[Nicolas06]

ClassA you are really funy. You imply the whole industry does it wrong, despite the fact that obviously the engineers working on this should have all a great knowledge on the subject.

You deny also the pratical experience of shooters that use filterless camera for quite some time and are all happy of it.

The example you give (like To Anti-Alias or not Anti-Alias ~ Frontal Lobbings) show a soft picture with false colors with a low pass filter (so the filter failed) and 2 oversharpened pictures with or without low pass filter full or artefact in both cases. I never seen any of my pictures looking like that.

in your explanations that a sensor with low pass filter need sharpening, you forget to say that the raw converter already does that and adapt the default sharpnening setting to it, thanks to the profile it associate with the camera. Typically if you go much futher than the default settings, the picture just look worse with much more artifacts, exactly as in your example. The sharpened picture with a low pass filter look really bad.

Finally, moire removal can be fully automated in post. DxO does just that (and I would guess lightroom too). if you wish too. You just need to enable it and that's done. You can add it to your default preset and add moire removal to the other stuff done to every image like geometrical corrections, removal of chromatic aberations, vigneting. If you wanted to batch it to 1000 pictures that of course possible.

---

Last edited by Nicolas06; 12-06-2015 at 03:46 AM.

[Rondec]

I would just say that it isn't a big deal either way. Cameras which have AA filters take more sharpening than cameras that don't, but in the end, you have a really hard time telling the difference between them -- except if you have moire.

On the other hand, there is no doubt that removing the AA filter improves lens performance on DXO Mark's testing. So I guess there is that.

[Fries]

I've noticed a considerable amount of moire in a jacket a guy was wearing. But that was with video. In the user setting reserved for video I've included the the AA filter simulation as a standard precaution.

[Class A]

ClassA you are really funy. You imply the whole industry does it wrong, despite the fact that obviously the engineers working on this should have all a great knowledge on the subject.

The "engineers working on this" have been using Bayer-AA filters for over a decade. Filterless options are a novelty, relatively speaking.

Do you think it took the engineers more than ten years to realise that they were adding two precision layer parts to each sensor thus adding material and production cost, for no good reason?

You may say hat increasing the number of pixels on a sensor made it possible to forgo the Bayer-AA filter but 16MP or 24MP on APS-C is nowhere near enough to rely on lens blur only.

Filterless options are catering to what customers ask for, they are not the better engineering option.

[Nicolas06]

The "engineers working on this" have been using Bayer-AA filters for over a decade. Filterless options are a novelty, relatively speaking.

Do you think it took the engineers more than ten years to realise that they were adding two precision layer parts to each sensor thus adding material and production cost, for no good reason?

You may say hat increasing the number of pixels on a sensor made it possible to forgo the Bayer-AA filter but 16MP or 24MP on APS-C is nowhere near enough to rely on lens blur only.

Filterless options are catering to what customers ask for, they are not the better engineering option.

You have to start somewhere and refine your design. The theory said it was better and the post processing software available for Raw processing was really crude.

Now you can get most issues corrected automatically, even in camera and raw processing software have profile for each camera/lens combination (like DxO) and adapt the algorithms accordingly. Removing moire is just a setting away when it happen (and it also happen with camera with low pass filter, it just a bit less intensive) and the likelyness that the default demoising algorithm got it wrong to begin with is much less than a few years ago.

I had issue with visible moire at time with K5, never got it with K3... Should be the contrary I suppose, except if the algorithms improved and that the resolution of 24MP actually help. I guess DxO is quite capable software that improved significantly over time.

I also think they didn't go direction from a full low pass filter to no pass filter overnight. They used weaker and weaker low pass filter over the years to get more sharpness and analysed the results it gave as well as of course their clients reaction. DSLR aren't also the only users of sensors and they capapitalized on the experience accumulated on photophones and smartphones. There more than 1 billion of such device sold each year with a global market of around 300Billions$/year. That's 50-100 times more unit sales than DSLR and still 10-20 times more than all cameras sold. That give quite some opportunity for R&D and now most sensor advences are announced with smartphones in mind rather than DSLRs.

[Class A]

You have to start somewhere and refine your design. The theory said it was better and the post processing software available for Raw processing was really crude.

