[BigMackCam]

It may well be that lightroom or it's settings influenced the end result in a bad way for some of the models in that test

This is almost certainly a profiling issue. The black points for red, green and blue channels can make a big difference regarding colour accuracy with heavy shadow recovery. If the black points set in the profile used by the application are slightly inaccurate - or there's a minor variance in the colour and luminance data recorded by different copies of the same model camera (which is a possibility) - then you'd get a colour cast when there's an extreme boost to exposure or low-end shadow recovery.

I see this with my Hasselblad HV in Lightroom 6 (one of the few raw converters that has a ready-made profile for this camera). If I have a shot with a deep black background - for example, the cloudless sky of a moon shot exposed for the moon - and boost the exposure by +4 or +5 EV, I get a magenta tint on the black background. The solution - not a perfect one, but it works quite well - is to adjust the shadow tint towards the green very, very slightly. RawTherapee, by comparison, offers full control over the individual black-points for red, green and blue channels. It's more difficult to find the precise settings, but possible to get it near-enough perfect.

Given that the issue only appears with big adjustments to exposure and/or shadows, the only time I can realistically see this being a practical issue is if you're shooting at a low ISO and relying on so-called "ISO invariance", then increasing exposure as required in the raw converter. It's an approach I've never liked, even on cameras that reviewers claim are ISO invariant (or very nearly so). There's always some visible loss in quality compared to shooting at the correct ISO...

---

Last edited by BigMackCam; 01-09-2022 at 09:34 AM.

[normhead]

A year ago people here loved that guy and always linked to his videos in support of Pentax, how things have changed.

So we loved him because he supported our brand, and now we don't because he doesn't. Dude, we never loved the guy, we loved the "Pentax is the best for me" message. Did you really think we were in love with the guy and follow him if he changed his mind? The simple fact is, he pointed out things about Pentax in a way no other site did. I know that bothers switchers and off brand users. Like those who continue to bash Pentax on a pro-Pentax site and are just looking for ammunition to throw back in the faces of those who successfully use Pentax gear. Boy did noses get out of joint when he went against the marketing narrative of the big boys. I can tell, some are still smarting. Hence the snootiness.

In every test where Pentax has been compared in the field, in a fair test with other brands, Pentax gear has won praise even when compared to brands costing 2 to 3 times as much, by surprised/confused testers who based on internet fueled bias, thought Pentax was much worse than it is.

This guy has switched brands so often, anyone with a brain knew this day would come, we expected it, and we don't care. But it was good while it lasted.

Now you off brand users can go back to parroting DPR and others in their ridiculously biased Pentax coverage, and pretend like their treatment of Pentax is fair confident in your ability to not be challenged, secure in your half truths and outright lies. You can continue to blame your gear for your photographic failures, even though others on the site succeed on a regular basis.

Funny how Pentax users put up with this kind of bias all the time. Users of other brands just can't handle it. They need hand holding and expert confirmation of their choices. Pentax users not so much. We buy our gear after dissecting and exposing the bias of most sites. We don't need no phony "experts" telling us we made the right choice. But it's still nice when it happens.

So shoot us.

[luftfluss]

So we loved him because he supported our brand, and now we don't because he doesn't.

No one should have been loving this dude at all, because his testing was sloppy. Pentax users, who have so often been witnesses to the pitfalls of poor testing methodology, should know better.

[house]

This is almost certainly a profiling issue. The black points for red, green and blue channels can make a big difference regarding colour accuracy with heavy shadow recovery. If the black points set in the profile used by the application are slightly inaccurate - or there's a minor variance in the colour and luminance data recorded by different copies of the same model camera (which is a possibility) - then you'd get a colour cast when there's an extreme boost to exposure or low-end shadow recovery.

I see this with my Hasselblad HV in Lightroom 6 (one of the few raw converters that has a ready-made profile for this camera). If I have a shot with a deep black background - for example, the cloudless sky of a moon shot exposed for the moon - and boost the exposure by +4 or +5 EV, I get a magenta tint on the black background. The solution - not a perfect one, but it works quite well - is to adjust the shadow tint towards the green very, very slightly. RawTherapee, by comparison, offers full control over the individual black-points for red, green and blue channels. It's more difficult to find the precise settings, but possible to get it near-enough perfect.

Given that the issue only appears with big adjustments to exposure and/or shadows, the only time I can realistically see this being a practical issue is if you're shooting at a low ISO and relying on so-called "ISO invariance", then increasing exposure as required in the raw converter. It's an approach I've never liked, even on cameras that reviewers claim are ISO invariant (or very nearly so). There's always some visible loss in quality compared to shooting at the correct ISO...

I agree completely with your analysis including that you have to push extreme amounts to get there. At least for most cases you can probably sort it out in Rawtherapee as you explained. This is a huge issue with the Adobe centric testing that completely dominates camera and lens reviews. Getting the Adobe settings right makes more of a difference than improving the actual products.

With cameraville this was only one of many, many issues with his reviews. He was equally bad when he was pro Pentax. Something any critical viewer/thinker should have realised. Thing is though that the tiny differences nowadays demand quite rigorous setups and review of results to ensure they are at all meaningful. Very few sites manage this. Talking head youtubers will inevitably fail but can provide other content.

