Ad
Ad
Ad
Pages: [1] 2 »   Bottom of Page
Print
Author Topic: Anti-aliasing filter effects demonstrated  (Read 34239 times)
luong
Full Member
***
Offline Offline

Posts: 208



WWW
« on: October 05, 2007, 01:11:58 PM »
ReplyReply

With all the talk (complaints ?) about AA filter effect on sharpness in some cameras versus others, I'd thought some would be interested in this link where images with and without AA from the *same* camera are shown: http://www.maxmax.com/nikon_d200hr.htm
Logged

John Sheehy
Sr. Member
****
Offline Offline

Posts: 838


« Reply #1 on: October 05, 2007, 01:22:07 PM »
ReplyReply

Quote
With all the talk (complaints ?) about AA filter effect on sharpness in some cameras versus others, I'd thought some would be interested in this link where images with and without AA from the *same* camera are shown: http://www.maxmax.com/nikon_d200hr.htm
[a href=\"index.php?act=findpost&pid=144060\"][{POST_SNAPBACK}][/a]

What the author calls more detail is often actually just artifact.

Also, it doesn't seem to me that the sharpest of lenses was used.  The non-AA images should have been sharper than they were.  Lack of an AA filter is not a big problem when the optics are not optimum.
Logged
feppe
Sr. Member
****
Offline Offline

Posts: 2909

Oh this shows up in here!


WWW
« Reply #2 on: October 05, 2007, 05:12:49 PM »
ReplyReply

Quote
With all the talk (complaints ?) about AA filter effect on sharpness in some cameras versus others, I'd thought some would be interested in this link where images with and without AA from the *same* camera are shown: http://www.maxmax.com/nikon_d200hr.htm
[a href=\"index.php?act=findpost&pid=144060\"][{POST_SNAPBACK}][/a]

I'd much rather see independent research, instead of carefully selected samples from a company which has a vested interest.
Logged

Monito
Jr. Member
**
Offline Offline

Posts: 96



WWW
« Reply #3 on: October 05, 2007, 10:56:25 PM »
ReplyReply

No photo of the obvious test subject:  fine cloth with a visible weave at an appropriate distance.  Another test is a screen window or a screen door.
Logged

MonitoPhoto (Landscape, Architecture, Portraits: Halifax, Nova Scotia)
feppe
Sr. Member
****
Offline Offline

Posts: 2909

Oh this shows up in here!


WWW
« Reply #4 on: October 06, 2007, 05:02:26 AM »
ReplyReply

Quote
No photo of the obvious test subject:  fine cloth with a visible weave at an appropriate distance.  Another test is a screen window or a screen door.
[a href=\"index.php?act=findpost&pid=144150\"][{POST_SNAPBACK}][/a]

Since removing an anti-aliasing filter only has a negative impact on those (and similar) subjects, would it not be prudent to remove it? Unless you're mainly shooting subjects subject to aliasing, one can fix the occasional aliasing artifacts with software. If the filter has such a negative impact on image sharpness, why not get rid of it? Right?
Logged

Quentin
Sr. Member
****
Offline Offline

Posts: 1123



WWW
« Reply #5 on: October 06, 2007, 07:20:28 AM »
ReplyReply

That's the reason why I used a Kodak 14nx for several years (no AA filter) and now use a Mamiya ZD (no AA filter).  AA filters turn detail to mush and make a mockery of supposedly higher pixel counts.  There are disadvantages with some subjects, but after working with non-AA filter cameras for 5years, I know how to deal with those occasions where colour aliasing causes problems.  I've done my research and the results are in.  No AA filter is better.

Quentin
« Last Edit: October 06, 2007, 07:21:50 AM by Quentin » Logged

Quentin Bargate, ARPS, Author, photographer entrepreneur and senior partner of Bargate Murray, Law Firm of the Year 2013
feppe
Sr. Member
****
Offline Offline

Posts: 2909

Oh this shows up in here!


WWW
« Reply #6 on: October 06, 2007, 08:08:30 AM »
ReplyReply

Quote
That's the reason why I used a Kodak 14nx for several years (no AA filter) and now use a Mamiya ZD (no AA filter).  AA filters turn detail to mush and make a mockery of supposedly higher pixel counts.  There are disadvantages with some subjects, but after working with non-AA filter cameras for 5years, I know how to deal with those occasions where colour aliasing causes problems.  I've done my research and the results are in.  No AA filter is better.

