[D1N0]

I think the noise cancelling chip doesn't do much for the K-1 II It's effect is only really visible on iso's above 12800 on normal viewing distances. You don't really want to go there anyway. Effects on the smaller 24mp aps-c sensor pixels is much greater at lower iso's so it is a good addition for those camera's. When a future full frame would have a 54mp sensor the noise cancelling chip would make more sense. So it enables the camera to have more megapixels. IBIS will have to be improved as well because smaller pixels need more accurate sr.

[Rondec]

Anyone able to replicate or seen demonstrated the clear discolouration of shadow detail on pushed under-exposed images from the K-1 II compared with the K-1, as this guy shows?: Pentax K1 MKII Image Quality Test - YouTube

I have not.

This is a 15 second iso 100 image:



This is the same image pushed 5 stops:



When I posted it earlier, I was told that it was only significantly longer exposures than that that would manifest the magenta-ish tinting demonstrated by some. Regardless, in my exposures up to 30 seconds, I have not seen it.

[Class A]

Purely digital, it's used in every audio system,...

It is not used in every audio system.

Noise shaping is a technique to reduce quantisation errors by by shifting noise into regions where it is less objectionable and/or distributing the energy to a larger band.
In audio, for instance, multi-bit D/A converters did not have to use any noise shaping, only so-called "1 bit converters" some of which use more than one bit, require noise shaping.

Downsampling alone wouldn't improve the resulting image IQ as the amount of noise vs signal would remain the same.

Of course "downsampling alone" already improves IQ (at the expense of resolution). That's why DxOMark's results often exceed theoretically possible values, as they are expressed in terms of normalised 8MP images.

[Not a Number]

Anyone able to replicate or seen demonstrated the clear discolouration of shadow detail on pushed under-exposed images from the K-1 II compared with the K-1, as this guy shows?: Pentax K1 MKII Image Quality Test - YouTube

As I recall using updated camera profiles rather than hacking the K-1ii RAW to be detected as a K-1 RAW eliminates these artifacts.

He also concludes that Hand Held Pixel shift doesn't work based on there being no difference between a "normal" RAW and a HHPS Raw compared in SilkyPix. He doesn't make it clear if he used SilkyPix or DCU. I'm assuming he is comparing JPEGs developed from the Raw files. A simple size difference between the PEF/DNG would tell you there is a difference. Regardless, Lee doesn't take into account that at the time of the video not even DCU would handle HHPS and would only use the first frame which just happens to be a "normal" RAW frame.

Early reviews should be taken with a grain of salt.

[Kunzite]

"Texture loss" sounds scary, if you don't realize not even a theoretically perfect camera could achieve zero "texture loss" as the light levels fall.
As for "sharpening", is there any camera that don't apply it? I had a quick look through some PDN reviews: X-T3 (the "best" according to DPR TV) does it, the "second best" A7 III does it, the EOS R does it, the Z7 does it...

"Pure RAWs" are an illusion; and Pentax would fall behind if they don't get better at image processing - because all the others are getting better.
Now, obviously I don't agree with any kind of processing. I don't think in extremes.

[reh321]

I had practically left this thread but stumbled across some measurements that help to understand that the K-1 II's "accelerator" unit is not able to magically only remove noise but leave signal (detail) intact. I thought I'd share these here, for those interested.

PDN measured the amount of artefacts in K-1 II images due to image processing (jump to "Texture loss" to see the corresponding graphs). This analysis uses the Imatest "Dead Leaves"/"Spilled Coins" charts which support the analysis of how much detail is lost in what spatial frequency band.

This confirms the analysis results by Bill Claff.

Yes, the loss of detail may be considered too small to matter.
Yes, the loss of detail may be considered a welcome trade off against lower noise.

No, it's not an imagined phenomenon.

PDN also confirms the presence of sharpening:

"The K1 Mark II sharpens images more than the K1 did, although both produce relatively little sharpening"

I was surprised to read that the K-1 already used sharpening. It would be very interesting to see the same analysis results for the K-1 and other DSLRs.

EDIT: I'm unsure whether the respective "Image Engineering" test procedures are based on RAW files or certain JPEG settings they describe. I only searched the document describing the procedures instead of reading every word. So technically, the artefacts measurements could be purely based on JPEG files. Seems unlikely but I wouldn't be able to exclude this possibility at this point.

Anyhow, let's hope Ricoh will give us the possibility of opting-out, should they choose to offer in-camera image processing again.
This would have to please everyone.

Blah, blah, blah.

"Just for the record", I am not a theoretician and I'm not going to spend time reading all this. I'm sure they could have done a similar study a year earlier on the KP - perhaps that would have alerted Pentax that someone cared about this, but that is all it would have done. My KP is giving me excellent ISO 12800 results, preserving the details I can see; I don't care about the others.

You may continue the theory, but me, I'm going to enjoy life and what I can see

[biz-engineer]

It is not used in every audio system.

You immediately reject, therefore you can't see. Relax sit back and instead of wanting to be right as primary goal, give yourself the goal to assess the advantages and disadvantages of this Pentax accelerator.

Of course "downsampling alone" already improves IQ (at the expense of resolution).

No, downsampling alone is a zero sum process: the noise is averaged out as much as resolution is decreased, but the SNR of the image is unchanged. Whereas band limited NR out of signal band does improve SNR.

[Andrea K]

Anyone able to replicate or seen demonstrated the clear discolouration of shadow detail on pushed under-exposed images from the K-1 II compared with the K-1, as this guy shows?: Pentax K1 MKII Image Quality Test - YouTube

NO. I tried in the past but with "no luck"

---------- Post added 08-01-19 at 14:08 ----------

He also concludes that Hand Held Pixel shift doesn't work based on there being no difference between a "normal" RAW and a HHPS Raw compared in SilkyPix. He doesn't make it clear if he used SilkyPix or DCU. I'm assuming he is comparing JPEGs developed from the Raw files. A simple size difference between the PEF/DNG would tell you there is a difference. Regardless, Lee doesn't take into account that at the time of the video not even DCU would handle HHPS and would only use the first frame which just happens to be a "normal" RAW frame.

