In praise of AA filters

Started Oct 23, 2013 | Discussions
Lee Jay Forum Pro • Posts: 53,178
Re: example

rovingtim wrote:

Lee Jay wrote:

rovingtim wrote:

Lee Jay wrote:

rovingtim wrote:

This is a bird's folded wing surrounded by feathers. The bold diagonals should be plain, but a couple of them have become 'ropey' with a pattern that matches the feathers above them. Why?

That's the point - there's no way to know.

I'm agreeing with you here, Lee.

You can't recover aliased information.

No, but software can 'guess'.

It doesn't. It has no idea what information is aliased and what isn't, and it can't create data above Nyquist because there's no place to put it even if it had that data.

Nyquist is 2x the sampling frequency.

Nyquist is 1/2 the sampling frequency.

On a sensor, that means a single pixel is undersampled.

Yes, if the optics and scene have the ability to provide spacial frequencies higher than half the spacial frequency of the pixels.

Thus the need for an AA. Take away the AA and print a dot for every pixel as if it has all the information and you are effectively imaging beyond Nyquist.

No, you are not.

So, if at 100% everything is crystal clear before you sharpen, then you know that what you are looking at is beyond Nyquist. I can list a number of cameras that do this.

There's no way to do this as it requires a non-causal filter.

How does the camera do this? I'm guessing software invention, but what would you say?

It doesn't do this.

-- hide signature --

Lee Jay
(see profile for equipment)

 Lee Jay's gear list:Lee Jay's gear list
Canon IXUS 310 HS Canon PowerShot SX50 HS Canon EOS 7D Mark II Canon EF 70-200mm F2.8L IS II USM Canon EF 8-15mm f/4L Fisheye USM +23 more
robert1955 Veteran Member • Posts: 3,898
Re: AA filters are necessary even for Sigma cameras

Lee Jay wrote:

Offering me a choice of with and without AA filter would always result in me choosing the camera with the AA filter.
--
Lee Jay
(see profile for equipment)

For now, with a D800/D800E choice maybe yes, with double that number of pixels probably not

 robert1955's gear list:robert1955's gear list
Fujifilm X-E3 Fujifilm XF 16mm F1.4 R WR Fujifilm 50mm F2 R WR Samyang 12mm F2 NCS CS
robert1955 Veteran Member • Posts: 3,898
Re: Pro AA filter users forget Sigma's Foveon sensor

Basalite wrote:

Pro AA filter users either forget or don't know about Sigma"s Foveon sensor. I mentioned it a couple of times in this thread but it needs to be emphasized that the Foveon sensors in Sigma's cameras do not need nor do they have such a filter, for they simply do not suffer from the fine detail drawbacks of Bayer sensors.

A sample image from a Sigma DP3 camera. Get a load of that detail.

You've shown that image several times now, but it is not an image in which moiré would occur

 robert1955's gear list:robert1955's gear list
Fujifilm X-E3 Fujifilm XF 16mm F1.4 R WR Fujifilm 50mm F2 R WR Samyang 12mm F2 NCS CS
Lee Jay Forum Pro • Posts: 53,178
Re: Pro AA filter users forget Sigma's Foveon sensor
2

Basalite wrote:

Tom Axford wrote:

I never said the Foveon couldn't.

Yes, you did.

"Foveon sensors in Sigma's cameras do not need nor do they have such a filter,..."

-- hide signature --

Lee Jay
(see profile for equipment)

 Lee Jay's gear list:Lee Jay's gear list
Canon IXUS 310 HS Canon PowerShot SX50 HS Canon EOS 7D Mark II Canon EF 70-200mm F2.8L IS II USM Canon EF 8-15mm f/4L Fisheye USM +23 more
Lee Jay Forum Pro • Posts: 53,178
Re: Pro AA filter users forget Sigma's Foveon sensor
2

Basalite wrote:

Lee Jay wrote:

Thanks for the example. That X3 image is covered with aliasing artifacts. It may lack most color moire but it's full of luminance moire. The lack of an AA filter is one reason of many I've never seriously considered a Sigma camera and the example above shows why.

I find it mind-boggling that anyone would choose the image on the left over the Foveon.

Who said they would?  I'd prefer a standard Bayer sensor with a quality AA filter over the horrible crunchiness, jaggies, and false detail of the Foveon (or a Foveon with an AA filter).  The one on the left has either a lousy AA filter or no AA filter.