No one should have been loving this dude at all, because his testing was sloppy. Pentax users, who have so often been witnesses to the pitfalls of poor testing methodology, should know better.

I agree.

---------- Post added 01-09-22 at 12:08 PM ----------

Here is a folder with four K-3 Mark III DNG files, each underexposed something like three stops as per the camera's indications. Just some random things around my house. The folder also includes jpegs processed in RawTherapee, three of them pulled up three stops, the one numbered 6718 was pulled about two and then shadows also pulled up. The one at ISO 100 (6719) also had shadows pulled up in addition to the three stops of exposure compensation.

I don't see any issue with the camera's dynamic range, I'd have no problem using those pictures. Pulling an ISO 400 shot three stops should be about like ISO 3200, and those look pretty clean to me without much loss in dynamic range or addition of noise. But there are the files for anyone who wants to play with them.

Excellent!

Pushed one of the darkest files from very dark to completely blown out without introducing any cast what so ever. A small amount of noise in the darkest parts was all that happened. This was Rawtherapee that currenly uses the embedded colour matrix so it's not even special on that front, someone with adobe software should try a 10 ev push and tell us what happens.

---

Last edited by house; 01-09-2022 at 12:09 PM.

[bobbotron]

someone with adobe software should try a 10 ev push and tell us what happens.

I feel like this is asking a lot of any sensor. If it did come up with really weird results I don't think you can really fault the hardware?

[house]

I feel like this is asking a lot of any sensor. If it did come up with really weird results I don't think you can really fault the hardware?

The point is that the sensor maintains good colour with Rawtherapee. If Adobe bails out it's due to the software not the camera which is useful to know.

[BigMackCam]

I agree completely with your analysis including that you have to push extreme amounts to get there.

See? We do agree on some things Actually, more than you probably realise

Pushed one of the darkest files from very dark to completely blown out without introducing any cast what so ever. A small amount of noise in the darkest parts was all that happened. This was Rawtherapee that currenly uses the embedded colour matrix so it's not even special on that front, someone with adobe software should try a 10 ev push and tell us what happens.

OK... so I can get the magenta tint in recovered shadows... sort of... in one of the under-exposure test files (many thanks to @ThorSanchez for these).

In Lightroom 6 (the latest version I have), I load the file and set the profile to "Embedded":



I bring up the exposure by +5 EV:



Looks good at this size. Let's zoom in 1:1 in the dark, top right corner:



Still looks OK at this size of reproduction. Let's try bringing up the shadows... and at +40 I start to see some magenta tint:



Taking the shadows all the way up to +100, I definitely see some magenta:



Here's the thing, though. If you look closely, the distribution of the magenta isn't even. I think what we're seeing is boosted effects of the image accelerator's colour noise reduction. It's a very clever bit of hardware / software imaging pipeline technology, but one of the things I don't like about it is that it tends to leave blotches of colour - typically magenta and green, but mostly magenta. You can see this if you look at K-1II high ISO raw files with dark areas, and the reason I don't like it is that it's harder to remove these blotches than regular randomly-distributed colour noise. It's less obvious in the KP, strangely. Anyway, I think this is what we're seeing...

... but really... +5 EV exposure boost, and +40 shadow boost on top of that before you see anything... and even then, it's not a problem. At +100 shadow boost on top of the exposure push, it might be considered unacceptable, but who does this in reality? Youtube testers, perhaps...

I think K-3III owners can rest easy... and, as they'd say on the Eurovision song contest: "Cameraville - no points... nul points... keine Pünkte!"

---

Last edited by BigMackCam; 01-09-2022 at 01:41 PM.

[Adam007]

It's not the K-1, particularly when it comes to highlights, but DR is good, and you can certainly recover those shadows.Let me put it this way. the K-iii is now the camera I grab 90% of the time (because the AF and weight are more important than absolute IQ most of the time). I can rarely see a difference when using great glass between this and the K-1.

[wkraus]

Here's the thing, though. If you look closely, the distribution of the magenta isn't even. I think what we're seeing is boosted effects of the image accelerator's colour noise reduction. It's a very clever bit of hardware / software imaging pipeline technology, but one of the things I don't like about it is that it tends to leave blotches of colour - typically magenta and green, but mostly magenta. You can see this if you look at K-1II high ISO raw files with dark areas, and the reason I don't like it is that it's harder to remove these blotches than regular randomly-distributed colour noise.

I totally agree with your analysis. But I think what you see is somewhat exaggerated by your colour balance. Assuming that the second book from left has a white spine, I get a temperature of 5300 and a tint of +18, rather than your +27. And the “Embedded” profile of the K-3 III is not very good in ACR. I get much better results with Adobe Standard since Adobe has provided its own profiles for the K-3 III. It has less saturated colours but that is easy to modify if you wish to do so.

[Hayashi]

Then look at this:

Attachment 561311




In the past I would have ditched this pic as soon as I looked at it: but now that I have started playing with DxO's Photolab 5, I realize just how much can be retrieved.
Now, admittedly this recovery was done through the exposure adjustment, not the shadow adjustment, but it shows how much detail can be recovered in an extremely underexposed pic.