Early reviews should be taken with a grain of salt.

I don't want to do a crosspost then you can go here to see Dynamic Pixel Shift at work: From a k-1 user real world impressions about the k-1 II - PentaxForums.com

[BarneyL]

As for "sharpening", is there any camera that don't apply it? I had a quick look through some PDN reviews: X-T3 (the "best" according to DPR TV) does it, the "second best" A7 III does it, the EOS R does it, the Z7 does it...

Lenstip doesn't confirm sharpening of RAWs in K-1 and K-1 II (nor in X-T3, EOS R, Z7). They detected some signs of possible RAW sharpening in A7 III, but they admit the premises aren't convincing enough, and if any sharpening takes place, its truly minimal. They also detected some sharpening in A7RIII's implementation of Pixel Shift (none in Pentax variant), but minimal as well.

[Ash]

As I recall using updated camera profiles rather than hacking the K-1ii RAW to be detected as a K-1 RAW eliminates these artifacts.



He also concludes that Hand Held Pixel shift doesn't work based on there being no difference between a "normal" RAW and a HHPS Raw compared in SilkyPix. He doesn't make it clear if he used SilkyPix or DCU. I'm assuming he is comparing JPEGs developed from the Raw files. A simple size difference between the PEF/DNG would tell you there is a difference. Regardless, Lee doesn't take into account that at the time of the video not even DCU would handle HHPS and would only use the first frame which just happens to be a "normal" RAW frame.



Early reviews should be taken with a grain of salt.

Thanks for that. So I gather this finding has then been debunked with the latest version of ACR/Lightroom camera profiles? Or has anyone still getting the magenta cast in shadow areas of underexposed shots pushed in PP?

[lsimpkins]

As for "sharpening", is there any camera that don't apply it?

Apparently not if they use a Sony BSI sensor. According to the block diagram of BSIs on the Sony Semiconductor website, noise cancellation techniques are applied to the signals on-chip both prior to and after A/D conversion.
CMOS Image Sensor Image Sensor for Camera | Sony Semiconductor Solutions Corporation
I was not interested enough to pursue the same information on the earlier non-BSI sensors.
So this begs the question, what is "pure raw" data?

[biz-engineer]

So this begs the question, what is "pure raw" data?

Raw data is always processed, because if it wasn't it would be the light that hit the front of the lens. If the accelerator was already used in the K1, there would be no discussion.

[Ash]

I think the noise cancelling chip doesn't do much for the K-1 II It's effect is only really visible on iso's above 12800 on normal viewing distances. You don't really want to go there anyway. Effects on the smaller 24mp aps-c sensor pixels is much greater at lower iso's so it is a good addition for those camera's. When a future full frame would have a 54mp sensor the noise cancelling chip would make more sense. So it enables the camera to have more megapixels. IBIS will have to be improved as well because smaller pixels need more accurate sr.

It's fascinating, this human nature thing... Now, I don't want to sound dismissive of your point, I actually agree that people would rarely shoot above ISO 12,800. However, for years members here have lamented over how Pentax cameras just didn't perform well at high ISO, and that photographers, esp. in the wedding business, absolutely NEEDED that high ISO performance. Low light situations demanded it!

Now look at us! We're saying high ISO is irrelevant!
Pentax must be confused by market demand.
But like many innovative technologies, people don't tend to know what they need until the technology has been developed and it's in their hands.
I would go as far as to say that shooting clean images at ISO 6,400 and 12,800 is a definite virtue above the barely perceivable loss of detail noted between the K-1 and K-1 II images. In fact, the microcontrast in the K-1 II images with PS and DPS may even have the edge over rivals.
But noise reduction was a big reason why people would go from APS-C to FF, so the effort makes sense even in a 36Mp camera, less dense pixels than in the K-3.

[reh321]

It's fascinating, this human nature thing... Now, I don't want to sound dismissive of your point, I actually agree that people would rarely shoot above ISO 12,800. However, for years members here have lamented over how Pentax cameras just didn't perform well at high ISO, and that photographers, esp. in the wedding business, absolutely NEEDED that high ISO performance. Low light situations demanded it!

Now look at us! We're saying high ISO is irrelevant!
Pentax must be confused by market demand.
But like many innovative technologies, people don't tend to know what they need until the technology has been developed and it's in their hands.
I would go as far as to say that shooting clean images at ISO 6,400 and 12,800 is a definite virtue above the barely perceivable loss of detail noted between the K-1 and K-1 II images. In fact, the microcontrast in the K-1 II images with PS and DPS may even have the edge over rivals.
But noise reduction was a big reason why people would go from APS-C to FF, so the effort makes sense even in a 36Mp camera, less dense pixels than in the K-3.

More graceful higher ISO performance was my main reason for purchasing a KP, and I am very pleased by what it delivers at 12800.
https://www.pentaxforums.com/forums/12-post-your-photos/379920-pets-first-photo-kp.html
https://www.pentaxforums.com/forums/12-post-your-photos/380693-misc-iso-25600-a.html

but I'm not comparing these to images from some other camera - I'm comparing details recorded to details I know are actually there.

[Ash]

I have not.

When I posted it earlier, I was told that it was only significantly longer exposures than that that would manifest the magenta-ish tinting demonstrated by some. Regardless, in my exposures up to 30 seconds, I have not seen it.

Have you purposely under-exposed at ISO 100 then pushed exposure to +4 or +5 to see the effect on shadow areas?