Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Moire problems from 5D  (Read 12389 times)
Ben Rubinstein
Sr. Member
****
Offline Offline

Posts: 1733


« on: May 07, 2006, 06:39:54 AM »
ReplyReply

This file is straight out of ACR, no sharpening, 5D with 24-70L and latest version ACR. The Moire shows in DPP though not so badly, Rawshooter is the best but the moire is still there. All images viewed at 100% of course, Help?

I need to sort out these images for a start, the client is waiting for them but I don't know how to fix moire in PS or if it is possible without too much hassle. I don't want to have to switch to Rawshooter, my entire workflow and camera profiles are set up for Bridge/ACR/PS. I also want to know how to avoid this in the future and it is scaring me.
« Last Edit: May 07, 2006, 02:26:00 PM by pom » Logged

Richowens
Sr. Member
****
Offline Offline

Posts: 847



« Reply #1 on: May 07, 2006, 09:09:23 PM »
ReplyReply

Hi Pom,

I used this technique to clear up the moire.

Color Gremlins by Christopher Smith

[attachment=547:attachment]

Think it will work for you?

Rich
Logged

drew
Sr. Member
****
Offline Offline

Posts: 477



WWW
« Reply #2 on: May 08, 2006, 03:44:23 AM »
ReplyReply

Interesting. A review of the 5D in this week's BJP indicates that you simply would not see this problem with the 1DS MKII.
Logged

Andrew Richards My Webpage
Ben Rubinstein
Sr. Member
****
Offline Offline

Posts: 1733


« Reply #3 on: May 08, 2006, 04:34:13 AM »
ReplyReply

Having posted on another forum I got this advice:

duplicate layer and set blend to colour. Add a layer mask (hide all) and just paint in the problem bit. Apply Gaussian blur till the problem goes away. Worked like a champ. There is still a colour cast but duplicating the layer mask to a color correction layer makes it fast to correct.

The important bit is to use layer masks otherwise the colours in the rest of the image go funky.
Logged

drew
Sr. Member
****
Offline Offline

Posts: 477



WWW
« Reply #4 on: May 08, 2006, 06:16:48 AM »
ReplyReply

Phase One Capture has a plug-in anti-moire filter. Never tried it, but that might work too.
Logged

Andrew Richards My Webpage
michael
Administrator
Sr. Member
*****
Online Online

Posts: 4925



« Reply #5 on: May 08, 2006, 07:08:19 AM »
ReplyReply

Some thoughts on the Moire issue.

It doesn't exist with film, because silver and dye particles (grain) are not regularly spaced. Pixels are.

Moire happens when the line frequency of a very fine pattern (usually in fabrics) creates an interference pattern with that of the sensor's pixels pitch.

For this reason almost all digital camera makers place an anti-aliasing filter (AA) in front of the sensor. This redeuces resolution somewhat but also reduces moire effects.

The problem starts to disappear at about 22 MP, though it can show up at any time if the pattern being photographed is fine enough. For this reason most medium format back makers do not use AA filters. This adds to their ability to record very high resolution images, and in the case of Phase One, their Capture One software comes with a Moire removal filter for those times when it does appear.

One can't really make generized statements about which cameras show moire more or less than others because it is totally dependant on the pattern in the fabric and the distance from the camera to the subject. At just the right pattern size and distance you'll always get moire with any digital sensor.

The only generalization that can be made is that the higher the pixel resolution of the sensor the less chance there is for moire, except with very small pattern spacing subjects.

Michael
Logged
William_Good
Newbie
*
Offline Offline

Posts: 14


WWW
« Reply #6 on: May 08, 2006, 08:46:10 AM »
ReplyReply

Its my understanding the 5D has a weaker  AA filter than the 1ds2.

I mainly use a kodak SLRc with no AA filter.
The cam is very sharp, but can produce image artifacts like moire.
Because of the lack of AA, Kodak users have had to learn how to deal with more image fixes than others....  
Slikypix software works very well on moire ( & NR ).  I believe they still have a 2 week trial download available.
There are MANY other methods to process moire - search the kodak forum at dpreview.
Logged

Website URL in profile
drew
Sr. Member
****
Offline Offline

Posts: 477



WWW
« Reply #7 on: May 08, 2006, 09:12:50 AM »
ReplyReply

Quote
One can't really make generized statements about which cameras show moire more or less than others because it is totally dependant on the pattern in the fabric and the distance from the camera to the subject. At just the right pattern size and distance you'll always get moire with any digital sensor.

Michael,
Sorry to disagree, but that is precisely what the British Journal of Photography article (written by Anders Uschold) does show. What William says is correct re the anti-aliasing filter. It is also true that the AA filter on the 1DS MKI is also weaker than on the MKII. My understanding of Moire is that it occurs when the frequency of a repeating pattern exceeds the frequency of the pixel rows on a Bayer sensor (all to do with Nyquist and his ?law). However, the BJP article also makes it clear that the signal processing that Canon employs in the 5D is significantly different from that in the 1DS MKII. The BJP article includes images of test charts specifically designed to show Moire. The 1DS MKII hardly shows Moire while the 5D most definitely does. Of course, the trade off is that the 5D images will probably appear sharper for those same fine details.
Logged

Andrew Richards My Webpage
michael
Administrator
Sr. Member
*****
Online Online

Posts: 4925



« Reply #8 on: May 08, 2006, 09:24:21 AM »
ReplyReply

Exactly what is it that you're disagreeing with?

