[Class A]

Have you seen this?

Have you seen images before and after denoising?

Also note that I did not write that I have proof of the K-3 III faking its advantages; I only expressed the sentiment that it might be better to keep one's expectations low in case the advantages are not all due to hardware improvements.

Oh, boy... here we go again...

Always nice to see someone embodying a friendly attitude without any attempt of confrontation, mocking, or othering.
No wonder moderators love such constructive remarks.

I said my bit and will stay out of any follow up discussion.

[Batman_]

Not really interested in these super high iso examples but if the improvement is real at say iso 12,800 then I'd be very happy. I set auto-iso on KP go to 6,400, so one full stop on this with equal or marginally better image quality would be a real gain for me.

AF improvement;
1.05 magnification ovf; and
Improved iso

are enough for me to get this camera if the price is within range.

[PePe]

Not really interested in these super high iso examples but if the improvement is real at say iso 12,800 then I'd be very happy. I set auto-iso on KP go to 6,400, so one full stop on this with equal or marginally better image quality would be a real gain for me.

AF improvement;
1.05 magnification ovf; and
Improved iso

are enough for me to get this camera if the price is within range.

Exactly!
I think this talk about image quality at ISO 102400 or 204800 is somewhat fallacious. It is still going to be useless for most purposes - only on a higher level ! If it is going to have a 1-1,5 stop advantage over K-3 II at usable ISOs, it is all very welcome. However, AF and all the other things are just as important.

[MMVIII]

Have you seen images before and after denoising?

Also note that I did not write that I have proof of the K-3 III faking its advantages; I only expressed the sentiment that it might be better to keep one's expectations low in case the advantages are not all due to hardware improvements.


Always nice to see someone embodying a friendly attitude without any attempt of confrontation, mocking, or othering.
No wonder moderators love such constructive remarks.

I said my bit and will stay out of any follow up discussion.

And no surprise. You claimed you would be supercritical towards this topic because you expect others to exaggerate it as a problem. And as I expected, you alone was the first to frame it in a "problem". I told you...

In German there is a phrase expressing that someone died due to his constant fear of death

[ThorSanchez]

Don't get your expectations too high up. I predict that most of that gain will be due to in-camera denoising, i.e., it won't present a real advantage (other than saving you to do the denoising in post-processing).

The K-1 II has a ~1.2 stop advantage over any APS-C camera with the same sensor performance (due to the larger sensor, provided in a comparison the same f-stop is used, resulting in less DOF for the K-1 II compared to the K-3 III).

The K-3 III apparently uses a BSI sensor which gives it a half- to one stop advantage over a camera with a regular sensor.

I think we can safely assume that the sensor technology in the K-3 III will not be significantly be better compared to the K-1 II sensor (beyond the BSI advantage), so that leaves us still with a slight advantage for the K-1 II.

If the K-3 III now is still ahead with "1.5 stops less noise" then that means that all this gain just comes from in-camera post-processing.

This is not to throw shade on the K-3 III, the latter will be an awesome camera, I'm just trying to not let expectation run wild. I'm sure after a while, we'll get some hard data about the sensor used in the K-3 III and then we'll definitely know for sure how much of the "noise advantage" is due to hardware and how much is due to "image processing" (which has been publicly announced for the K-3 III and which the Pentax engineers are very proud of).

I don't particularly care how they get there as long as they get there. It's a false purity to think hardware = real, and software = fake. The only thing that matters is the end result. The K-3ii has a max ISO setting of 51,200 and that's only usable for reh321's security cam photos of bigfoot raccoons. The example they've shown of 102,400 looks better than 12800 on the K-3ii. Shooting youth sports at night it's commonplace to have to use 12,800 or higher, with compromises of shutter speed, so there are real-world implications of ISO settings at these levels.

Of course I want to see more examples and actual RAW files and reviews. I'm not expecting 2, 3, 4 stops of magic. But if I can get better-than-KP results at very high ISO that will be nice.

[reh321]

I don't particularly care how they get there as long as they get there. It's a false purity to think hardware = real, and software = fake. The only thing that matters is the end result. The K-3ii has a max ISO setting of 51,200 and that's only usable for reh321's security cam photos of bigfoot raccoons. The example they've shown of 102,400 looks better than 12800 on the K-3ii. Shooting youth sports at night it's commonplace to have to use 12,800 or higher, with compromises of shutter speed, so there are real-world implications of ISO settings at these levels.

Of course I want to see more examples and actual RAW files and reviews. I'm not expecting 2, 3, 4 stops of magic. But if I can get better-than-KP results at very high ISO that will be nice.

My usage is not “security cam” - I show the 800K image only to show it.

I regularly use my KP in the 100-12000 range without expectation that anyone can tell the difference between 100 and 12000.
For me, ‘TAv’ mode was useless, because I never ventured above 800.

[Wheatfield]

Don't get your expectations too high up. I predict that most of that gain will be due to in-camera denoising, i.e., it won't present a real advantage (other than saving you to do the denoising in post-processing).

The K-1 II has a ~1.2 stop advantage over any APS-C camera with the same sensor performance (due to the larger sensor, provided in a comparison the same f-stop is used, resulting in less DOF for the K-1 II compared to the K-3 III).

The K-3 III apparently uses a BSI sensor which gives it a half- to one stop advantage over a camera with a regular sensor.

I think we can safely assume that the sensor technology in the K-3 III will not be significantly be better compared to the K-1 II sensor (beyond the BSI advantage), so that leaves us still with a slight advantage for the K-1 II.

If the K-3 III now is still ahead with "1.5 stops less noise" then that means that all this gain just comes from in-camera post-processing.

This is not to throw shade on the K-3 III, the latter will be an awesome camera, I'm just trying to not let expectation run wild. I'm sure after a while, we'll get some hard data about the sensor used in the K-3 III and then we'll definitely know for sure how much of the "noise advantage" is due to hardware and how much is due to "image processing" (which has been publicly announced for the K-3 III and which the Pentax engineers are very proud of).

The K1 sensor is somewhere around 8 years old now. I realize the technology isn’t improving as fast as it was 20 years ago, but it isn’t a stretch to think that sensor noise levels have improved over the past almost a decade.

[reh321]

The K1 sensor is somewhere around 8 years old now. I realize the technology isn’t improving as fast as it was 20 years ago, but it isn’t a stretch to think that sensor noise levels have improved over the past almost a decade.

People talk about noise, but DR and color fidelity are also issues - the KP is better in all three at higher ISO values, and I think viewers will be satisfied with how the K-3iii handles all three.

[Class A]

I know I said I'd stay out of a follow up discussion and I still won't respond to non-technical content, but I feel there is a need for clarification and I'd like to make a concession.

It's a false purity to think hardware = real, and software = fake.

I don't get the "purity" argument.

My point is that the "image processing" software improvements can be had from any camera. These do not constitute an advantage of any specific camera. On the contrary, out of camera processing can make use of more computational power and improving algorithms, and they work the better the less adjustments have been made to the image data already.

Let's avoid the debate of whether or not the K-3 III does any in-camera processing. Just in case it did, a fair comparison to a different camera would have to involve equivalent processing of the images of the other camera.

Some regard it as an advantage, if the camera takes care of the post-processing and I fully see that point, but one cannot say one camera has better image quality than another just because the other camera gives the user freedom over which, if any, processing to apply.

I realize the technology isn’t improving as fast as it was 20 years ago, but it isn’t a stretch to think that sensor noise levels have improved over the past almost a decade.

Indeed, you are correct in acknowledging that sensor improvements have been modest. The K-5's sensor is even older than the K-1's but already realises a performance close to the theoretical limit (of its sensor type).

The concession I'd like to make concerns the possibility that the K-3 III's sensor has a dual-gain design. In this case, high-ISO performance would be boosted in a way that post-processing cannot emulate. This would give the K-3 III an advantage over the K-1 II (which only emulates the dual-gain behaviour via in-camera post-processing) which I didn't account for in my earlier comparison.

Still, though, if your hypothesis about the K-1 II's sensor being outmached by more modern sensors where true then this would imply that modern FF sensors from Canisonies, outperforn the K-1 II's sensor by about 2.5 stops. That's not what we are seeing, is it? To be honest, I haven't followed recent sensor performances but I cannot see how the aforementioned implication could be true.

[Kunzite]

Have you seen images before and after denoising?

Also note that I did not write that I have proof of the K-3 III faking its advantages; I only expressed the sentiment that it might be better to keep one's expectations low in case the advantages are not all due to hardware improvements.

So, they're getting better detail and less noise than a larger format camera. You've made a "prediction" that there's no real advantage.
At the same time, AFAIK nobody so far was able to replicate the K-1 II's results through whatever software denoising. It's all empty claims.

Then there's using words like "faking".

[Kunzite]

My point is that the "image processing" software improvements can be had from any camera. These do not constitute an advantage of any specific camera. On the contrary, out of camera processing can make use of more computational power and improving algorithms, and they work the better the less adjustments have been made to the image data already.

Hardware improvements can be had from any camera as well. They just have to be implemented

You're unable to replicate the results on a PC, despite your claims like the above (you basically say you can get superior results). Stop making claims you cannot support.
This isn't some philosophical dispute; you have to present evidence.

By the way, one of the patents I've read last year was about being able to precisely identify the noise generated by the SR mechanism - which had to be done simultaneously with the actual exposure.
I don't see how you could replicate the same results just by algorithmic guessing.

[Pål Jensen]

If I can get superior results from my camera without involving a PC (not to mention lots of work), I'm all for it...

[ThorSanchez]

I don't get the "purity" argument.

My point is that the "image processing" software improvements can be had from any camera. These do not constitute an advantage of any specific camera. On the contrary, out of camera processing can make use of more computational power and improving algorithms, and they work the better the less adjustments have been made to the image data already.

Let's avoid the debate of whether or not the K-3 III does any in-camera processing. Just in case it did, a fair comparison to a different camera would have to involve equivalent processing of the images of the other camera.

Some regard it as an advantage, if the camera takes care of the post-processing and I fully see that point, but one cannot say one camera has better image quality than another just because the other camera gives the user freedom over which, if any, processing to apply.

You're assuming the equivalence of external software doing the processing that Pentax (or other cameras) do as part of the process of turning sensor data into RAW files. I don't know that that is true. I think it's plausible that the camera's embedded software/firmware has very specific things it does to correct for noise, color fidelity, etc that other external software would not be able to.

[reh321]

Exactly!
I think this talk about image quality at ISO 102400 or 204800 is somewhat fallacious. It is still going to be useless for most purposes - only on a higher level ! If it is going to have a 1-1,5 stop advantage over K-3 II at usable ISOs, it is all very welcome.

The reason for talking about performance at ISO values that most of us will seldom use is continuity - yes, the KP does have a 'jump' when the 'accelerator' cuts in {I would personally prefer that it operate at all ISO levels just to reduce talk} - but a camera which does provide better noise, DR, and color fidelity at 800K will probably provide more graceful performance at lower values also. That is the real value of photos such as the last pair at
Pentax KP First Shots: Going head-to-head against the Nikon D500 at ISO 819,200!

[gatorguy]

The reason for talking about performance at ISO values that most of us will seldom use is continuity - yes, the KP does have a 'jump' when the 'accelerator' cuts in {I would personally prefer that it operate at all ISO levels just to reduce talk}
Pentax KP First Shots: Going head-to-head against the Nikon D500 at ISO 819,200!

Pentax engineers would apparently agree since it will kick in at base ISO now on the K3III rather than at 800 as on your KP.