Time to drop the low pass filters?

[T N Args]

I was browsing the blurb on the new Nikon D810, and it highlights the absence of an OLPF (optical low pass filter).

Why doesn't everyone drop the OLPF from their high end cameras? With the recent decisions by Nikon, they seem to have overcome the need for it, to their own satisfaction at least.

 

[OzRay]

The E-M1 (as well as many other cameras) does not have a low pass filter. It looks like Nikon is playing catch-up.

 

[T N Args]

I'm aware it's not the first.

Is an OLPF and an anti-aliasing (AA) filter the same thing? Because the Nikon D800e is, I believe, AA-free and that was 2.5 years ago.

My question isn't about who was first. I am asking whether some tech progress has occurred, so that the need for these things is completely removed. Or are we into compromise territory? (in which case the advantage is debatable).

 

[OzRay]

AA and low pass are pretty much the same thing. The only reason these filters have been used is because of concerns with moiré, but lately, it seems to be less of a concern; users prefer the better image quality available from cameras without the AA/low pass filter. I think Leica was the first (one of the first) to drop the AA filter.

 

[jnewell]

I'm aware it's not the first.

Is an OLPF and an anti-aliasing (AA) filter the same thing? Because the Nikon D800e is, I believe, AA-free and that was 2.5 years ago.

My question isn't about who was first. I am asking whether some tech progress has occurred, so that the need for these things is completely removed. Or are we into compromise territory? (in which case the advantage is debatable).

AA and low pass are pretty much the same thing. The only reason these filters have been used is because of concerns with moiré, but lately, it seems to be less of a concern; users prefer the better image quality available from cameras without the AA/low pass filter. I think Leica was the first (one of the first) to drop the AA filter.

The need for the OLPF has not changed, but it appears that perceptions about how often it's needed have changed. Moire isn't a frequent problem for most photographers/photographs, whereas maximizing sharpness is nearly always desirable.

 

[OzRay]

The need for the OLPF has not changed, but it appears that perceptions about how often it's needed have changed. Moire isn't a frequent problem for most photographers/photographs, whereas maximizing sharpness is nearly always desirable.

That was basically what I was suggesting. When digital first came out, there was all this concern about moiré and reviewers were highlighting this as something dreaded. When they began to realise the compromises the AA filters created, many started to ask why are we worried about moiré when it appears so infrequently in photographic subjects. Software, I believe, is also able to correct for moiré to some degree as well, so there's really no reason to compromise quality for such small gains.

 

[T N Args]

If the moire risk is overrated, I'm also wondering if the extra sharpness is overrated? The more I think about it, the more I think it's a tradoff that is very small in either case....

 

[OzRay]

Removing the AA filter allows you to capture more detail and improves resolution. It can actually be a big deal for some.

 

[T N Args]

Which one has no AA filter?

I think it is incredibly hard to tell the difference. Here are two 100% crops from RAW test files, I call them The Red and The Green. One of them is a camera with no AA filter, they are otherwise identical (16 MP µ4/3 sensor, same lens, same aperture, same exposure, ISO 200). RAW exported to JPEG at 100% Quality, no other processing ( I did not even edit white balance to match -- just what came out of Raw Therapee).

Can anyone tell which has more detail and resolution?

Subscribe to see EXIF info for this image (if available)

Subscribe to see EXIF info for this image (if available)

[EDIT: I re-posted the above images when I saw a non-zero exposure applied. I don't think it changed the relativities though]

 

[OzRay]

I don't think it's as easy as that to make comparisons. To me the one on the right appears to resolve better, but that could be due to a number of factors (and they don't look anywhere near red or green to me).

 

[Medley]

Here's a picture that I've posted elsewhere that is a classic example of an opportunity for moire, taken with my E-M10, which has no AA filter. This picture is straight out of the camera. No moire on the grill....

Subscribe to see EXIF info for this image (if available)

I don't think they're necessary any longer, but engineers may think otherwise. It always seemed to me that moire was a result of sensors that didn't have sufficiently dense resolution to capture high contrast, fairly frequently alternating pixel brightness. I think (although this is just a gut feeling) that with much higher res sensors available today with pixels of sufficient physical size, this has pretty much disappeared as a major concern. In the old days of digital photography moire was a frequent complaint and so the manufacturers likely dealt with it by adding "anti-aliasing filters", even though no-one knew how the hell they actually worked (probably by optically softening edges). My first digital camera was a Canon Digital Elf with a whopping 2 megapixel resolution. Yes, I had moire.

