[Class A]

It's only proof of smoothing under two conditions:

1) the scene is indistinguishable from noise.

2) the accelerator is a linear filter.

@photoptimist, you seem to be missing the fact that the low ISO shots show no signal correlation but the higher ISO shots (past ISO 640) do. If you trust that all the shots were done consistently then the difference in the 2D FT plots shows that smoothing is applied (only) at higher ISO levels.

It makes me a bit uncomfortable that Rishi Sanyal said that he took the shots that bclaf then analysed, so we are in the hands of someone who has trouble understanding the problem with changing lenses when comparing two cameras, but if we assume that he did not mess up the shots (and how would he have done that in a manner that correlates with the ISO setting?) then the results posted by bclaf leave no room for doubt/interpretation.

It's going to take a lot more that an FFT of the noise to characterize what the chip is doing and whether it affects images in any detrimental way.

It only takes an FT to understand when higher spatial frequencies (aka "detail") have been tampered with.

Whether the images are affected in a "detrimental" way is another question, but there is no doubt that processing is occurring and to RAW purists that is a problem in itself.

Having just read about the quiet, behind the scenes changes of DPReview images, I lost pretty much all trust into the material that bclaf was given by Rishi Sanyal. God knows what happened. I'm not saying the results are bogus for sure, but unless the analysis is confirmed by someone else, I would reserve some scepticism.

---

Last edited by Class A; 05-11-2018 at 05:49 AM.

[normhead]

@photoptimist, you seem to be missing the fact that the low ISO shots show no signal correlation but the higher ISO shots (past ISO 640) do. If you trust that all the shots were done consistently then the difference in the 2D FT plots shows that smoothing is applied (only) at higher ISO levels.

It makes me a bit uncomfortable that Rishi Sanyal said that he took the shots that bclaf then analysed, so we are in the hands of someone who has trouble understanding the problem with changing lenses when comparing two cameras, but if we assume that he did not mess up the shots (and how would he have done that in a manner that correlates with the ISO setting?) then the results posted by bclaf leave no room for doubt/interpretation.

Ya, I'm waiting for the IR images. These guys simply stated, are a clown show.

[monochrome]

No RAW change, they changed the processed JPEG from RAW. Adding the different and defective lens I'm with you about "At this point they should take down the entire article, confess, and start over.".

Corrected, I think.

They nullify Pentax's standard raw adjustments in lightroom and apply their own WB and exposure compensation. First pic is wat DPReview shows us, second is what is loaded into lightroom by default with noise and sharpness adjustments:

They probably would claim some kind of equivalence justification. Or something.

I don’t really have a dog in this hunt. I hope to limit my comments, but I can’t abide academic sloppiness that looks like a conflict of interest. The appearance of a conflict is often worse than the reality, and then you end up in this kind of situation - lost credibility and no way to correct the record.

[Mallee Boy]

So.... the Ricoh legal Dept doing a bit of overtime ? IF any of this can be proved someone is going to take a hit.

[normhead]

So.... the Ricoh legal Dept doing a bit of overtime ? IF any of this can be proved someone is going to take a hit.

We should be so lucky. But now that they are owned by Amazon, it could be a lucrative way of raising money for lens development.

[photoptimist]

@photoptimist, you seem to be missing the fact that the low ISO shots show no signal correlation but the higher ISO shots (past ISO 640) do. If you trust that all the shots were done consistently then the difference in the 2D FT plots shows that smoothing is applied (only) at higher ISO levels.

There's zero evidence of "smoothing" in bclaff's analysis because bclaff's analysis does not look at an image with any signal in it. There's only evidence of noise reduction. There's an assumption that noise reduction requires smoothing but that is unproven and may be false.

Attenuation of higher spatial frequencies in a noise image says nothing about attenuation of higher spatial frequencies of a signal if the filter is a non-linear. A nonlinear filter can have strong attenuation of noise and near-zero attenuation of signal. The only signal that is in danger of attenuation from these kinds of filters is signal of an amplitude so low that it's indistinguishable from noise. But if the signal is indistinguishable from noise, it will also be lost in an image without NR anyway. And even weak signals may be preserved by a well-designed nonlinear filter because the spatial and chromatic structure of signal is different from the spatial and chromatic structure of noise.

[Shivaess]

There's zero evidence of "smoothing" in bclaff's analysis because bclaff's analysis does not look at an image with any signal in it. There's only evidence of noise reduction. There's an assumption that noise reduction requires smoothing but that is unproven and may be false.

Attenuation of higher spatial frequencies in a noise image says nothing about attenuation of higher spatial frequencies of a signal if the filter is a non-linear. A nonlinear filter can have strong attenuation of noise and near-zero attenuation of signal. The only signal that is in danger of attenuation from these kinds of filters is signal of an amplitude so low that it's indistinguishable from noise. But if the signal is indistinguishable from noise, it will also be lost in an image without NR anyway. And even weak signals may be preserved by a well-designed nonlinear filter because the spatial and chromatic structure of signal is different from the spatial and chromatic structure of noise.

The thing that I think gets lost in all of this talk about the tests is that we wouldn’t be having this contentious of a discussion if the NR effect had been optional. In my view that decision still represents a serious miscalculation by our beloved manufacturer. Especially after prior fiascos involving NR.