-- hide signature --

Lee Jay
(see profile for equipment)

 Lee Jay's gear list:Lee Jay's gear list
Canon IXUS 310 HS Canon PowerShot SX50 HS Canon EOS 7D Mark II Canon EF 70-200mm F2.8L IS II USM Canon EF 8-15mm f/4L Fisheye USM +23 more
Lee Jay Forum Pro • Posts: 53,178
Re: Is this what photography has come down to??
2

Basalite wrote:

Here's another image from that Sigma camera. Good luck trying to produce that with a Bayer sensor camera.

The window screens on the left show exactly why a Foveon sensor desperately needs an AA filter.  It would be hard to produce an image like this with a Bayer sensor that has a decent AA filter because it wouldn't product the horrible look of those screens.

-- hide signature --

Lee Jay
(see profile for equipment)

 Lee Jay's gear list:Lee Jay's gear list
Canon IXUS 310 HS Canon PowerShot SX50 HS Canon EOS 7D Mark II Canon EF 70-200mm F2.8L IS II USM Canon EF 8-15mm f/4L Fisheye USM +23 more
Lee Jay Forum Pro • Posts: 53,178
Re: AA filters are necessary even for Sigma cameras
2

robert1955 wrote:

Lee Jay wrote:

Offering me a choice of with and without AA filter would always result in me choosing the camera with the AA filter.

For now, with a D800/D800E choice maybe yes, with double that number of pixels probably not

Before I'd rely on diffraction to act as an AA filter, I'd want pixels at 1.5 microns (384MP on full-frame).

-- hide signature --

Lee Jay
(see profile for equipment)

 Lee Jay's gear list:Lee Jay's gear list
Canon IXUS 310 HS Canon PowerShot SX50 HS Canon EOS 7D Mark II Canon EF 70-200mm F2.8L IS II USM Canon EF 8-15mm f/4L Fisheye USM +23 more
(unknown member) Veteran Member • Posts: 5,593
Re: Is this what photography has come down to??

Lee Jay wrote:

Basalite wrote:

Here's another image from that Sigma camera. Good luck trying to produce that with a Bayer sensor camera.

The window screens on the left show exactly why a Foveon sensor desperately needs an AA filter. It would be hard to produce an image like this with a Bayer sensor that has a decent AA filter because it wouldn't product the horrible look of those screens.

LOL. What the hell are you talking about??

In addition to the detractors being in denial or delusional I'll add the possibility of visually impaired.

-- hide signature --

Lee Jay
(see profile for equipment)

(unknown member) Veteran Member • Posts: 5,593
Re: example

Lee Jay wrote:

rovingtim wrote:

Lee Jay wrote:

rovingtim wrote:

Lee Jay wrote:

rovingtim wrote:

This is a bird's folded wing surrounded by feathers. The bold diagonals should be plain, but a couple of them have become 'ropey' with a pattern that matches the feathers above them. Why?

That's the point - there's no way to know.

I'm agreeing with you here, Lee.

You can't recover aliased information.

No, but software can 'guess'.

It doesn't. It has no idea what information is aliased and what isn't, and it can't create data above Nyquist because there's no place to put it even if it had that data.

Nyquist is 2x the sampling frequency.

Nyquist is 1/2 the sampling frequency.

On a sensor, that means a single pixel is undersampled.

Yes, if the optics and scene have the ability to provide spacial frequencies higher than half the spacial frequency of the pixels.

Thus the need for an AA. Take away the AA and print a dot for every pixel as if it has all the information and you are effectively imaging beyond Nyquist.

No, you are not.

So, if at 100% everything is crystal clear before you sharpen, then you know that what you are looking at is beyond Nyquist. I can list a number of cameras that do this.

There's no way to do this as it requires a non-causal filter.

How does the camera do this? I'm guessing software invention, but what would you say?

It doesn't do this.

You're both advocates of AA filters and you still can't agree with each other.

Do you two ever get to enjoy your pictures?

-- hide signature --

Lee Jay
(see profile for equipment)

(unknown member) Veteran Member • Posts: 5,593
Re: Alias artifacts don't only occur in man-made objects

robert1955 wrote:

Basalite wrote:

That doesn't apply to Sigma cameras with their Foveon sensors.

Though they do have monochrome aliasing

Read more carefully what I have written.

(unknown member) Veteran Member • Posts: 5,593
Re: Speaking of delusional...

Great Bustard wrote:

Basalite wrote:

Great Bustard wrote:

Basalite wrote:

rovingtim wrote:

Once you take out the AA, the software of the camera necessarily much finish off partial signals. If it gets it 'right', it is rarely noticed. However, when it gets is wrong, we used to call that an artifact. Regardless, it is software invented 'detail'.

