Luminous Landscape Forum
Raw & Post Processing, Printing => Digital Image Processing => Topic started by: Slobodan Blagojevic on April 01, 2010, 10:22:07 pm
-
It is one of those things that you think happens only to other people (or cameras)... until it happens to you. You hear about moiré problems with MFDB and/or Leica M8s and M9s, given that they do not have anti-aliasing filters, and you think you are safe with your "blurry" camera (i.e., the one containing such a filter). Until one day you encounter this (check the black louvres on the blue building) :
[attachment=21219:Screen_s...42.10_PM.png]
Well, maybe I shouldn't, but I am surprised. Any explanation for this (other than "nothing is perfect"), and especially how to get read of it? The out-of-camera file was processed in LR 3 Beta 2.0 with everything on default (and I checked Canon's own DPP, the effect is still there, though slightly less pronounced). Trying to desaturate it (using adjustment brush) doesn't help much as it still leaves darker/lighter diagonal pattern.
On a side note, I never cease to be surprised just how sharp Canon's 17-55/2.8 is
-
Nothing is perfect
You might have some luck using Hue/Saturation, working on each offending part of the pattern to bring the colors and luminosities closer together. It won't disappear completely, but you should be able to hide it quite well.
I did this with Hue/Sat utilizing all three sliders for both magenta/red and green/cyan. The trick is to use the bottom slider (forget it's exact name...darnit) to select the specific areas of color you want to change.
(http://www.chuckkimmerle.com/test/moire2.jpg)
(http://www.chuckkimmerle.com/test/HS1.jpg)
(http://www.chuckkimmerle.com/test/HS2.jpg)
-
Thanks Chuck... it seems more labor-intensive and trial-and-error type of process than I hoped for. Sigh!
I remember seeing quite some time ago, in ads for a new raw developer (perhaps it was DxO?), they used a similar example to show how theirs is superior in removing moiré.
Anyone having any experience in that respect with other raw developers (i.e., other than LR and DPP)?
-
Thanks Chuck... it seems more labor-intensive and trial-and-error type of process than I hoped for. Sigh!
I remember seeing quite some time ago, in ads for a new raw developer (perhaps it was DxO?), they used a similar example to show how theirs is superior in removing moiré.
Anyone having any experience in that respect with other raw developers (i.e., other than LR and DPP)?
Hi Slobodan,
I haven't tried and compared all possible offerings, but this is my workflow recipe:
1. Avoid Moiré as much as possible (AA-filter). Prevention is better than cure. Try to predict the chance on moiré and shoot 2 exposures, one with slight defocus or diffraction, and one optimally focused. Blend locally after alignment/resizing. Unfortunately moiré usually appears after the fact or on dynamic scenes, so
2. If that is not good enough, use a built-in moiré suppression tool (the C1 Pro one I use is not perfect, hence recipe no.1, but it helps).
3. If no moiré tool available try a lot of noise reduction in one of 2 raw conversions and locally blend later.
4. If everything else fails, prepare to spend a lot of time on the local processing (e.g. inverted high-pass layer with appropriate blending).
Cheers,
Bart
-
It is one of those things that you think happens only to other people (or cameras)... until it happens to you. You hear about moiré problems with MFDB and/or Leica M8s and M9s, given that they do not have anti-aliasing filters, and you think you are safe with your "blurry" camera (i.e., the one containing such a filter). Until one day you encounter this (check the black louvres on the blue building) :
[attachment=21219:Screen_s...42.10_PM.png]
Well, maybe I shouldn't, but I am surprised. Any explanation for this (other than "nothing is perfect"), and especially how to get read of it? The out-of-camera file was processed in LR 3 Beta 2.0 with everything on default (and I checked Canon's own DPP, the effect is still there, though slightly less pronounced). Trying to desaturate it (using adjustment brush) doesn't help much as it still leaves darker/lighter diagonal pattern.
On a side note, I never cease to be surprised just how sharp Canon's 17-55/2.8 is
Moire can sometimes be an issue of raw conversion. Is the raw file available?
-
Moire can sometimes be an issue of raw conversion. Is the raw file available?
The file can be downloaded here (http://files.me.com/sblagojevic/alqsst) (12.8 MB)
-
The file can be downloaded here (http://files.me.com/sblagojevic/alqsst) (12.8 MB)
In LR3b2, the color noise reduction really whacked it pretty good - as did using the brush tool with -ve clarity, -ve sharpening and/or -ve saturation.
-
The file can be downloaded here (http://files.me.com/sblagojevic/alqsst) (12.8 MB)
There is moire in individual RGGB channels of the Bayer raw data, so it's not the converter.
-
File attached processed in Capture 1 with Moire amount at 50 (of 100). I have not previously used moire correction and I have a question. The processed image, when opened in CS4, shows no moire at 100% and larger but does show moire at less than 100%. Why?
-
The processed image, when opened in CS4, shows no moire at 100% and larger but does show moire at less than 100%. Why?
Hi Jack,
The re-appearance of moiré in the downsampled version is caused by the sub-standard resampling algorithm used. It's usually a choice between speed and avoiding artifacts, and software developers making wrong choices or not offering options. The downsampling algorithms used in Lightroom 3 and ACR 6 are probably better, we'll have to see what materializes in the final commercial releases.
Cheers,
Bart
-
Thank you, Bart, it is much as I suspected. Having virtually no experience with moire, I appreciate your tutorial!
-
File attached processed in Capture 1 with Moire amount at 50 (of 100)...
Thanks, Jack... good to know there is a simpler, one-slider solution (than trial-and-error, time-consuming ones). The bad news is the price of C1 and the fact that I am so used to Lightroom
On a side note, isn't it high time for LR to have a similar capability? It took LR quite some time to catch up with Canon's DPP in terms of noise processing (and even that is still in Beta).
-
Hi Slobodan, moire (and tools to deal with it) is something that we are hearing about more frequently now. Adobe is aware of the request. Please note this is in quite a different league (in terms of scope) compared to noise, in terms of % of images affected. This doesn't make it unimportant, but realistically it does affect the manner and order in which we tackle problems.
-
Thanks, Jack... good to know there is a simpler, one-slider solution (than trial-and-error, time-consuming ones). The bad news is the price of C1 and the fact that I am so used to Lightroom
On a side note, isn't it high time for LR to have a similar capability? It took LR quite some time to catch up with Canon's DPP in terms of noise processing (and even that is still in Beta).
You are welcome, Slobodan. Actually there is also a "Pattern" slider but, on your image, I saw no effect from its use.
-
File attached processed in Capture 1 with Moire amount at 50 (of 100).
That is very impressive. I wonder what their method is, it is typically very difficult to separate moire from natural image variation (for instance, why doesn't it also remove the slats that are the cause of the moire -- they are a regular banding pattern too!)... are there any unwanted side effects when this slider is cranked up?
-
File attached processed in Capture 1 with Moire amount at 50 (of 100).
That is very impressive. I wonder what their method is, it is typically very difficult to separate moire from natural image variation (for instance, why doesn't it also remove the slats that are the cause of the moire -- they are a regular banding pattern too!)... are there any unwanted side effects when this slider is cranked up?