[Pål Jensen]

But you got a point: who needs more Mpix? Not a lot of people indeed. Most would trade the added Mpix for better high ISO performance anytime, and that's why there are cameras like the D700 out there.

Less than 1% of all images are shot at extreme ISO values. Everyone would benefit from higher resolution. Thats why megapixels sell.

[Class A]

As has been stated, there is evidence that diffraction issues are showing up in higher MP APSc cameras.

Diffraction occurs independently of the sensor resolution. If we had a sensor that outresolved all lenses in existence then we wouldn't need to increase resolution anymore. But we are nowhere near that. If you take your "diffraction issues" argument to its conclusion then you must go back to 2 (two) MP, because any higher resolution will show softening due to diffraction below (i.e., aperture ratios smaller than) f/22.

My K10D is my macro camera and produces a sharper ISO100 image than the K20D can. I can easily see it in prints and on the screen. That may be CCD vs CMOS to a degree as well but I suspect MP has a lot to do with it.

Resolution cannot be the culprit if you compare at the same output size. If you look at 100% crops, of course the higher resolution sensor will look worse. There are a number of factors that could come into play even when you compare at the same output size (which requires downscaling of the higher resolution image which increases contrast and removes noise), e.g., sensor technology, but resolution as such cannot be the culprit.

But that being said, I've printed K20D images up to 60" on the long side.

To print this with 200dpi (300dpi is the professional standard) you would need a 96MP sensor. I realise that such a big print will imply a larger than usual viewing distance and that you don't need 200dpi to make it look good from a distance, but if you are getting close your 76dpi won't look good.

What benefit is higher MP?

To print 60" at more than 76dpi and to be able to crop an image and still print it large. Sometimes you want to crop because you found an image within an image. Sometimes you want to crop to gain reach. If you have a very good lens then cropping to get more reach is much better than degrading image quality by using a teleconverter.

I'm more inclined to listen to someone's opinion about photographic equipment if they have some work to show that they are serious about the art.

Sure, that's a valid position; you are of course free to disregard any opinions by non-photographers. However, one also has to accept that some people are more about the gear than the images and that they are fully entitled to that attitude. Surely the camera manufacturers don't mind the business with them. N.B., this has nothing to do with Ogl. I'm not making a judgement about him being a gearist or not. All I'm saying is he is entitled to his opinion independently of his qualities as a photographer (which may be excellent).

[Digitalis]

"My K10D is my macro camera and produces a sharper ISO100 image than the K20D can"

I have seen the same thing with My k-7, but I suspect the pixel count has nothing to do with it. I suspect the AA filter is too hard, this is most apparent with my use of the Pentax 50mm f/1.2 on the K-7 the images just look soft, and lacking per pixel accutance. But on the k10 with it's weaker AA filter the images are crisp, low in contrast as an inherent characteristic of the SMC 50mm f/1.2 but the images have higher pixel acutance on the K10.

[ogl]

Resolution cannot be the culprit if you compare at the same output size. If you look at 100% crops, of course the higher resolution sensor will look worse. There are a number of factors that could come into play even when you compare at the same output size (which requires downscaling of the higher resolution image which increases contrast and removes noise)

Naked resolution is not interested at all.
Strong АА filter can spoil IQ very easy.
K10D/K200D/K-m/K-x are the best cameras from Pentax in terms of sharpness.
Why? Weak AA filter.

If 21 MP APS-C camera won't have AA filter, it could be cool...
But I doubt that we will see 21 MP APS-C.

[cbaytan]

If 21 MP APS-C camera won't have AA filter, it could be cool...
But I doubt that we will see 21 MP APS-C.

Bigger pixels or more photon collecting micro-lens design will be more sensitive to the light, so, a 12 or 15 MP FF camera with a weak AA filter would be cooler, and that would deserve the name either K-5 or K-3, if a 21 or more MP FF camera arrives in the future that'd be K-1. Just dreaming though.

I wish I could have had my K-7's AA filter taken out to see what happens.

Nikon D200HR

[Class A]

Strong АА filter can spoil IQ very easy.
K10D/K200D/K-m/K-x are the best cameras from Pentax in terms of sharpness.
Why? Weak AA filter.

What is the best approach to having no AA filter at all?
You guessed it: Having a high resolution sensor.

P&S cameras often have no AA filter because the lens already performs that job. The sensors are much more resolving than the tiny lenses used. The higher the resolution, the weaker the AA filter can become.

[falconeye]

I wish I could have had my K-7's AA filter taken out to see what happens.

Ah, that's easy.

Just look at a frame from a 1536x1024 K-7 video.

No AA filter means lot's of ugly color moiré and artifacts. The appearant softness from an AA filter is easily removed with an USM radius 0.5 filter. But not vice versa. dSLR video shooters are desperate to find an AA filter replacement. Guess why?

But as long as uneducated photographers cry for a weak or no AA filter the industry will listen and give us dirty colors in fine textures. Thank's everybody.


Note:
I attach the sample from the Nikon D200HR site, but with applying proper sharpening to the AA side (the artifacts are from the JPG compression, not the sharpening). I largely prefer the AA result: Crisp detail with proper color rendition. Look at the letters from the missing AA side: all text letters have an ugly color cast. Which is typical.
BTW, the 4th and 5th line pair field in the samples is beyond Nyquist and a perfect rendition is smooth gray. The extra texture you see in the non-AA sample is called "Moiré", not "detail" ...