So let’s get some more testing and let’s get a firmware update.

[photoptimist]

The thing that I think gets lost in all of this talk about the tests is that we wouldn’t be having this contentious of a discussion if the NR effect had been optional. In my view that decision still represents a serious miscalculation by our beloved manufacturer. Especially after prior fiascos involving NR.

So let’s get some more testing and let’s get a firmware update.

That's a good point. Yet I can understand why Ricoh did what they did for the following reasons:

1. the use of NR on K-3 RAWs did not seem to prompt any outrage.
2. the use of the accelerator NR on K-70 RAWs did not seem to prompt any outrage.
3. the use of the accelerator NR on KP RAWs did not seem to prompt any outrage.
4. if accelerator NR could be turned-off, how much shall we wager that DPR would have turned it off to "prove" the K-1ii was no better than the K-1?

[normhead]

The thing that I think gets lost in all of this talk about the tests is that we wouldn’t be having this contentious of a discussion if the NR effect had been optional. In my view that decision still represents a serious miscalculation by our beloved manufacturer. Especially after prior fiascos involving NR.

So let’s get some more testing and let’s get a firmware update.

And the thing that gets found in this is that so far, there is no credible evidence that there is any need to do that.
The fact that some "chicken little's" pronounce that there have been fiasco's doesn't make it true.

What I see is less resolution in some parts of the images, more resolution in other parts of the images, some really bad studio work by the testers at DPR, and seriously, as a guy who spent some time in various studios, I'd be embarrassed to have published the work they did. Overall the images are cleaner, no question about that. The loss of resolution thing is debatable.

I'd be disappointed if Pentax took the time and money to put on an un-necessary switch on a feature just to degrade it's performance. They've got bigger fish to fry on a limited budget. I trust the judgement of real photography engineers over website attention grabbers any day.

[DennisP]

I would like to see the option to select the ISO level that it kicks in. I’m also curious as to how the accelerator affects AF performance since it’s been implied that it plays a role somehow. My initial tests with my K-1 II and 24-70 lens have been very encouraging; the AF feels more responsive and “confident”. I’m not sure if this is measurable, but it feels faster than the K-1.

[Not a Number]

That's a good point. Yet I can understand why Ricoh did what they did for the following reasons:

1. the use of NR on K-3 RAWs did not seem to prompt any outrage.
2. the use of the accelerator NR on K-70 RAWs did not seem to prompt any outrage.
3. the use of the accelerator NR on KP RAWs did not seem to prompt any outrage.
4. if accelerator NR could be turned-off, how much shall we wager that DPR would have turned it off to "prove" the K-1ii was no better than the K-1?

For years Pentax DSLRs came with mandatory Anti-Aliasing filters. Nobody complained about the loss of detail. Then Hoya gave you a choice with the K-5ii and K-5iis - WIth AA filter or without, respectively. Ricoh then released the K-3 without AA filter with "simulated AA filter" via Shake-Reduction. I don't see a huge outcry that Hoya/Ricoh offer a free upgrade service to remove the AA filter on older models. People seem perfectly happy with their K-5 or earlier with the loss of detail from the mandatory AA filters. It seems telling the K-70 and KP did stir up any outrage but rather praise for the high ISO performance. Maybe of Ricoh said nothing at all about the accelerator chip nobody would have noticed anything other than better high ISO.

[house]

Wow they really have switched the images. The K-1 II now looks pretty darn good. Still some loss of detail at very high iso but a much more reasonable tradeoff.

[SSGGeezer]

For years Pentax DSLRs came with mandatory Anti-Aliasing filters. Nobody complained about the loss of detail. Then Hoya gave you a choice with the K-5ii and K-5iis - WIth AA filter or without, respectively. Ricoh then released the K-3 without AA filter with "simulated AA filter" via Shake-Reduction. I don't see a huge outcry that Hoya/Ricoh offer a free upgrade service to remove the AA filter on older models. People seem perfectly happy with their K-5 or earlier with the loss of detail from the mandatory AA filters. It seems telling the K-70 and KP did stir up any outrage but rather praise for the high ISO performance. Maybe of Ricoh said nothing at all about the accelerator chip nobody would have noticed anything other than better high ISO.

This^^^ Telling them and showing a new chip on the board gave them something to create image boogeymen with. Poor execution of their testing, and assumptions of facts not in evidence. Storm of clickbait and trolls.

[beholder3]

Wow they really have switched the images.

Yep. Quite interesting to see how they make a headline announcement once they add a handful poor to average snapshots to a sample image gallery, but if they swap out underlying data from an opinion piece called "test" or "review" (which obliterates key statements from it) it is done like selling a black market item in a dpr dark alley. Sssh. Click. Click. Sssh.

[monochrome]

The problem is, their results are not repeatable; they should publish the complete range of camera and RAW developer settings they used to generate the image files, appended to each image post. Then a second party could independently repeat the experiment.

Having to individually dig this information out of EXIF is clearly unworkable.

Unless the results can be repeated they are not scientifically valid - and independent, scientific comparison is the core of their claim to a valid rating process.