[normhead]

If you can see the images taken with this Camera (which you can) if people have demonstrated the color and moire issues (which they have) I'm having a bit of trouble seeing how the term "risk" applies. The sample shots on the japan site, the K-5 II's images look remarkably like my K-5 images. The K-5 IIs images look like a few I saw years ago that I was assured were impossible on APS-c. Just on the K-5 II images alone, I'd wait for the end of the product run... then consider it, but I usually I'd wait for the next model after that to be released, and drop in price a bit. A 3-4 year cycle.

The K-5 IIs on the other hand offers a noticeable improvement over what I use now.

Sometimes I think whether or not someone thinks upgrading is worthwhile is completely dependant on whether or not they are looking for a new system. Personally I need an upgrade. At this point my plan is either Nikon D600 for landscape, or K5 IIs.

The advantages of staying with one system and putting money away for the Pentax FF is pretty much a no-brainer, since I've sold images taken with an Optio 9 WR,( I'm not convinced I need an FF camera to produce saleable images.) but I think I'll wait and see if the price on the KIIs falls as we get closer to Christmas. Between now and Christams is sort fo a black hole for decent looking landscapes anyway.In the meantime I'm going to try and not let the jaded opinions of those not actually looking to buy dampen my enthusiasm. Tess and I came back from a shoot recently her with the K-5 and me with my K20D, (a sunset) her images were gorgeous, mine were trash, mine suffered from terrible low light performance, my shadows were black, hers still had lots of detail.

I'm getting a new body, soon. Honestly I've read all the doubts I've heard about the need for no AA filter on APS-c etc and it's not going to make any difference...

If you can benefit from buying a new body... the advice of others who are actually of the same mind is obviously looking back through this post, way more relevant than those who have no interest one way or the other. One group is looking to buy, the other group is looking for reasons not to buy... and never the twain shall meet.

[westmill]

A question for the boffins If i were to have the AA filter removed from my K5 !!!!!! Would the K5 then be able to focus correctly under artificial light ? It seems as if it is something like this that throws it out. Or is it a different filter thats the problem ? Just a thought ! since any competant camera engineer should be capable of removing the AA filter.

[hansenn]

A question for the boffins If i were to have the AA filter removed from my K5 !!!!!! Would the K5 then be able to focus correctly under artificial light ? It seems as if it is something like this that throws it out. Or is it a different filter thats the problem ? Just a thought ! since any competant camera engineer should be capable of removing the AA filter.

good question...
And I'd like to know how you clean the sensor? Normally the sensor is protected by the AA filter when you wipe over it with something.
Without the protective filter will the chance that you do some harm to the sensor increase?

[greg_77]

A question for the boffins If i were to have the AA filter removed from my K5 !!!!!! Would the K5 then be able to focus correctly under artificial light ? It seems as if it is something like this that throws it out. Or is it a different filter thats the problem ? Just a thought ! since any competant camera engineer should be capable of removing the AA filter.

The AA filter is not related to that issue. From what I understand the little lenses over/on the AF sensor aren't achromatic, so they give slightly different focus depending on the redness or blueness of the light.

[normhead]

Just to further that point Pentax added an extra element to the AF system in the K-30 and it continues in the K-5 II s. There's a hardware fix to the AF issues. So, no, it won't be the same thing. You want the new AF system, it's simply better and faster than the old.

[westmill]

The AA filter is not related to that issue. From what I understand the little lenses over/on the AF sensor aren't achromatic, so they give slightly different focus depending on the redness or blueness of the light.

Thank you greg It was worth a thought ! I knew it was some thing along the lines of the different colours focussing at different places etc. I was hoping it was the AA filter. Pity ! I would have had it removed.

[kh1234567890]

Pity ! I would have had it removed.

Being a quite chunky lump of crystal you'd have to replace it with something of the same mass otherwise your shake reduction probably wouldn't work anymore, beside the fact that your OVF focusing would be out unless you put something in its place with about the same refractive index. Your colours would get screwed because you'd be removing the IR filter and any dust and crap would land directly on the sensor, becoming wonderfully in focus and difficult to remove. People who rip the AA filter out usually amuse themselves with taking long exposures of stars, so none of this really matters to them

[Class A]

Pixel pitch isn't the crucial variable if it comes to equivalent cameras. It is MP really.