---

Last edited by falconeye; 06-15-2011 at 05:29 AM.

[gazonk]

Therefore, a Bayer sensor's RGB output normally looks very close to a full color sensor's one. Because the human eye primarily judges sharpness in the luminosity channel. The human eye actually plays similiar tricks to a Bayer sensor

Thanks for the interesting comment. But I still wonder: Will this apply to texture of colored (especially natural) objects as well? If you look at 100% crops of typical Canon images on e.g. imaging-resource, black on white (e.g. small letters) looks crazy sharp, but textured objects get that famous "plastic look".

As a rule of thumb it has now become a tradition to treat a Bayer sensor with N pixels and a full color sensor of 1.5N pixels to be of equivalent resolution. So, you would need a 18MP camera to exploit the full potential of a 12MP display.

Even if you consider the types of subjects where the 645D is supposed to excel - e.g. textures of leaves?

Anyway, even with those numbers, and since I expect 12mp displays to be common in the not too distant future, more than 15 mp in the camera will soon be useful even for on-screen viewing of uncropped images. But for my current laptop display (1.7mp) a 3 mp sensor would suffice... (an aside: if only those supercompact P&S cameras were limited to 3 mp, just think how good they would have been!).

[Class A]

But as long as uneducated photographers cry for a weak or no AA filter the industry will listen and give us dirty colors in fine textures.

I find it hard to believe that the industry will act according to the whims of uneducated photographers. There must be other reasons at play.

Do you think Leica did a user poll when they decided to not use an AA-filter in the M9? This is not a rhetorical question; I really don't know what their real considerations were (they claim they didn't want to lose sharpness).

Do you disapprove of Pentax's decision to omit an AA filter for the 645D? Perhaps it is relatively safe to do with the old MF glass? Leica should be in more trouble with their glass but seem to get away with it.

In principle, I agree with you. I'd rather capture sharpen than frequently face ugly colour moiré. However, given sufficient resolution, the lens does enough blurring to obviate the use of a further AA filter. So a bayer array sensor should better have loads of MP (say 50 ). A Foveon type sensor should be able to get away with less MP since luminance moiré is less frequent/disturbing than colour moiré, isn't it?

[ogl]

I think 21 MP camera from rumor is fake.

[ogl]

Ah, that's easy.


No AA filter means lot's of ugly color moiré and artifacts.

For me, M8/M9/645D/K10D/K200D and DMF cameras's (without and with weak AA filters) IQ are outstanding.
Moire is not problem. I even like moire.

[wll]

If the new sensor is 21 MP, in the aps-c mode how many pixels would you have ?

I think all FF cameras allow for aps-c lenses to be used ?

wll

[cbaytan]

As an uneducated user I sharpened other picture from Nikon HR page, with Photoshop.

Amount 200 %
Radius 0.5 Pixels
Threshold: 50 Levels

Check with the Jackson's cape!

[falconeye]

black on white (e.g. small letters) looks crazy sharp, but textured objects get that famous "plastic look".

I think the plastic look is rather due to color noise reduction overdoing its job.

But you're right, colored textured objects like a flower bloom have less resolution.

There must be other reasons at play.

Do you think Leica did a user poll when they decided to not use an AA-filter in the M9?
[...]
Do you disapprove of Pentax's decision to omit an AA filter for the 645D?

Ok, once we get away with the "AA filter is evil" attitude we can discuss


There are five factors which play against each other:

1. Luminance Moiré. This is a problem which goes away by itself with increasing number of sensor pixels.
2. Color Moiré. This creates plain ugly images, even when scaled down to web size. It can be ugly enough to not enter a photo competition.
3. Demosaicing algorithm. It is much more challenging to write a good demosaicer for a weak AA filter. But progress has been made in this domain and e.g. Leica and Hasselblad claim to recognize and eliminate color aliasing in their firmware demosaicers. Pentax does not claim so.
4. Importance of ideal situation resolution. Some cameras are sold only to deliver twice the resolution at 5x the price. E.g., in fashion photography, if a texture creates ugly color moiré (and it does, I have seen MF samples!), you reshoot and change the lightning or texture.
5. Yes, some lenses act like an AA filter. But in general, lenses still outresolve the sensors by a large margin (in the image centers, I mean). But it's not just the lenses. With increasing pixel density, the normal blur (non flash and mild defocus) act as a weak AA filter too.

So, you're right, all factors combined mean that AA filters are getting less important. But they are not evil.

BTW, you can order the 645D with or without an AA filter. So, there is no Pentax decision I could disapprove.

For me, M8/M9/645D/K10D/K200D and DMF cameras's (without and with weak AA filters) IQ are outstanding.
Moire is not problem. I even like moire.

You don't like color Moiré. If you think you do you haven't seen it yet in real photos. Think of beautiful girls with legs in glowing dirty colors ('cause they wear silk stockings)

[RBellavance]

I think all FF cameras allow for aps-c lenses to be used ?

Canon EF-S lenses cannot be used on the 24x36mm cameras, because their rear-element design would "interfere" (i.e. collide) with the mirror. I believe they cannot even be mounted on these cameras.