Quentin
[a href=\"index.php?act=findpost&pid=144200\"][{POST_SNAPBACK}][/a]

I did some reading online about them, and your experience seems to back that up. Moire is fixable with software, and you get to choose whether to use anti-aliasing, instead of it being forced upon you.

If only the maxmax site was navigable, as it's such a jumbled mess I can't find whether they remove AA filters from 30D/40D without any IR funkiness.
Logged

Morgan_Moore
Sr. Member
****
Offline Offline

Posts: 2221


WWW
« Reply #7 on: October 06, 2007, 03:20:50 PM »
ReplyReply

Quote
That's the reason why I used a Kodak 14nx for several years (no AA filter) and now use a Mamiya ZD (no AA filter).  AA filters turn detail to mush and make a mockery of supposedly higher pixel counts.  There are disadvantages with some subjects, but after working with non-AA filter cameras for 5years, I know how to deal with those occasions where colour aliasing causes problems.  I've done my research and the results are in.  No AA filter is better.

Quentin
[a href=\"index.php?act=findpost&pid=144200\"][{POST_SNAPBACK}][/a]

As a user of and SLRN, a 22mp DSLR and a D200 I think this clarifies why my D200 always looks soft

Whats the score with the D3

And yes the first two do suffer moiree on occasion

S
Logged

Sam Morgan Moore Cornwall
www.sammorganmoore.com -photography
marcmccalmont
Sr. Member
****
Offline Offline

Posts: 1729



« Reply #8 on: October 07, 2007, 01:48:57 AM »
ReplyReply

Quote
I did some reading online about them, and your experience seems to back that up. Moire is fixable with software, and you get to choose whether to use anti-aliasing, instead of it being forced upon you.

If only the maxmax site was navigable, as it's such a jumbled mess I can't find whether they remove AA filters from 30D/40D without any IR funkiness.
[{POST_SNAPBACK}][/a]


It is called a visible light Hot Rod

[a href=\"http://maxmax.com/hot_rod_visible.htm]http://maxmax.com/hot_rod_visible.htm[/url]

send them an email I'm sure he can cut a visible light filter to fit any camera
sales@maxmax.com

Marc

A couple of images taken with a 5D w/visible light modification (AA filter removed) sorry for the jpegs I wish you could see the 16 bit prints.
Logged

Marc McCalmont
marcmccalmont
Sr. Member
****
Offline Offline

Posts: 1729



« Reply #9 on: October 07, 2007, 01:51:09 AM »
ReplyReply

quote=feppe,Oct 6 2007, 03:08 AM]
I did some reading online about them, and your experience seems to back that up. Moire is fixable with software, and you get to choose whether to use anti-aliasing, instead of it being forced upon you.

If only the maxmax site was navigable, as it's such a jumbled mess I can't find whether they remove AA filters from 30D/40D without any IR funkiness.
[{POST_SNAPBACK}][/a]
[/quote]


It is called a visible light Hot Rod

[a href=\"http://maxmax.com/hot_rod_visible.htm]http://maxmax.com/hot_rod_visible.htm


send them an email I'm sure he can cut a visible light filter to fit any camera
sales@maxmax.com

Marc

A couple of images taken with a 5D w/visible light modification (AA filter removed) sorry for the jpegs I wish you could see the 16 bit prints.

[attachment=3494:attachment][attachment=3495:attachment][attachment=3496:attachm
ent]
Logged

Marc McCalmont
ErikKaffehr
Sr. Member
****
Offline Offline

Posts: 7888


WWW
« Reply #10 on: October 07, 2007, 03:23:29 AM »
ReplyReply

Hi,

I just wonder how much of the contrast/resolution lost in the AA filter that can be regained using capture sharpening? I have actually downloaded two of the examples on the "Maxmax site" taken with EOS 5D , processed in Lightroom and saw no big difference.

I enclose cropped JPEGs. Pictures were normalized (with auto-exposure in LR) sharpened
and cropped. Sharpening is on both pictures:

amount 61
radius 0.60
detail  50
masking 48


Best regards

Erik



[attachment=3498:attachment]
« Last Edit: October 07, 2007, 03:42:45 AM by ErikKaffehr » Logged

Quentin
Sr. Member
****
Offline Offline

Posts: 1123



WWW
« Reply #11 on: October 07, 2007, 03:57:45 AM »
ReplyReply

An ideal solution would be a user removable AA filter.  I don't know of any currrent 35mm format dslr that has one.  The ZD does have an optional user-replaceable AA filter , but its pretty expensive - but at least its available if you feel you need it for some shots, maybe portraits or wedding where avoiding moire might be more important than absolute resolution.  