As someone point out to me, ever taken a shot of a stubbled man with a no AA camera and seen the occasional massive beard hair? Ever noticed how a camera with no AA will happily capture sharp detail of brickwork and then suddenly it go blank like someone blurred it in photoshop? Then a little further on, more sharp detail. No gentle transition.

Ever notice that some leaves on the trees look 'crunchy'? Ever notice that the colour shifts in busy areas of a landscape? For example, a clump of trees looks to be in a faintly purple shadow where none exists.

Ever notice that some 'photographers' are constantly striving to improve their ability to overlook the increasing alias artifacts of modern cameras?

That doesn't apply to Sigma cameras with their Foveon sensors.

Indeed. But they do have to contend with other issues which are, arguably, more serious:

http://www.dpreview.com/forums/post/52373209

Nonsense. Anyone who looks at that amazing sample image I posted from a Sigma DP3 and concludes that it has "more serious" issues than the typical blurry mush produced by Bayer sensor cameras is either in denial or delusional.

...who posts a photo of a scene from Camera A and no corresponding photo of the scene from Camera B, and concludes that the sensor in Camera A is superior?

No need to, as there is no 15 megapixel Bayer sensor camera that can even begin to deliver image quality like that.

That said, Foveon sensors, as a general rule, resolve as well as Bayer sensors with double the pixel count. So, we'd compare the DP3 to 30 MP Bayer, in terms of resolution, which only a few current cameras are sporting.

No, you would not. The proper comparison is a 15 megapixel camera against the Sigma to produce the same image size. That just goes to show how crappy Bayer sensor cameras resolve detail, even those without an AA filter.

Also, the lens does play a role, as you may have guessed, and the DP3 has an excellent lens, which makes it an excellent camera for base ISO photography if you like that single focal length that you are locked into ('tis a pity that Sigma does not make those cameras with interchangeable lenses).

That's what the DSLR SD1 with the same sensor is for. There are also a number of lenses for the SD1 that deliver the same sharpness as the DP lenses.

But to deny that Foveon has issues, well, speaking of delusional...

Feel free to quote me even suggesting that.

The Foveon sensor is not good at high ISO. Does that make you feel better?

In short, it's pretty clear that you're a Foveon fanboy, based on your posts in this thread, the likes of which are even known to Dr. Fossum:

You actually expect to be taken seriously calling people such childish names?

No, I am someone that can recognize and appreciate superior image quality when I see it. Sadly most of you would rather talk about "Nyqvist frequency," "spacial frequencies," "sampling frequency," "non-causal filters," "single pixels," etc, etc, ad nauseam. I don't know whether to laugh or to feel sorry for you people.

(unknown member) Veteran Member • Posts: 5,593
Re: Pro AA filter users forget Sigma's Foveon sensor

Lee Jay wrote:

Basalite wrote:

Tom Axford wrote:

I never said the Foveon couldn't.

Yes, you did.

"Foveon sensors in Sigma's cameras do not need nor do they have such a filter,..."

You need to work on your reading comprehension.

-- hide signature --

Lee Jay
(see profile for equipment)

(unknown member) Veteran Member • Posts: 5,593
Re: Pro AA filter users forget Sigma's Foveon sensor

Lee Jay wrote:

Basalite wrote:

Lee Jay wrote:

Thanks for the example. That X3 image is covered with aliasing artifacts. It may lack most color moire but it's full of luminance moire. The lack of an AA filter is one reason of many I've never seriously considered a Sigma camera and the example above shows why.

I find it mind-boggling that anyone would choose the image on the left over the Foveon.

Who said they would? I'd prefer a standard Bayer sensor with a quality AA filter over the horrible crunchiness, jaggies, and false detail

I don't think you have any concept of what fine photographic detail is.

The one on the left has either a lousy AA filter or no AA filter.

A strong enough blur filter to hide that mess doesn't exist.

-- hide signature --

Lee Jay
(see profile for equipment)

(unknown member) Veteran Member • Posts: 5,593
Re: Pro AA filter users forget Sigma's Foveon sensor

robert1955 wrote:

Basalite wrote:

Pro AA filter users either forget or don't know about Sigma"s Foveon sensor. I mentioned it a couple of times in this thread but it needs to be emphasized that the Foveon sensors in Sigma's cameras do not need nor do they have such a filter, for they simply do not suffer from the fine detail drawbacks of Bayer sensors.

A sample image from a Sigma DP3 camera. Get a load of that detail.