That's apparently what you believe.

The only problem with it is that it is wrong and that you won't be able to quote any reliable sources to corroborate this and your other theories.

If what you say were true, why did Canon recently release both Rebel T6i and T6s with Bayer AA-filters?

Why did both Nikon and Pentax release standard models (D800 & K-5 II) next to their filterless options (D800E & K-5 IIs)?

Why do you think Pentax developed an AA-filter-simulator, if the answer is in RAW conversion?

All that doesn't make sense if RAW converters could handle the problem properly.

Please read about what "aliasing" means in information theoretic terms and you'll see that it cannot be undone after it occurred. Analog filters in audio equipment were not reduced before oversampling became feasible.

The equivalent of audio oversampling in photography is to increase the number of pixels on the sensor. 16 MP (K5- IIs) or 24 MP (K-3) are nowhere near enough to forgo a Bayer-AA filter, as the recent 24 MP Canon T6 models with filters demonstrate.

[pinholecam]

I did some checks between a modified no AA filter A7 and one that has one.



Pentax SMC K24/2.8

Full Image




Center Crop







IMO, there is a 'hint of fuzziness' with the AA filter.
However, w/o a side by side comparison, its often hard to tell.

As for moire, its all up to the individual.
I do find it often on repeated patterns like tiling, tiled roofs, repeated building lines which become small/close enough to trigger it from a distance.
However, I've never found it to be particularly bothersome to an image as a whole.
YMMV.

[Nicolas06]

That's apparently what you believe.

The only problem with it is that it is wrong and that you won't be able to quote any reliable sources to corroborate this and your other theories.

If what you say were true, why did Canon recently release both Rebel T6i and T6s with Bayer AA-filters?

Why did both Nikon and Pentax release standard models (D800 & K-5 II) next to their filterless options (D800E & K-5 IIs)?

Why do you think Pentax developed an AA-filter-simulator, if the answer is in RAW conversion?

All that doesn't make sense if RAW converters could handle the problem properly.

Please read about what "aliasing" means in information theoretic terms and you'll see that it cannot be undone after it occurred. Analog filters in audio equipment were not reduced before oversampling became feasible.

The equivalent of audio oversampling in photography is to increase the number of pixels on the sensor. 16 MP (K5- IIs) or 24 MP (K-3) are nowhere near enough to forgo a Bayer-AA filter, as the recent 24 MP Canon T6 models with filters demonstrate.

1) Canon is quite late to the game, their first model will be the newest FF and it will be more expensive than the one without low pass filter. This was the case for D800e and K5-IIs too back in time.
2) New models for Nikon/Sony/Pentax are now filterless by default.Pentax offer the option because it is easy to have it and it allow them to shut up people like you that are never happy even if most people just never enable the feature.
3) The filter only help with color. It doesn't affect how aliasing is handled for luminance.

[Rupert]

I would have never guessed that there would be so much controversy over a filter vs no filter! I guess it's sorta like squirrels....you like them or you don"t.

Regards!

[Aksel]

I would have never guessed that there would be so much controversy over a filter vs no filter!...

Isn't it a good opportunity for producers to prepare similar cameras: the only difference - some with AA filters, other without it!

[GlennG]

I would just say that it isn't a big deal either way. Cameras which have AA filters take more sharpening than cameras that don't, but in the end, you have a really hard time telling the difference between them -- except if you have moire. On the other hand, there is no doubt that removing the AA filter improves lens performance on DXO Mark's testing. So I guess there is that.

Back in early 2013 I finally found a good used K5 classic at a great price. At that time, the K5II and K5IIS were still relatively new and pricey, and I wondered if I should just wait a lot longer for the K5IIS price to come down, but I was ready to step up at that time. I checked the no AA K5IIS photo examples that were being posted at the time and concluded that there seemed to be very little visual difference that I could tell, and most of that difference could be treated with good sharpening techniques. So I thought at the time that if that is all there is - just go ahead and buy the good value used K5. It was really only the AA vs no-AA difference that I was concerned with. I'm very happy with the K5 classic.