Michael
Logged
drew
Sr. Member
****
Offline Offline

Posts: 477



WWW
« Reply #9 on: May 08, 2006, 10:39:23 AM »
ReplyReply

That you cannot make generalized statements.
« Last Edit: May 09, 2006, 04:09:56 AM by drew » Logged

Andrew Richards My Webpage
opgr
Sr. Member
****
Offline Offline

Posts: 1125


WWW
« Reply #10 on: May 08, 2006, 11:09:26 AM »
ReplyReply

Let's not confuse the two types of "frequency" discussed here:

Frequency in sampling theory denotes a simple periodic wave in a Fourier series to rebuild a complex, non-periodic wave in the spatial domain. If you convert a (spatial) scene to the frequency domain, ANY object has an infinitely high frequency component for its edge.

In a hypothetical ideal camera system, any single edge or line will exhibit aliasing artifacts. In practice all camera systems are limited regarding sharpness thereby blurring edges (by diffraction for example). At high enough resolution, the lens will act as an automatic anti-aliasing filter, which is why high resolution systems are less likely to exhibit problems.

Given the above, if a system creates aliasing problems in an edge, then a repeat of edges will result in a fairly large area of inconsistent colors usually visible as irregular patterns known as moire. This repetition of edges is also referred to as frequency, but can be very much below the sampling resolution of the CCD and still result in the moire artifacts...!

Following are very small crops from 5d images from the dpreview site. (I did not ask permission to copy these so I would like you to see them as "quotes" which I presume is allowed under normal copyright laws.)

Images 1 shows aliasing artifacts, image 2 and 3 show how a single image-line (larger than a single scan-line) can already produce discoloration. Given enough single lines relatively close together (but obviously well below sample resolution), one may see moire...





Logged

Regards,
Oscar Rysdyk
theimagingfactory
drew
Sr. Member
****
Offline Offline

Posts: 477



WWW
« Reply #11 on: May 08, 2006, 11:32:54 AM »
ReplyReply

Quote
Given the above, if a system creates aliasing problems in an edge, then a repeat of edges will result in a fairly large area of inconsistent colors usually visible as irregular patterns known as moire.

Well, it would be all to easy to confuse a simple soul like me......, but my point at the beginning was that as a generalized statement, these Moire artifacts are much more likely to be evident with the 5D, than with the 1DS MKII. According to the BJP article, this is down to a combination of the AA filter and signal processing. This is well demonstrated by the shots of test objects which are designed specifically to induce these artifacts. Perhaps I should not have implied that you will never see this problem with the 1DS MKII, but I really think this is splitting hairs.
Logged

Andrew Richards My Webpage
61Dynamic
Sr. Member
****
Offline Offline

Posts: 1442


WWW
« Reply #12 on: May 08, 2006, 12:33:05 PM »
ReplyReply

Quote
Michael,
Sorry to disagree, but that is precisely what the British Journal of Photography article (written by Anders Uschold) does show. What William says is correct re the anti-aliasing filter. It is also true that the AA filter on the 1DS MKI is also weaker than on the MKII. My understanding of Moire is that it occurs when the frequency of a repeating pattern exceeds the frequency of the pixel rows on a Bayer sensor (all to do with Nyquist and his ?law). However, the BJP article also makes it clear that the signal processing that Canon employs in the 5D is significantly different from that in the 1DS MKII. The BJP article includes images of test charts specifically designed to show Moire. The 1DS MKII hardly shows Moire while the 5D most definitely does. Of course, the trade off is that the 5D images will probably appear sharper for those same fine details.
[a href=\"index.php?act=findpost&pid=64779\"][{POST_SNAPBACK}][/a]
Quote
That you can make generalized statements.
[a href=\"index.php?act=findpost&pid=64788\"][{POST_SNAPBACK}][/a]
Quote
According to the BJP article, this is down to a combination of the AA filter and signal processing.
Your first statement I quoted does nothing to contradict anything Michael said. The Anders test is comparing a 1Ds Mk II camera which has more resolution than the 5D and thus falls in line with Michael's statements on higher resolution cameras not having as great of an issue with moire. Sure, the processing is different in the 5D, but that is a moot point since it neither confirms nor denies the generalization regarding resolution.

Anders provides a test that basically confirms the generalization you are not fond of. The test would have to be conducted between two different cameras of the same pixel resolution but different processing and AA filters (such as the 30D and the 1D Mk II) for your final conclusion to be valid.
Logged
Ray
Sr. Member
****
Offline Offline

Posts: 8948


« Reply #13 on: May 08, 2006, 08:54:06 PM »
ReplyReply

Would I be right in assuming that this moire problem is only likely to be an issue at f stops smaller than f11, and then only when the fine pattern on the subject is in precise focus? If this is true, then one could avoid the problem entirely at the shooting stage.
Logged
drew
Sr. Member
****
Offline Offline

Posts: 477



WWW
« Reply #14 on: May 09, 2006, 03:18:08 AM »
ReplyReply

Quote
Anders provides a test that basically confirms the generalization you are not fond of

Sorry, I may have mislead you, but I am 'fond' of that generalization. The BJP test objects are designed to bring out Moire artifacts regardless of the resolution of the device used to capture them. The 1DS MKII does not show troublesome artifacts to anything like the same extent as the 5D. The generalization I am 'fond' of is that this will hold true in the real world just as someone else has observed that the Kodak DCS Pro 14n is very prone to this kind of problem.

Now why is it that I am starting to feel like a dissected frog?......
Logged

Andrew Richards My Webpage
drew
Sr. Member
****
Offline Offline

Posts: 477



WWW
« Reply #15 on: May 09, 2006, 04:10:58 AM »
ReplyReply

I have edited my post above to show 'cannot' instead of 'can'. Hope this is clear.
Logged

Andrew Richards My Webpage
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad