• 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Forums > Back > Could AA anti-aliasing be done in camera firmware?
#1
Quote:Here the Nikon D200 as example:

http://www.maxmax.com/nikon_d200hr.htm

The above example photos, and the interesting arguments by Lomskij and others in the currently raging Sigma SD1 thread, remind me how much detail we may all be losing thanks to our cameras' AA anti-aliasing filters. Klaus has also published useful information on this matter.



I wonder if, with the increasing CPU power every year, AA could be implemented in software. Something along the lines of an optional blurring function, to be applied to ordinary JPEGs and/or raw files. Have not thought this through, maybe one of you brainiacs have.



And, does anybody know how to remove the Anti-Aliasing filter on my Sony Nex?



And does Klaus and other Photozone folks have an opinion on whether or not it would be field significant to have no AA filters on, say, 30cm x 40cm (hmm that's pretty big) uncropped final prints?
  Reply
#2
[quote name='RussellB' timestamp='1286230042' post='3479']

The above example photos, and the interesting arguments by Lomskij and others in the currently raging Sigma SD1 thread, remind me how much detail we may all be losing thanks to our cameras' AA anti-aliasing filters. Klaus has also published useful information on this matter.



I wonder if, with the increasing CPU power every year, AA could be implemented in software. Something along the lines of an optional blurring function, to be applied to ordinary JPEGs and/or raw files. Have not thought this through, maybe one of you brainiacs have.



And, does anybody know how to remove the Anti-Aliasing filter on my Sony Nex?



And does Klaus and other Photozone folks have an opinion on whether or not it would be field significant to have no AA filters on, say, 30cm x 40cm (hmm that's pretty big) uncropped final prints?

[/quote]

No, you can not really filter afterwards. The aliasing and moire create false detail and patterns that you can not just undo and recreate what should have been there.



It is not really about CPU power, but about procedures that can detect aliasing artifacts, and about procedures that can then "mend" the aliasing... Things are just not that simple.



You do not lose a lot of detail through AA-filters, even though it might seem that way. The example I showed was fro a D200, there are cameras with a bit less strong AA-filter. Most is false detail, though, edges where there are no edges (just because the pixels end there). You can get that apparent detail back with sharpening in software... And that software usually does not create the same amount of aliasing and normally no moire patterns at all.



You should not remove the AA-filter on your NEX. If you actually would want an AA-filter-less NEX, you should get into contact with that company I linked to.



30 x 40 is a 4/3rds or compact camera size <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/wink.gif' class='bbc_emoticon' alt='Wink' />. You have enough detail from any NEX to print that size, no problem. Also when you would use an AA-filter-less camera, the aliasing and moire would not really show up. You would need to print a lot bigger for that to become a problem.



In my opinion, it is a better idea to learn about sharpening in "correct" ways for good prints, which can deliver better results that shooting without AA-filter.



AA-filters do serve a purpose...
  Reply
#3
[quote name='RussellB' timestamp='1286230042' post='3479']

The above example photos, and the interesting arguments by Lomskij and others in the currently raging Sigma SD1 thread, remind me how much detail we may all be losing thanks to our cameras' AA anti-aliasing filters.

[/quote]

Well as you might have noticed my saying in that thread, I have a AA-filterless 50D an I really see no improvements in image quality, even with the best lenses out there. What you see at 100% are slightly blocky details in areas of high-frequency detail (e.g. grass in far distances). Removing the AA filter helps when the AA filter is too strong and the lenses can resolve enough detail. Otherwise, the lens blur acts like a blur filter of its own.





[quote name='RussellB' timestamp='1286230042' post='3479']

I wonder if, with the increasing CPU power every year, AA could be implemented in software. Something along the lines of an optional blurring function, to be applied to ordinary JPEGs and/or raw files. Have not thought this through, maybe one of you brainiacs have.

[/quote]

Have you seen this article?

http://en.wikipedia.org/wiki/Spatial_anti-aliasing#Signal_processing_approach_to_anti-aliasing





[quote name='RussellB' timestamp='1286230042' post='3479']

And, does anybody know how to remove the Anti-Aliasing filter on my Sony Nex?

[/quote]

I can think of 2 reasons why this is not a good move:



1. Very few lenses out there can resolve to the level of individual pixels on a Nex. Without such lenses, you're not really going to see improvements.



2. There are far more important things that affects image quality than the AA filter, in the pipeline. For example, which RAW converter do you use and like what Brightcolours said, how do you sharpen your images? Also how do you stabilise the camera and most importantly which lenses do you use?