Since that time, however, I've heard of nothing but raves from K5IIS owners about how much better and clearly noticeable the difference is compared to the filtered K5. And, DXO stating that it is some 20% better. If it was that much better, shouldn't I have been able to easily discern the difference in early 2013? Seems like I should have been. And yet, I did not. And, in addition, you would have thought that if the improvements were that readily visible, we would not have "endured" a decade or more of the horrible AA-filtered cameras that the engineers were gearing up for us. Were they duping us this whole time? And, their companies? I don't know, man, maybe they were the same engineers that were designing diesel engines for VW. I think there probably is a difference from removing the AA filter - it is just not that much.

I would have never guessed that there would be so much controversy over a filter vs no filter! I guess it's sorta like squirrels....you like them or you don"t.

Yes, I like them. You know, however, that they are just rodents with furry tails, don't you? Pretty cute, however. One fell out of a tree near me recently. That does not happen very often. After a short while he got up, shook it off, and went back to his business.

[Rupert]

I'm very happy with the K5 classic.

I was happy with mine too but the K5IIs pleases me more in the AF arena....the IQ may or my not be better......I can't honestly say I see a ton of difference....20%?........that would be hard for me to see. My K5 was very good, no excuses needed for it!

Yes, I like them.

That's good!

You know, however, that they are just rodents with furry tails, don't you?

May be.....but they are smarter than a lot of people I know, I guarantee!

One fell out of a tree near me recently. That does not happen very often. After a short while he got up, shook it off, and went back to his business.

I've seen it here many times over the years.....lots of squirrels here. I saw one fall from over 50 feet up in a big oak tree and land with a loud thud on my deck. Certain he was dead, in a few minutes I went to my shop to get a shovel to bury him. As I picked him up, he shook off and scurried away up the same tree he fell out of. They are extremely resilient!

Best Regards!

[Nicolas06]

Since that time, however, I've heard of nothing but raves from K5IIS owners about how much better and clearly noticeable the difference is compared to the filtered K5. And, DXO stating that it is some 20% better.

The whole point then is to be consistant. That 20% more "perceptual MP" on great lenses like DA35 macro. Not all. So 10% resolution.

If you don't care if you lens is doing 2000 or 2200 on photozone or other (10% difference) and if you don't care if your sensor is 13MP or 16MP (10% resolution) or 16MP vs 20Mp (again 10% resolution) then there no reason for you to see that much more 10% comming from the removal of a low pass filter.

But this also combine of course, 10% from removal of low pass filter, 10% from a better lens, 20% from 24 vs 16 MP and suddently it is 40% gain, the equivalent of TC or the capacity to print 40% larger. This is no longer that negligible.

But really if you don't crop or print huge you'll not see that much of a difference at all between an original K5 with a zoom and K3 with a prime.

As for sharpening, this add artefacts if abused and can only applied for a limited effect... Sharpening doesn't allow a 55-300 to have the level of detail of DA300. This cannot repeated and tend to kill the bokeh too. In fact one of the things I really like is that I don't need to sharpen results now anymore and I don't get the associated artefacts. This give me much better details that look more natural.

[kh1234567890]

Could someone explain what are these 'weaker' AA filters ?

It's been a long time since I did Physics for a living and I can't get my head round this concept. Isn't the whole idea of the filter to replicate a point image over all four sensels in the colour quad ? How do you make that 'weaker' ?

As for the AA filter itself - I'm happy with my K-5ii. As far as I can tell (real life comparisons seem to be non-existent) any extra sharpness is largely illusory.

[savoche]

Could someone explain what are these 'weaker' AA filters ?

It's been a long time since I did Physics for a living and I can't get my head round this concept. Isn't the whole idea of the filter to replicate a point image over all four sensels in the colour quad ? How do you make that 'weaker' ?

As for the AA filter itself - I'm happy with my K-5ii. As far as I can tell (real life comparisons seem to be non-existent) any extra sharpness is largely illusory.

Can't help you with the "weak" filters. But yeah, the difference in sharpness (false or not) is very small IMO, and what difference there is can be made up for with correct sharpening. What's more, correct sharpening does not affect the out-of-focus areas.

I won't let the (lack of) AA filters influence the choice of my next camera.