Quentin
Logged

Quentin Bargate, ARPS, Author, photographer entrepreneur and senior partner of Bargate Murray, Law Firm of the Year 2013
marcmccalmont
Sr. Member
****
Offline Offline

Posts: 1729



« Reply #12 on: October 07, 2007, 05:34:33 PM »
ReplyReply

The bottom line is you can use more software sharpening to regain a sharp image with a AA filter inplace but images look more realistic/better with less software sharpening and no AA filter. I think this is one reason MFDB images look so good. The question remains is it worth $450? On my 5D I think it was worth the $ on my D80 it was not. On a 1Ds III yes I would do the mod.
Marc
Logged

Marc McCalmont
nma
Full Member
***
Offline Offline

Posts: 161


« Reply #13 on: October 07, 2007, 09:12:53 PM »
ReplyReply

Quote
The bottom line is you can use more software sharpening to regain a sharp image with a AA filter inplace but images look more realistic/better with less software sharpening and no AA filter. I think this is one reason MFDB images look so good. The question remains is it worth $450? On my 5D I think it was worth the $ on my D80 it was not. On a 1Ds III yes I would do the mod.
Marc
[a href=\"index.php?act=findpost&pid=144458\"][{POST_SNAPBACK}][/a]

This topic comes up from time to time, with writers always suggesting, implicitly or explicitly, that camera manufactures who include (or omit) an AA filter don't know what they are doing. Nonsense!  Particularly with cameras using a Bayer-type sensor, careful attention must be given to matching the digital sampling of the sensor and the resolution of the lens. It is instructive to consider two case -- 1. the 3MP sensor and 2. the 3000 MP sensor (at least conceptually   ).  In the first case, most would agree that aliasing would be a severe problem and it would likely affect all images, not just the ones with repetitive patterns.  In the second case, it is likely that the sensor would not exhibit any noticible aliasing because of the high spatial sampling. Put another way, the Nyquist frequency equivalent for the sensor would be higher than the Nyquist frequency for the lens.  So, it is this balance between sampling and lens resolution that must be properly struck. The AA filter is supposed to limit the high frequency information passed by the lens so that the image can be "reconstructed" up to the Nyquist frequency.  Fabricating a perfect AA filter is apparently a problem.  Too strong and you loose resolution you are entitled to and too weak leads to aliasing and Moire patterns. It has to be just right! Notice that with a Bayer sensor, the poorer sampling of red and blue can lead to problems.  High quality lenses can be a problem for lower MP DSLRs made without a proper AA filter.


Contrary to assertions made in this thread, there is no general algorithm capable of undoing Moire or aliasing.  Aliasing puts high frequency details in the wrong place in the image. Without detailed apriori knoweldge of the scene it can't be corrected, though perhaps it can be mitigated in some cases.

As a speculation, I suggest that cameras like the 1DsIII may be reaching the point where they would function well without an AA filter for all but the best lenses.
Logged
John Sheehy
Sr. Member
****
Offline Offline

Posts: 838


« Reply #14 on: October 07, 2007, 09:56:54 PM »
ReplyReply

Quote
The bottom line is you can use more software sharpening to regain a sharp image with a AA filter inplace but images look more realistic/better with less software sharpening and no AA filter.[a href=\"index.php?act=findpost&pid=144458\"][{POST_SNAPBACK}][/a]

More realistic?  The real world runs out of resolution, it doesn't alias.

There is nothing "realistic" about aliased imaging.
« Last Edit: October 07, 2007, 10:13:52 PM by John Sheehy » Logged
marcmccalmont
Sr. Member
****
Offline Offline

Posts: 1729



« Reply #15 on: October 07, 2007, 10:55:49 PM »
ReplyReply

Quote
More realistic?  The real world runs out of resolution, it doesn't alias.