I now see that approaching the question using equivalent images makes more sense and then obviously MP matter, not pixel-pitch. Thanks.

Personally, I consider a sub-100MP camera w/o Bayer-AA filter technically flawed which is and must be sold with a disclaimer.

I'll quote you with that.

I think the way resolution of Bayer sensors is measured is accurate.

Thanks for the many good points you are making, but I think that my basic reservation still somewhat stands.

My problem is that I understand that the typically measured Bayer sensor resolution limit includes a level of detail that already produces false colours (when using an AA-filter less sensor).

Maybe some RAW converters, such as ACR, even optimise B&W pattern cases and optimistically remove false colours (which obviously doesn't work for full colour patterns)?
Maybe measurement software ignores false colours by using (special) B&W versions of test shots (which seems OK, if you want to measure lens resolution)?
Maybe some tester believe that the best way to measure resolution is to measure the RAW files directly (without demosaicing)?

In any of the above case resolution figures would be exaggerated, AFAIC. I'm just a little (not too much ) concerned that a resolution comparison between say a K-5 IIs and K-5 II could be unfair.

But if an image has been blurred already (btw much more likely with high MP), e.g., defocus, shake, motion, aberration, noise, then a Bayer-AA filter can remove image information which cannot be recovered via sharpening.

Good point, thanks.

I take it you are happy with your 24MP sans AA-filter? Or do you sometimes wish you had went for the D800?

[falconeye]

My problem is that I understand that the typically measured Bayer sensor resolution limit includes a level of detail that already produces false colours (when using an AA-filter less sensor).

...

I take it you are happy with your 24MP sans AA-filter? Or do you sometimes wish you had went for the D800?

This isn't exactly what I understand when it comes to resolution. You are right that a regular pattern near the Nyquist frequency would produce color moiré and wrong resolution readings. But this isn't exactly the same as resolution.

Resolution is more closely related to the response function for input with fine detail (with high spatial frequency components), like eye iris, hair, sand etc. Tests measure it as the slanted edge spread function and edges produce no false color. The funny thing is that point spread functions (PSFs) would produce false color but ESFs and LSFs do not, due to smart enough demosaicing algorithms. And for 2D images, LSFs are a good representation of resolution. Esp. as the human eye resolves line pairs better than point pairs.

As an example, I like to post a link to my image here:
http://photos.smugmug.com/photos/i-pKBNq4n/0/O/i-pKBNq4n.jpg
It is from a camera w/o Bayer-AA filter (D800E) and I think one can see that the visible resolution at 1:1 is no fake. It looks as crisp at 100% as do smaller versions. And the lens wasn't even a prime lens.

The unavoidable unfair comparison of K-5II vs. K-5IIs will come from testers who fail to adjust sharpening to the strength of the Bayer-AA filter. Not from ignoring false color moiré.

...

24MP? Would be the D600 and it has a Bayer-AA filter. You probably meant the 36MP D800E. Yeah, I am generally happy with it. But if I mainly worked on contract (esp. fashion), I would probably prefer the D800 which has a weak Bayer-AA filter and which actually is well able to still produce color moiré. OTOH, the K-5II has a stronger Bayer-AA filter and fewer MP. So, it might not fit the bill where the D800 would. The K-5IIs then may be an answer.

[Class A]

Tests measure it as the slanted edge spread function and edges produce no false color.

I'm assuming you are saying that because you are implying that a smart demosaicing algorithm is used, is that correct?

The funny thing is that point spread functions (PSFs) would produce false color but ESFs and LSFs do not, due to smart enough demosaicing algorithms.

But when you are relying on smart enough demosaicing algorithms, you are supporting my point, aren't you?

In cases where the demosaicing algorithm has to be smart, in cannot reliably reconstruct the original. Of course, a demosaiced image will never be a 100% reconstruction of the original, but if you respect the resolution limits of the sensor, the deviations will not matter in practice.

My point is that relying on the demosaicing algorithm to avoid false colour or operating on RAW data directly (before demosaicing) creates resolution figures that are exaggerated (because the reconstructed resolution may or may not correspond to the original; an approach like this works if you know what the original looked like so that you can do the correct reconstruction and this is fine for lens measurements, or optimal reconstruction of B&W scenes, etc, but cannot work in general). I appreciate that in practice smart demosaicing works fine but I wonder whether it may impact on a fair comparison between sensors with and w/o AA filters.

