even make my mind up which is better. The big deal is, as always, how incredible well Pentax compares to the top of the line from the other brands. When price and pixelshift is taken into account it gets silly. Sony pixelshift looks pretty poor in comparison to Pentax.

The logical question, then, must be if the IQ improvement at normal K-1 shooting levels up to 12800 is only marginal - at best - are HHPS and as yet unproven AF improvements worth $500?

I’ve agreed with DPR conclusion and Cons right along. I abhor their sloppy work and snarky tone.

[BigMackCam]

'a bit' of detail loss? It looks totally mushy on the K-1 II shot to me.

With respect, I think that's an exaggeration... But then, it's highly subjective, and our perspectives are different. Way back when, you described one of my shots as being "mushy" when it had a level of post-processing noise reduction that I personally was very happy with. The same applies here... the noise reduction will please some, disappoint others, and the rest won't care.

That said, I've changed my mind on the basis of my post-processing tests, and would probably go with the K-1 also for IQ alone... though I'd be very happy with either camera, and of course the K-1II offers additional advantages.

They could resolve this entirely if they had a way to disable the baked in NR somehow.. I really hope the K-3 II successor has that since I'm pretty sure it will have some flavor of this chip built in.

What's interesting is, almost nobody complained about this with the KP, but if you compare DPR's high ISO shots for the K-3 / K-3II and the KP, the exact same difference in detail (I call it "a bit", you call it "totally mushy") can be observed.

In all the time the KP has been around, I can only think of one member who decided the KP wasn't for him because of this noise reduction, and - perhaps coincidentally, perhaps not - his discovery was made right after all the discussions about the K-1II's noise reduction kicked off.

Aside from all that, like you I hope that any future Pentax cameras allow for this NR to be switched on and off, or - better still - configurable.

[D1N0]

What's weird in the samples leekil posted above is that the two Pentaxes render the scene very similarly but the Sont, and especially the Nikon seem t bring out very different details- not more or less necessarily, just different

Leekil didn't use comparative size. Since the Nikon and the Sony have more megapixels 1:1 is more zoomed in.

[rawr]

Looking at comparisons between bodies that use the same 36MP sensor shows that the Accelerator seems to concentrate on removing chroma noise.



These sort of comparisons also show how much of what we are seeing in these tests is very likely lens dependent, not a sensor issue. Here the Nikon and Sony lenses are perhaps simply a bit sharper than the FA77.

Or an ant bumped the DPR testing rig.