You've shown that image several times now, but it is not an image in which moiré would occur

Obviously that image wasn't a demonstration for moire. Did you not read the rest of the text??

I've also showed another image that perfectly shows the difference in moire.

(unknown member) Veteran Member • Posts: 5,593
Re: Pro AA filter users forget Sigma's Foveon sensor

Lee Jay wrote:

Tom Axford wrote:

rovingtim wrote:

Example:

In the middle of this crop, look at how the detail dissolves in patterns that do not exist in the fabric. In other words, the sensor is trying to resolve detail with partial information. The software constructing the missing 'detail' is getting it wrong.

Thanks, I see what you mean. I was thrown by the use of the word 'moire' which I think perhaps should be reserved for the larger-scale patterns produced by aliasing.

Yes, I said moire when I meant aliasing.

These effects seem to be fairly well muted, however. I think Sigma must have taken steps to alleviate the problems or they could be far worse.

The only step they take is large pixel areas which are themselves a very poor form of AA filter.

On which sensor would that be? Foveon has had three sensor sizes, 3MP, 4.6MP and currently a 15.3MP. The "pixel areas" have only gotten smaller.

At the very least educate yourself about the product you, and others, have been trying so hard to criticize and discredit.

-- hide signature --

Lee Jay
(see profile for equipment)

Lee Jay Forum Pro • Posts: 53,178
Re: Pro AA filter users forget Sigma's Foveon sensor

Lee Jay wrote:

Basalite wrote:

Lee Jay wrote:

Thanks for the example. That X3 image is covered with aliasing artifacts. It may lack most color moire but it's full of luminance moire. The lack of an AA filter is one reason of many I've never seriously considered a Sigma camera and the example above shows why.

I find it mind-boggling that anyone would choose the image on the left over the Foveon.

Who said they would? I'd prefer a standard Bayer sensor with a quality AA filter over the horrible crunchiness, jaggies, and false detail

I don't think you have any concept of what fine photographic detail is.

The one on the left has either a lousy AA filter or no AA filter.

A strong enough blur filter to hide that mess doesn't exist.

You don't know a thing about what you are taking about. You're just a Sigma fanboy posting very poor samples taken with Sigma cameras.
--
Lee Jay
(see profile for equipment)

 Lee Jay's gear list:Lee Jay's gear list
Canon IXUS 310 HS Canon PowerShot SX50 HS Canon EOS 7D Mark II Canon EF 70-200mm F2.8L IS II USM Canon EF 8-15mm f/4L Fisheye USM +23 more
Lee Jay Forum Pro • Posts: 53,178
Re: Is this what photography has come down to??
2

Basalite wrote:

Lee Jay wrote:

The window screens on the left show exactly why a Foveon sensor desperately needs an AA filter. It would be hard to produce an image like this with a Bayer sensor that has a decent AA filter because it wouldn't product the horrible look of those screens.

LOL. What the hell are you talking about??

You honestly don't see any trouble here?

-- hide signature --

Lee Jay
(see profile for equipment)

 Lee Jay's gear list:Lee Jay's gear list
Canon IXUS 310 HS Canon PowerShot SX50 HS Canon EOS 7D Mark II Canon EF 70-200mm F2.8L IS II USM Canon EF 8-15mm f/4L Fisheye USM +23 more
Great Bustard Forum Pro • Posts: 41,905
Re: Speaking of delusional...

Basalite wrote:

Great Bustard wrote:

Basalite wrote:

Great Bustard wrote:

Basalite wrote:

rovingtim wrote:

Once you take out the AA, the software of the camera necessarily much finish off partial signals. If it gets it 'right', it is rarely noticed. However, when it gets is wrong, we used to call that an artifact. Regardless, it is software invented 'detail'.

As someone point out to me, ever taken a shot of a stubbled man with a no AA camera and seen the occasional massive beard hair? Ever noticed how a camera with no AA will happily capture sharp detail of brickwork and then suddenly it go blank like someone blurred it in photoshop? Then a little further on, more sharp detail. No gentle transition.

Ever notice that some leaves on the trees look 'crunchy'? Ever notice that the colour shifts in busy areas of a landscape? For example, a clump of trees looks to be in a faintly purple shadow where none exists.

Ever notice that some 'photographers' are constantly striving to improve their ability to overlook the increasing alias artifacts of modern cameras?

That doesn't apply to Sigma cameras with their Foveon sensors.

Indeed. But they do have to contend with other issues which are, arguably, more serious:

http://www.dpreview.com/forums/post/52373209

Nonsense. Anyone who looks at that amazing sample image I posted from a Sigma DP3 and concludes that it has "more serious" issues than the typical blurry mush produced by Bayer sensor cameras is either in denial or delusional.