24MP? Would be the D600 and it has a Bayer-AA filter. You probably meant the 36MP D800E.

Sorry, I meant your 36MP D800E.

I erroneously wrote 24MP because in the back of my mind I was thinking about the K-3 (24MP, APS-C sensor) and wondered whether an AA-filter less version of it would be satisfactory.

[falconeye]

My point is that relying on the demosaicing algorithm to avoid false colour [...] creates resolution figures that are exaggerated

I had seen your point and it is perfectly legitimate.

However, it isn't mine. Bayer sensors (with or w/o their AA filter) use a trick to record color and fine detail at the same time.

If you convert each 2x2 Bayer cell into 1 RGB pixel, then you not only reduce linear resolution by half, you create edges with ugly color fringes too. The Bayer sensor simply isn't meant to be regarded this way. The separate reconstruction of a full resolution luminance signal and a separate chrominance signal is at the heart of the Bayer filter invention. There is nothing exaggerated.

While your reservation may apply in a theoretical way, I don't think it does so in practice (which is why I supplied a sample photo). The reason is that the human eye is using a similiar trick, i.e., to have a higher luminance than chrominance resolution, and a higher resolution for line than point structures.

[JohnBee]

Is it a wise choice to choose the S model and have moire on your shots? I tried some PP with Photoshop and lightroom 4, and it is very hard if not impossible to correct without a lot of PP like cloning.
The small amount of sharpness over the stand model, can't you sharpen in K5II in PS or lightroom and get the same results?

of course it is all opinions, but all are welcomed!

I'm going to venture: Yes, if you're doing landscape type shooting and: No if you're leaning toward studio and/or portrait type shooting. Having said that, the casual shooter could benefit from the IIs as well.

[Class A]

Bayer sensors (with or w/o their AA filter) use a trick to record color and fine detail at the same time.

We both agree that this works in practice.

If you convert each 2x2 Bayer cell into 1 RGB pixel, then you not only reduce linear resolution by half, you create edges with ugly color fringes too. The Bayer sensor simply isn't meant to be regarded this way.

For the record, I never meant to imply that this approach should be followed. It would be unreasonable to throw away all the extra spatial information.

The reason is that the human eye is using a similiar trick, i.e., to have a higher luminance than chrominance resolution, and a higher resolution for line than point structures.

Point taken, but if resolution figures are expressed in objective technical terms -- as opposed to perception scales -- they need to stand on their own without reference to human physiology.

Overall we do agree, though. Thanks again for your (always) interesting contributions.

[UnknownVT]

Each serves its purpose - the K-5II is for general photography - the K-5IIs is more specialized to capture more detail -
BUT it has some disadvantages without the anti-aliasing filter - in that fine patterns can produce aliasing or moire patterns - it may be possible to remove the moire with post processing - but it is not generally easy and any perceived advantage will probably be negated by the pp.

This is a rather nicely put article on Aliasing and Moire by someone accredited.

Generally I personally would stick with a camera that is general purpose
- in this case the choice would be the K-5II - unless I have a specific specialized need.

If extra detail is really needed - then probably instead look at a higher resolution sensor -
eg: not that I am saying for a moment that these are anywhere near the K-5IIs -
but the Nikon D3200 has a 24Mp sensor and that is a budget dSLR selling for less than $490 with kit lens;
the Sony SLT A65 also 24Mp is less than $690 -
I would venture to say both will out resolve the K-5IIs without the possible aliasing and moire problems
- hope that puts it somewhat in context.

[emtor]

The small amount of sharpness over the stand model, can't you sharpen in K5II in PS or lightroom and get the same results?

No, you can't, since the sharpness was never there (on the sensor) to begin with.
Sharpening in post isn't really sharpening at all- It's just increasing contrast between ligher and darker areas which gives an illusion of sharpness.
The image was sharp when going through the lens, but as soon as it went through the AA-filter that sharpness was lost forever, and no sharpening algorithm in the world can bring the original sharpness back. If you shoot mainly landscape I wouldn't think twice about buying the K5 IIS.