These are from the same file: I cropped when producing the final result.
I am not sure why the extreme underexposure occurred: I think I may have thought my flash was on, and had set exposure to manual thinking I could use the low ISO and rely on fillin flash.

This is wonderful !!

May I ask how much EV level you pushed up the raw file ?

[K2 to K50]

This is wonderful !!

May I ask how much EV level you pushed up the raw file ?

The exposure setting is showing 3.51 increase Hayashi. Must admit when I first loaded the file in (to DxO Photolab 5) I was not expecting to get anything useful. I was most pleasantly surprised.

[2351HD]

Here is a folder with four K-3 Mark III DNG files, each underexposed something like three stops as per the camera's indications. Just some random things around my house. The folder also includes jpegs processed in RawTherapee, three of them pulled up three stops, the one numbered 6718 was pulled about two and then shadows also pulled up. The one at ISO 100 (6719) also had shadows pulled up in addition to the three stops of exposure compensation.

I don't see any issue with the camera's dynamic range, I'd have no problem using those pictures. Pulling an ISO 400 shot three stops should be about like ISO 3200, and those look pretty clean to me without much loss in dynamic range or addition of noise. But there are the files for anyone who wants to play with them.

Thanks for sharing these.

I have just loaded them into Capture One Pro V22, pushed the exposure +4 stops (maximum) and raised the shadows to +100. What can I say? The results for ISO400 are nothing short of exceptional. No banding, no pink, noise is a very fine grain (with no playing around with NR other than the C1 default settings. I can only imagine how good ISO 100 files must look.

I would expect these results from a full frame camera such as a K1 or D850 or Canon R5 of which I have tested all of them (owned a K1 and now have an R5).

Now, not that I think I have ever added +4 to exposure and +100 to shadows. But I have added say +1 or +1.5 to an exposure and then added say +80 to shadows on a difficult landscape scene where grad filters were not an option. It is good to know that the K3-III is capable of this sort of image quality.

I might just have to buy one when they come up on sale next.

[ffking]

It's a sad thing about YouTubers - out in the real world there are probably far more people who dislike Sony, for instance than dislike Pentax, but on YuTube Pentx is ignored (Northrup et al) or mocked (Polin et al) or generally disparaged (Cameraville now) in a way that is totally out of proportion to its current status. Ricoh has done nothing to offend these people, they just pick on Pentax for cheap laughs or to be on the right side or whatever - most of us grew up in a time when you defended the little guy and didn't smarm up to the big guy by taking his side against the little guy and find it distasteful. We know Pentax cameras are not perfect and not for everybody - like every other brand - all most of us want is fair criticism and credit where it's due.

[BigMackCam]

It's a sad thing about YouTubers - out in the real world there are probably far more people who dislike Sony, for instance than dislike Pentax, but on YuTube Pentx is ignored (Northrup et al) or mocked (Polin et al) or generally disparaged (Cameraville now) in a way that is totally out of proportion to its current status. Ricoh has done nothing to offend these people, they just pick on Pentax for cheap laughs or to be on the right side or whatever - most of us grew up in a time when you defended the little guy and didn't smarm up to the big guy by taking his side against the little guy and find it distasteful. We know Pentax cameras are not perfect and not for everybody - like every other brand - all most of us want is fair criticism and credit where it's due.

I don't even care that much if folks criticise and disparage Pentax - so long as they don't do it here in these forums (unless, of course, it's balanced, constructive and warranted, as opposed to personal opinion for the hell of it). That I do find distasteful, inconsiderate and inflammatory...

[repaap]

I’m happy that Thor did give those samples. What canI say, I told so too…

I have Panasonic S5 mostly for video, and I thought that why not to put these cameras head to head(K-1,K-3III and S5) there is just a tad more DR in FF. K-3III is king with ISO and colour precision even when you have to ’push’ shadows. Or whole image.

Now with all other improvements (I still have PLM lenses on my shopping list and will come back after covid stop killing my bank account) it is very capable camera.

S5 is great with video. Colour precision and high ISO with photography is not enough for me.

I stil can recommend K-3III and I’m waiting for K-1III to give us very good FF with AF like K-3III and great other features(I also like how instant and responsive K-3III is

Thanks for sharing these.

I have just loaded them into Capture One Pro V22, pushed the exposure +4 stops (maximum) and raised the shadows to +100. What can I say? The results for ISO400 are nothing short of exceptional. No banding, no pink, noise is a very fine grain (with no playing around with NR other than the C1 default settings. I can only imagine how good ISO 100 files must look.

I would expect these results from a full frame camera such as a K1 or D850 or Canon R5 of which I have tested all of them (owned a K1 and now have an R5).

Now, not that I think I have ever added +4 to exposure and +100 to shadows. But I have added say +1 or +1.5 to an exposure and then added say +80 to shadows on a difficult landscape scene where grad filters were not an option. It is good to know that the K3-III is capable of this sort of image quality.

I might just have to buy one when they come up on sale next.