...who posts a photo of a scene from Camera A and no corresponding photo of the scene from Camera B, and concludes that the sensor in Camera A is superior?

No need to, as there is no 15 megapixel Bayer sensor camera that can even begin to deliver image quality like that.

That said, Foveon sensors, as a general rule, resolve as well as Bayer sensors with double the pixel count. So, we'd compare the DP3 to 30 MP Bayer, in terms of resolution, which only a few current cameras are sporting.

No, you would not. The proper comparison is a 15 megapixel camera against the Sigma to produce the same image size. That just goes to show how crappy Bayer sensor cameras resolve detail, even those without an AA filter.

As for a "proper comparison", well, that depends on the two cameras that you're thinking of buying. For example, if you were debating between an RX100 and DPM3, then a comparison between those two would be "proper".

But when comparing resolving power between Foveon and Bayer, a Foveon sensor, on average, resolves as well as a Bayer sensor with double the pixel count, all else equal.

Also, the lens does play a role, as you may have guessed, and the DP3 has an excellent lens, which makes it an excellent camera for base ISO photography if you like that single focal length that you are locked into ('tis a pity that Sigma does not make those cameras with interchangeable lenses).

That's what the DSLR SD1 with the same sensor is for. There are also a number of lenses for the SD1 that deliver the same sharpness as the DP lenses.

But to deny that Foveon has issues, well, speaking of delusional...

Feel free to quote me even suggesting that.

The Foveon sensor is not good at high ISO. Does that make you feel better?

It is not even as good at base ISO, when comparing to Bayer sensors with double the pixel count.

In short, it's pretty clear that you're a Foveon fanboy, based on your posts in this thread, the likes of which are even known to Dr. Fossum:

http://www.dpreview.com/forums/post/52378381

So, aside from "entertainment value", I don't think there's much more to be gained here.

You actually expect to be taken seriously calling people such childish names?

No, I am someone that can recognize and appreciate superior image quality when I see it.

Superior to what and under what circumstances, exactly?  You see, it's hyperbolic unqualified statements like that that mark you as a fanboy.

Sadly most of you would rather talk about "Nyqvist frequency," "spacial frequencies," "sampling frequency," "non-causal filters," "single pixels," etc, etc, ad nauseam. I don't know whether to laugh or to feel sorry for you people.

As well as statements like that.

crames Regular Member • Posts: 192
Re: Is this what photography has come down to??

Lee Jay wrote:

Basalite wrote:

Lee Jay wrote:

The window screens on the left show exactly why a Foveon sensor desperately needs an AA filter. It would be hard to produce an image like this with a Bayer sensor that has a decent AA filter because it wouldn't product the horrible look of those screens.

LOL. What the hell are you talking about??

You honestly don't see any trouble here?

Rather than screens, they appear to be textured glass panels with a horizontal pattern. The second one from the top has a coarse fishnet behind it. No "trouble" that I can see, unless you think they actually are screens.

Tony Beach Forum Pro • Posts: 11,039
Re: ready to see your artifacts?

Basalite wrote:

Tony Beach wrote:

Basalite wrote:

I simply enjoy by far the best detail that can be captured with any camera.

Hasselblad users take note, this guy's camera captures more detail (not).

For an equal amount of pixels it certainly does.

You need to be more careful with your incessant fanboy proclamations because you said your camera captures the "best detail" of any camera. If half the photographers in the world actually agreed with what you claimed then half the photographers in the world would be either using or thinking about using your camera, and that is currently not the case.

There is no higher resolution (not megapixel count) sensor per area than a Foveon sensor.

If you want to play that game then I bet tiny little compacts capture more detail per sensor area than your camera; but I would say the more relevant comparison is how much detail the camera captures with the lens you use on it (i.e., the whole system), and that being the case your camera falls short of the D800/D800E as well as Pentax and Hasselblad medium format cameras. Nonetheless, playing your game and looking at DPR's review of Nikon's D7100, I would say that actual resolution between that camera and the SD 1 Merrill are very nearly identical and pretty much constrained by the lenses and apertures (which gets back to the OP and why Nikon took out the AA filter on the D7100). Resolution, especially resolution as measured by sensor area, is not what I would consider a persuasive reason to choose a camera.

If you backed down from your clear fanboy mentality, and said "best in class" resolution then your proclamations would be far more credible if not any less tiresome.

Keyboard shortcuts:
FForum MMy threads