[quote name='RussellB' timestamp='1286230042' post='3479']

And does Klaus and other Photozone folks have an opinion on whether or not it would be field significant to have no AA filters on, say, 30cm x 40cm (hmm that's pretty big) uncropped final prints?

[/quote]

Spatial visual information that wasn't there on screen is not suddenly going to appear in prints. What you should really be thinking about is how your printer does any up-scaling. Calculate the pixel resolution needed to print an image to a certain magnification at a certain DPI and then up-scale your source image to that size using the best up-scaling algorithm you can find. This way you're in control of how pixels are interpolated rather than letting the printer decide it. Also you got to take attributes of the paper in to consideration when it comes to all this and how sharpening, etc. should be done.



GTW
  Reply
#4
[quote name='genotypewriter' timestamp='1286242902' post='3481']

Well as you might have noticed my saying in that thread, I have a AA-filterless 50D an I really see no improvements in image quality, even with the best lenses out there. What you see at 100% are slightly blocky details in areas of high-frequency detail (e.g. grass in far distances)...GTW

[/quote]

That answers my question perfectly, and saves a lot of time and thought. If you took the anti-aliasing filter off of your 50D, and no angels appeared, then removing AA filters are not worth thinking about. Certainly a bit less blocky details in high-frequency parts of the image, only visible at huge magnification, are not worth pursuing by removing the AA filter, and needlessly creating moire problems in many different images.



Thanks for the rather elevated discussion, folks.



And it is rather amusing to see in general, how casually people talk about whether or not their particular camera system can produce what sounds to me like flawless 1 meter x 1 meter prints. Am scratching my head to remember when any client of mine in the last 35 years has wanted a perfect-quality final print bigger than 8.5 x 11 inches (25x35cm). Certainly doesn't come up in magazine or web journalism, or wedding albums, or marketing materials (well maybe the occasional two-page spread) that people hold in their hand, or ordinary portraiture. Now I suppose tons of extra resolution or pixels from an unnecessarily-large-format sensor would let you crop a picture a bunch and still have it look great. But statistically am getting a higher percentage of perfectly sharp photos than ever, from the new small (APS-C 16x24mm) decent-ISO-800 sensors. With wonderful depth of field for a given natural light level, F-stop and shutter speed.



Well maybe lots of hobbyists and intense artists are making huge prints for personal pleasure or special effects.

/Ru
  Reply
#5
[quote name='RussellB' timestamp='1286250567' post='3486']

That answers my question perfectly, and saves a lot of time and thought. If you took the anti-aliasing filter off of your 50D, and no angels appeared, then removing AA filters are not worth thinking about. Certainly a bit less blocky details in high-frequency parts of the image, only visible at huge magnification, are not worth pursuing by removing the AA filter, and needlessly creating moire problems in many different images.

[/quote]

The way MPs are headed in Bayer CFAs, I'm sure there will come a time when the necessity to have an AA filter in lower-end DSLRs will be reduced. Maybe that's what manufacturers are trying to do, afterall? <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='Wink' />





[quote name='RussellB' timestamp='1286250567' post='3486']

Well maybe lots of hobbyists and intense artists are making huge prints for personal pleasure or special effects.

[/quote]

For me, it's about making things challenging and getting it right. I don't crop images (or even rotate to fix verticals) or machine gun to capture action. I use only fast primes and I manual focus a lot, even action. I like high resolution because it's easier to get a shot in focus on a lower res camera than on a higher res one... it adds to the challenge like everything else <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='Big Grin' />



GTW
  Reply
#6
[quote name='genotypewriter' timestamp='1286326597' post='3518']

For me, it's about making things challenging and getting it right. I don't crop images (or even rotate to fix verticals) or machine gun to capture action. I use only fast primes and I manual focus a lot, even action. I like high resolution because it's easier to get a shot in focus on a lower res camera than on a higher res one... it adds to the challenge like everything else <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='Wink' />

GTW

[/quote]

- grrrrr -



for two other points of view -



1. if i lose an instant in time for technical or gear reasons, i say why did i bother spending all of my time being there -

2. if i want sharp pictures i buy sharp lenses, they are available for current sensors (well up to 135 anyway[Image: tongue.gif]) . . . and if i'm allowed to get my hands on something longer that is also good it will be nice, thank you -



. . . and this just another pov gtw - nothing more
  Reply
#7
[quote name='anyscreenamewilldo' timestamp='1286351645' post='3520']

- grrrrr -



for two other points of view -



1. if i lose an instant in time for technical or gear reasons, i say why did i bother spending all of my time being there -

2. if i want sharp pictures i buy sharp lenses, they are available for current sensors (well up to 135 anyway[Image: tongue.gif]) . . . and if i'm allowed to get my hands on something longer that is also good it will be nice, thank you -



