[teiki arii]

When I read tests of a camera should be interpreted with the actual lenses the brand can produce, I completly agree with this kind of logical reasoning, but this should be explicit mentionned in reviews. The conclusion can apply to a system ( camera+lens) but not to the camera we'd like to be reviewed...

---

Last edited by teiki arii; 05-17-2018 at 12:52 AM.

[Ian Stuart Forsyth]

Even Bill, who has produced compelling evidence, leaves a tiny bit of probability that Ricoh doesn't apply smoothing.

I have been looking at some of the raw files in FRV and it look like there is some smoothing being done in.
If we look in the green channel and look at the dust found on the board you can see what look like grain but I it is dust on the surface of the board
If we look at the K1m2 shot at iso 100 you can see this dust


While with the K1 at iso you can see some noise introduced but not very much and with the K1m2 @ iso 800 you can see the smoothing

Just to remind people this is look at the raw files without NR other than what is done in camera

Here is the K1m2 iso 100 to show the dust ( again in a raw viewer)

[Class A]

Photozone had the 77 on 10mp K10D and the d50 2.8 on 16mp k5. So actual numbers can't be compared unfortunately,

Yes, you are right, this makes a numerical comparison impossible and sorry for having posted the numbers without realising that two different cameras were used. I swear I tried to convince myself that the same cameras had been used; not sure how I could mistake the K-5 on the front page with a K10D.

However, in this particular case, I think we are not totally clueless as Klaus (from opticallimits) tested the FA 43/1.9 on both K10D and the K-5. The latter test came up with lower resolution numbers.

There could be all sorts of reasons for the lower resolution numbers (one is that Klaus' measurements are not scientifically accurate in an absolute sense, they really only support relative comparisons between lenses on the same camera) but if it were true that all lenses would deliver lower resolution numbers on the K-5 (in Klaus' tests) compared to the K10D then that would mean the 50/2.8 macro is even better than the figures posted suggest.

[Class A]

While with the K1 at iso you can see some noise introduced but not very much and with the K1m2 @ iso 800 you can see the smoothing

Yes, I've observed a similar effect on other images. Rarely, but sometimes, the K-1 II image lacks some information that the K-1 retains (albeit in a noisy form).

I don't want to argue against anyone who prefers the K-1 II results. All I'm asking is that we unite in terms of asking Ricoh for making the adjustments optional. If Ricoh did that, everyone would win and DPReview would have to give the K-1 II a better score.

[Rondec]

It would make sense if the shots were only available in the context of the article.

However, one can use DPReview's camera comparison tool outside any article context and it will be referenced by other DPReview articles. This will inevitably lead to the K-1 II being compared to say a Sony A7 III and then it matters how sharp the lens used on the K-1 II is.

The whole camera comparison tool doesn't make much sense, given they don't use a reference lens model (like Imaging Resource did with the Sigma 70/2.8 macro) across all camera brands. With the current limitations in place, the best you can hope for a camera you'd like others to appreciate is that they use the best possible lens available for it. I love the 77/1.8 as a portrait lens, but I think there should be better choices for the purpose of capturing a studio scene with the best resolution possible. A proper copy of the 50/2.8 macro would be preferable, don't you think?

And yes, they should reshoot the K-1 with the 50/2.8 macro as well, to support comparability in this manner.

You are right, if you have a good copy of the 50 macro. But apparently theirs has decentering or something as the right upper quadrant of their test image is really soft throughout. As it currently stands, people are constantly pulling bits out of that part of the image and you can't draw much conclusion from them as a result.

As to the general question, I have yet to see someone take a photo which was "damaged" by the noise reduction offered by the K-1 II. Obviously this is a new camera and so we don't have much in the way of examples, but I see people blowing up pixels and pointing out specks of dust that may be missing at iso 800. But when I shoot at iso 800 it isn't generally with the intention of printing big enough that dust pixels will show up and the bigger issue is whether the overall image is effected and my guess is that it isn't.

I still am waiting on the astro photographers among us to test if this will affect star fields (I don't shoot astro myself), but that is the one application where I think this might have a negative aspect to it.

In general, I have said before and I will say again, the accelerator more than matches the majority of photographers ability to denoise an image without losing detail, with minimal time and effort involved. And that second part is really important as well and not to be glossed over. Even if you can get roughly the same results running your images through DXO Mark's PRIME noise reduction software, that takes time -- quite a bit of time to run each image through. From an efficiency stand point, this is a definite winner with minimal clear loss.

It feels as though you and several others are arguing more on the principle of the thing. RAW files just should be pure with minimal pre-processing done on them. Even if you could show that no detail was lost at all it would just feel wrong to do noise reduction of some kind on the RAW file.

I wouldn't be surprised if Pentax does eventually allow some tweakability to this to allow it to kick in at different isos, but certainly at the upper end of the iso spectrum there is no point to even shooting those unless you have some noise reduction, whether in camera or out of camera.

[MJKoski]

Japanese style of innovation does not allow admitting design mistakes. Fuji have taken different approach lately. Sony is good example, providing uncompressed RAW took 2 years and even then it was half-assed solution which cannot do lossless compressed. Instead Sony gives 16-bit container and every uncomp RAW from 42MP cameras take whopping 90MB of storage!

[Tan68]

four times 24mp aps-c two with and two without nr chip:
Image comparison: Digital Photography Review

I suppose these are offered as direct comparisons to the KP
Do we agree the K-1 to the K-1 II are the same except for the accelerator?

All these sensors in the cameras you linked are 24 MP. Of course it is possible all four sensors are the same with no revision but haven't read anyone is certain, so I would like to set aside the 24 MP-ness these cameras have in common.

On this forum and others, Pentax has been lauded for doing a better job than even Sony at getting the most out of any Sony sensor. I presume this extra image quality Pentax manages has to do with the electronic doodads surrounding the sensor and how they process the signal.

If we accept K-3 as a direct comparison (in the context of my previous comment : K-1 to K-1 II) to KP, then we are accepting there has been no advancement in the doodads or processing Pentax uses. Although it is possible there has been no advancement, I think it is very unlikely the doodads and etc. are the same between KP and K-3 models. The KP and K-70 could be the same but who knows !!! We are guessing that these 4 models are as alike to each other as are the K-1 & K-1 II :^)

Bill Claff has recently written people might be surprised by notable difference between camera models because of the different electrical doodads in the different models. I paraphrase him and the comment is in a DPR thread and I will not go look it up for reference :^)

We're deep in the pixel peeping jungle. Within the same two DPR sample images for K-1 & K-1 II **, some portions favor the K-1 while some portions favor the K-1 II. The differences are small but at least ..fairly noticeable. And if they weren't, there wouldn't be so much discussion!

Bill has written there are differences between camera models and I kind of expect those differences may not be very great; he probably has different IQ thresholds than I for his tests. But, let's accept the differences are there and that they are small but at least ..fairly noticeable.

Considering there may have been advancements by Pentax to the electrical doodads and that Bill notes differences between models, I think it is not possible, in the context of my post, to make a direct comparison of 4 models manufactured over 4 or so years.

Your comparison is useful for some topics but I am going to stick to my guns that there is no direct comparison for KP as there is between K-1 and K-1 II. I hope my comments have helped you understand why I am stubborn on this point :^)

** of course, comparison made in the portions of the sample images considered reasonably suitable for comparison

[Class A]

It feels as though you and several others are arguing more on the principle of the thing. RAW files just should be pure with minimal pre-processing done on them. Even if you could show that no detail was lost at all it would just feel wrong to do noise reduction of some kind on the RAW file.

I don't understand why it feels that way to you.

I would have absolutely zero problem with any non-destructive RAW treatments (I'm pretty sure that we don't see pure data from the sensor in any event).

Even though I'd have to run all my images through a RAW converter anyhow as I would never degrade a camera of the calibre of a K-1 to a JPG machine, any improvement to RAWs is welcome as they help to reduce processing time.

If anyone could demonstrate that no detail is lost (AFAIC, the opposite has been done already) then I would not complain in the slightest. On the contrary.

I presume this extra image quality Pentax manages has to do with the electronic doodads surrounding the sensor and how they process the signal.

It is a long-held assumption that a camera manufacturer can squeeze more out of a sensor by tweaking it. You need to understand, though, that a modern Sony sensor is a black box that spits out digital data. You cannot reach inside and optimise anything on an analogue level. Even providing super stable voltages to the sensor does not help, as the sensor should not be viewed as an analogue device. I'll admit that I'm not passing on first-hand knowledge. Instead, I'm drawing on comments made by user falconeye, a physicist who dived much deeper in the matter than I did. I'm not invested/interested enough to study sensor spec sheets, but if you do, I'm sure you'll find that you can confirm falconeye's statements. Falconeye, back when he was a Pentax user, made many scientifically sound contributions, such as investigating the K-7's shutter blur issue. You can read about his work on his LumoLabs pages. Although I much prefer to not parrot what others are saying, I feel highly confident in passing on his assessment on the matter of "squeezing quality out of a sensor".

As to why Sony did worse using their own sensors, there are a number of potential explanations. Perhaps Pentax uses a different CFA which is less discriminating but lets more light through overall. Perhaps Pentax always performed some RAW baking (they most certainly did for a number of camera models beyond ISO 1600, for instance). Another potential source for a difference could be thermal conditions. Sony has to run their sensors full-time which I don't think is helpful with respect to noise. Sony, with their compact cameras, may also provide a worse cooling environment. I don't claim to know all the potential sources, but I think it is pretty much certain that there is no way Pentax has found a way to use the "accelerator" to truly only reduce system noise, without affecting potential information in the scene at the same time.

[reh321]

I don't understand why it feels that way to you.

Statements like the one below are exactly why someone might agree with statement by Rondec.

I would never degrade a camera of the calibre of a K-1 to a JPG machine

Some of the finest photographers in the world shoot the Olympics for Getty, and yet news reports clearly state that this work is done in JPEG only.

[MJKoski]

Yeah, there are timeframes in which material must be loaded to (photo) agency servers. There is no time to waste by processing the images for an art exhibition. Raise your hands if you are working against such limits with a Pentax :O

[reh321]

Yeah, there are timeframes in which material must be loaded to (photo) agency servers. There is no time to waste by processing the images for an art exhibition. Raise your hands if you are working against such limits with a Pentax :O

You are another one who demonstrates bias {in this case, against Pentax} with every breath. Find a productive use of your time.

[Rico]

Yeah, there are timeframes in which material must be loaded to (photo) agency servers. There is no time to waste by processing the images for an art exhibition. Raise your hands if you are working against such limits with a Pentax :O

It doesn't matter what brand you are using you will always be wasting time in post processing. The DPR Studio Scene attests to that. It is clear how bad the noise really is for the K-1, D850, A7III and even the Canon 5DVI. There is no denying how much better the K-1MKII is doing in creating a much finer grain image with better continuous tone and better value retention as ISO increases right from ISO 100. That translates into better bokeh. Better turn and shadow reflected light transitions. A more natural rendition for printing. Right out of the camera.

---------- Post added 05-17-18 at 10:42 AM ----------

I have been looking at some of the raw files in FRV and it look like there is some smoothing being done in.
If we look in the green channel and look at the dust found on the board you can see what look like grain but I it is dust on the surface of the board
If we look at the K1m2 shot at iso 100 you can see this dust


While with the K1 at iso you can see some noise introduced but not very much and with the K1m2 @ iso 800 you can see the smoothing

Just to remind people this is look at the raw files without NR other than what is done in camera

Here is the K1m2 iso 100 to show the dust ( again in a raw viewer)

The K-1MKII crop is doing a way better job maintaining what is actually there. You only think it is smoothing. The K-1 image has a blanket of noise you think is "more" detail when it is not.

[MJKoski]

Oh gosh...how many of you are really shooting Olympic games for an agency with a Pentax? You need either super fast WLAN of LAN on-site to get the stuff to a hub which handles all the traffic forward. Before that you need to actually capture the sportsman doing their stuff. Good luck.

[angerdan]

The photozone (now "opticallimits") charts come to the conclusion that the FA 77 resolves higher at f/5.6 than the D FA 100mm Macro. So with DPReview using the 77mm stopped down, the results should be better.

..77mm.f1.8 @ f/5.6 center: 2303 lw/ph
100mm.f2.8 @ f/5.6 center: 2194 lw/ph

. 77mm.f1.8 @ f/5.6 border: 2214 lw/ph
100mm.f2.8 @ f/5.6 border: 2126 lw/ph

[DennisP]

The photozone (now "opticallimits") charts come to the conclusion that the FA 77 resolves higher at f/5.6 than the D FA 100mm Macro. So with DPReview using the 77mm stopped down, the results should be better.

..77mm.f1.8 @ f/5.6 center: 2303 lw/ph
100mm.f2.8 @ f/5.6 center: 2194 lw/ph

. 77mm.f1.8 @ f/5.6 border: 2214 lw/ph
100mm.f2.8 @ f/5.6 border: 2126 lw/ph

I've never used the DFA 50 Macro, but I have used the 100 and the 77 and have always felt that the 77 at F5.6 is about as sharp as possible. I'd put it up against glass from any manufacturer at that setting. I'm very surprised and still have doubts that the DFA 50 Macro is sharper in the center than the Limited; it just doesn't make sense given the design intentions with typical macro lenses vs. other lenses (max resolution in center for Limited, corner to corner emphasis for macros at the expense of a smidgen of resolution in the center).