I don't know if you see it on your particular monitor, but the notebook that I'm using to type this has a 1920 X 1080 screen. When I scroll the picture above I see moire, but when I stop it disappears. Presumably this is because the resolution and refresh rate are not sufficient to accurately render the detail of the picture as it's moving. I'm guessing that if you have a higher res monitor with a faster refresh rate, you may not. If you're using a CRT monitor ("a what?", ask the youngsters) you almost certainly won't see any moire. What I see is effectively moire. I think this is the same basic cause of moire in captured pictures; insufficient resolution combined with insufficiently large pixels on the sensor to capture and present the image at a sufficiently high resolution that my eyes can't see the rough edges that are actually there and the weird patterns that result when they start to mush together. That's just my theory, but my real-world observations are consistent with it.

 

[T N Args]

Well if it's a big deal, like Nikon wants us to believe, the differences have to be evident.

Let see if we can get a few more opinions on the 100% crops in post #9. If there is an even split in opinion, then the usefulness comes into question.

 

[OzRay]

I understand that the benefits of no AA/low pass filter come when pixel densities increase, and m4/3 pixel densities are fairly high. I also understand that you get the full benefits of high quality lenses. I'd need to do some research, as these are things that I've read in the past on sites such as Luminous Landscape (which has often discussed large format sensors that have never had low pass filters). Why would you want to intentionally make you images fuzzy, if you could avoid it? Here's an article from Nikon on the first point: https://nikoneurope-en.custhelp.com...d/56368/~/what-is-an-optical-low-pass-filter?. And an example of the difference between the D800 and D800E: http://www.lifepixel.com/blog/anti-aliasing-low-pass-filter-removal.

 

[T N Args]

Why would you want to intentionally make you images fuzzy, if you could avoid it?

Yes, sound compelling --- like a sales pitch.

But the pragmatic reality lies in the images. Which one is 'made fuzzy' again?

 

[OzRay]

Yes, sound compelling --- like a sales pitch.

But the pragmatic reality lies in the images. Which one is 'made fuzzy' again?

It's not a sales pitch, it's a fact. The AA/low pass filter is designed to soften the image.

 

[dhazeghi]

Here's a picture that I've posted elsewhere that is a classic example of an opportunity for moire, taken with my E-M10, which has no AA filter. This picture is straight out of the camera. No moire on the grill...

There may not be, but the only way to know is to view at 100%. Otherwise, the moire isn't from the image, but from the down-sampling algorithm of your JPEG converter.

All the people who think that getting rid of the AA filter is a big deal should shoot an E-M5 and an E-M10 side-by-side for a bit. The difference between a light AA filter (E-M5) and no AA filter (E-M10) is really very small.

 

[OzRay]

There may not be, but the only way to know is to view at 100%. Otherwise, the moire isn't from the image, but from the down-sampling algorithm of your JPEG converter.

All the people who think that getting rid of the AA filter is a big deal should shoot an E-M5 and an E-M10 side-by-side for a bit. The difference between a light AA filter (E-M5) and no AA filter (E-M10) is really very small.

What about a full AA filter and none, or light AA filter? Also, what sort of scenes are we considering in comparisons?

 

[dhazeghi]

What about a full AA filter and none, or light AA filter? Also, what sort of scenes are we considering in comparisons?

Why does it matter? No manufacturer has used a strong AA filter for some time. Yes, a poorly designed AA filter (like the one on the Olympus E-3) can be strong enough to rob significant detail, but that's not the choice anybody faces today.

 

[OzRay]

So camera manufacturers should revert back to AA filters, as they make no difference to image quality?

 

[T N Args]

It's not a sales pitch, it's a fact.

Many a sales pitch is built on a truth. They just make it more compelling than it needs to be. "Oh, an SACD player doesn't need to have a digital DAC? Wow, who wouldn't want that?" -- "I can plug my Canon camera straight into a Canon printer and push one button to print? Wow, who wouldn't want that?" All facts, but also sales pitches.

The AA/low pass filter is designed to soften the image

.

If I get the idea of an AA filter, then it's designed to have no visible effect at all on sharpness or detail, but that is really hard to achieve in practice. But when someone really perfects an AA filter, it will only function above the Nyquist frequency with no 'spillage' into the lower spectrum. I think.