. . . and this just another pov gtw - nothing more

[/quote]

Yup... it's great that we're all interested in different things <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/smile.gif' class='bbc_emoticon' alt='Smile' />



But with that said... here are a few "moments" I managed to capture manual focused (none of these are cropped too):

http://www.flickr.com/photos/genotypewriter/tags/lisabarroncomau/

http://www.flickr.com/photos/genotypewriter/tags/loreal600mf/

http://www.flickr.com/photos/genotypewriter/4603049471/

http://www.flickr.com/photos/genotypewriter/4496618199/

http://www.flickr.com/photos/genotypewriter/4438229106/

http://www.flickr.com/photos/genotypewriter/4434495917/

http://www.flickr.com/photos/genotypewriter/4431101709/

http://www.flickr.com/photos/genotypewriter/4428389667/

http://www.flickr.com/photos/genotypewriter/4427818563/

http://www.flickr.com/photos/genotypewriter/4365192906/

http://www.flickr.com/photos/genotypewriter/4270783321/

http://www.flickr.com/photos/genotypewriter/3852064595/

http://www.flickr.com/photos/genotypewriter/3808514412/

http://www.flickr.com/photos/genotypewriter/3788275497/

http://www.flickr.com/photos/genotypewriter/3789085646/

http://www.flickr.com/photos/genotypewriter/3770940406/

http://www.flickr.com/photos/genotypewriter/3684695918/

http://www.flickr.com/photos/genotypewriter/3681498843/



<img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/wink.gif' class='bbc_emoticon' alt='Smile' />
  Reply
#8
congrats gtw, these (or at least the ones i've looked at so far) are super pictures, very nice indeed, and thank you -



and i do agree that manual focus is very possible and really not so hard at all with a sharp and fast lens and a good viewfinder (i do find my cz's really easy to manually focus) -



however i was referring more to candids, a face in a moving crowd at random to be picked out in an instant in time, otherwise said face has walked away and is lost forever, and i really don't like the loss of a perfect picture . . . so dare i suggest that auto focus cz's (in my case) have their place too - and i have looked and wish it weren't true but so far currently for me there is nothing else that works, so for me . . . sadness [Image: sad.gif] . . .



but i think i've gone off topic yet again . . . i think we were talking about removing the AA filter - well on the test i looked at i couldn't see the benefit, and so my current point is that a quality sharp lens will do the job on the current sensors, no problem . . . but these lenses that will see through the grime are few not many . . .and i'd like a longer one please [Image: blink.gif]



anyway ta gtw, lovely pictures
  Reply
#9
It's possible to do AA in software (or otherwise, in the signal chain) but to do this the data has to be "oversampled", in other words the sample has to be taken at a frequency (or in this case, resolution) much higher than that which is desired for the final output. I doubt this is something we want at this point in camera evolution.



Haha - I see I'm a newbie again.
  Reply
#10
I saw a few threads about this on other forums. Part of my profession is signal processing (I'm a geophysicist) and I think I should comment here.



When a new camera comes out, people react in two different ways. One group says: "18 megapixels on a crop camera, that's crazy! Useless! You'll run into all sorts of trouble with diffraction and such". Another group says: "Great! Now I must have a set of lenses that squeezes 18 megapixels worth of information out of that sensor on every possible shot. How do I yank out the AA filter?"



The more reasonable way to go about it is like this. I like to make prints a certain side, and 12 megapixels works well for this. So, your best bet is to buy a 18 megapixel camera. It's oversampled but that's okay. This way the blur due to the AA filter, Bayer de-mosaicing, or diffraction become non-issues.



Suppose that, as a geophysicist, I want to record signals up to 80 hertz. I'll sample at 4 milliseconds which gives me a Nyquist frequency of 125 Herz. I'm way oversampled but that's okay. Disk space is cheap. I know that my AA filter has ramps, and while it completely suppresses 125 Hz and up, it still attenuates stuff in the 80-125 Hz range. Being oversampled I can use simpler sample interpolation schemes (think of this being analogous to demosaicing). There are other advantages. Never would I want to design my experiment so that I'm close to Nyquist.



Finally, once you're aliased you've got a big problem. Removing Moire is not a trivial task. Mathematically it's a problem involving non-uniqueness. Like phase unwrapping, there are schemes going back fopr decades in the literature but all are cumbersome, all produce their own artifacts, and none are satisfactory. If you can avoid aliasing in the first place then it's worth it.
  Reply


Forum Jump:


Users browsing this thread: 5 Guest(s)