There is nothing "realistic" about aliased imaging.
[a href=\"index.php?act=findpost&pid=144511\"][{POST_SNAPBACK}][/a]

I guess I was not specific enough, I should have written "to my eye, although images from my 5D with an AA filter inplace could be sharpened in software, once I had the AA filter removed the images looked more real, palpable and 3 dimensional, using much less software sharpening." Just my personal observation from having my 5D modified.
Marc
Logged

Marc McCalmont
marcmccalmont
Sr. Member
****
Offline Offline

Posts: 1729



« Reply #16 on: October 08, 2007, 02:09:39 AM »
ReplyReply

Quote
This topic comes up from time to time, with writers always suggesting, implicitly or explicitly, that camera manufactures who include (or omit) an AA filter don't know what they are doing. Nonsense!  Particularly with cameras using a Bayer-type sensor, careful attention must be given to matching the digital sampling of the sensor and the resolution of the lens. It is instructive to consider two case -- 1. the 3MP sensor and 2. the 3000 MP sensor (at least conceptually   ).  In the first case, most would agree that aliasing would be a severe problem and it would likely affect all images, not just the ones with repetitive patterns.  In the second case, it is likely that the sensor would not exhibit any noticible aliasing because of the high spatial sampling. Put another way, the Nyquist frequency equivalent for the sensor would be higher than the Nyquist frequency for the lens.  So, it is this balance between sampling and lens resolution that must be properly struck. The AA filter is supposed to limit the high frequency information passed by the lens so that the image can be "reconstructed" up to the Nyquist frequency.  Fabricating a perfect AA filter is apparently a problem.  Too strong and you loose resolution you are entitled to and too weak leads to aliasing and Moire patterns. It has to be just right! Notice that with a Bayer sensor, the poorer sampling of red and blue can lead to problems.  High quality lenses can be a problem for lower MP DSLRs made without a proper AA filter.
Contrary to assertions made in this thread, there is no general algorithm capable of undoing Moire or aliasing.  Aliasing puts high frequency details in the wrong place in the image. Without detailed apriori knoweldge of the scene it can't be corrected, though perhaps it can be mitigated in some cases.

As a speculation, I suggest that cameras like the 1DsIII may be reaching the point where they would function well without an AA filter for all but the best lenses.
[a href=\"index.php?act=findpost&pid=144504\"][{POST_SNAPBACK}][/a]

I think the large Japanese manufactures cater to a wide audience with many uses for their cameras, they also started making DSLR’s with low mega pixel sensors.  The MFDB manufacturers cater to a smaller market with higher IQ expectations and work with higher mega pixel sensors. From my experience with mostly nature/landscape images I rarely see moiré using my modified 5D. My conclusion is that for my purposes, the AA filter was not necessary and my image quality improved with it’s removal. I like Mamiya’s solution to make it an option.
Marc
Logged

Marc McCalmont
jing q
Sr. Member
****
Offline Offline

Posts: 594


WWW
« Reply #17 on: October 08, 2007, 04:03:33 PM »
ReplyReply

the samples posted on their website reflect my experience using medium format digital backs. There's a crispness (an over crispness almost...) that can be abit distracting sometimes but when printed big the 3-dimensional effect is immediately visible when compared to an image from a camera with an AA filter..

moire becomes abit of a problem but it's a small price to pay.
Will they mod a 1ds MkII? I would be interested
Logged
feppe
Sr. Member
****
Offline Offline

Posts: 2909

Oh this shows up in here!


WWW
« Reply #18 on: October 08, 2007, 05:11:16 PM »
ReplyReply

Quote
the samples posted on their website reflect my experience using medium format digital backs. There's a crispness (an over crispness almost...) that can be abit distracting sometimes but when printed big the 3-dimensional effect is immediately visible when compared to an image from a camera with an AA filter..

moire becomes abit of a problem but it's a small price to pay.
Will they mod a 1ds MkII? I would be interested
[a href=\"index.php?act=findpost&pid=144696\"][{POST_SNAPBACK}][/a]

I emailed them, and they said they currently only offer 5D conversions for Canons, as they don't have the correct-size filters to make the Hot Rod mod (ie. removing the anti-aliasing filter along with whatever else it filters, and replacing it with somethingsomething).
Logged

Anders_HK
Sr. Member
****
Offline Offline

Posts: 1001



WWW
« Reply #19 on: October 09, 2007, 02:17:52 AM »
ReplyReply

Quote
A couple of images taken with a 5D w/visible light modification (AA filter removed) sorry for the jpegs I wish you could see the 16 bit prints.

[attachment=3494:attachment][attachment=3495:attachment][attachment=3496:attachm
ent]
[a href=\"index.php?act=findpost&pid=144341\"][{POST_SNAPBACK}][/a]

Marc,

This is interesting. How do you experience the change of your 5D in the dark near black shadow regions and brightest highlights such as the sun on that one photo? Is there more detail similar to MF?

Much thanks.

Regards
Anders
Logged
Pages: [1] 2 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad