How to remove moire?

<p><img src="C:\Photos\Architecture\LOTT%20Clean%20Water%20Alliance\LOTT-5.jpg" alt="" />Hello- I experienced moire for the first time with my Nikon D800. See attached. This shot was taken with a Nikon 24mm tilt-shift stopped down to f8 or so. I have tried multiple techniques to remove it in Photoshop and Capture One with no success. I'm not crazy about the blur technique that some suggest since I don't want to lose detail where moire occurs, if possible. Any suggestions or tips?<br>

Thanks!<br>

Andrew</p>

 

<p>Sorry guys....here's the attachment: <img src="C:\Photos\Architecture\LOTT%20Clean%20Water%20Alliance\LOTT-6.jpg" alt="" /></p><div></div>

<p>You don't have any moire on this photograph. Check the image on your monitor larger os smaller view, it my be your computer video car and the monitor communication.<br /> Checked on a MAC to EIZO calibrated monitor.<br /> It is a nice high quality architectural image.</p>

<p>Andrew, I just wanted to confirm that I see the moire pattern as well, especially in the sunlit part of the right side of the building, but in other places too. I brought the photo into GIMP and saw the pattern, faintly but definitely, throughout a wide zoom range. I suspect the vertical lines on the building panels were setting up a pattern with the pixels in the camera. It doesn't look like a monitor issue to me. Unfortunately, I have no suggestions on how to get rid of it.</p>

<p>Andrew, the moire is definitely there, as Mark said, but the pattern changes if you increase or decrease the size of the image on the screen. This tends to make me think the moire results from interaction between the vertical lines on the wall and the monitor scan rate.</p>

<p>One way to find out for sure: Make a print.</p>

<p>Sorry but I don't see any moire on my monitor either. I'll recommend that you make a print first to see if it's really there.</p>

<p>I'm seeing variations in luminance between the vertical structure of the wall panels, but I'm not sure that is moire. Moreover, the D800 has both very high pixel density <em>and</em> an anti-aliasing filter, so I would expect it to occur very rarely.<br>

It would be nice to see a 100% crop of a a part of the image in which you see the pattern.</p>

I can only barely see the moiré pattern in the sunlit part of the building. My guess is that the effect is actually related to the interaction of the vertical lines in the image with the dot pitch of people's individual monitors, not a moiré pattern in the image itself. I would have thought the Nikon D800's optical low-pass filter would avoid moiré patterns. I echo the suggestion to make a print and proceed from there.

<p>It has NO moire on the image, It is your MONITOR.<br>

Still not sure, print the image and you will see it.</p>

<p>I believe it is a monitor issue, and like others have suggested, make a print to confirm.<br>

<br />I myself am more concerned with the what appears to be poor use of the Tilt-Shift lens.<br />The Verticals on both of the upright elements seem to be over compensated for and in fact are leaning towards the viewer rather than perfectly upright.<br>

<br />Just my ¢2 worth.</p>

<p>I don't see any Moire - retina screen MBP. Would suggest the OP view the image in PS at either 25, 50 or 100% and see if it is still there. Let us know. Was also wondering about the curvature I see in the building. Is that really there?</p>

<p>That is so cool!</p>

<p>On this Windows monitor, for a fraction of a second as the image is loading, I see a really bad, oval moire on the right-most illuminated panel of the building. Once the image is fully loaded, that moire goes away. When I look at that same panel at 100%, I see a weak moire on the lower half of that same panel. It's a roughly vertical series of lines which curve to a more upper-right to lower-left orientation toward the bottom. With all the various observations, I thought another monitor was called for.</p>

<p>On my iPad ("Retina display"), I see nothing at "fit-to-screen" size. When I enlarge it "a lot" I don't see it. If I go from "fit" to "large" in small steps, there is a small range where I see roughly the same pattern as I saw on this monitor. It seems like it is indeed at the size where the original pixels are the same size (or at least about the same size) as the monitor pixels. </p>

<p>That seems to support the "it's the monitor" argument. It never occurred to me that this could happen without being an inherent part of the image -- something that would show up at any degree of enlargement over 100%. </p>

 

<p>DxO Optics Pro does an excellent job of removing moire, when it occasionally occurs with my Canon 5DsR. You can try it for 30-days free.</p>

<p>Hi Andrew, a couple things are needed to help you with the Moiré<br>

1) You and we need to be looking at a 100% crop of your original image in either an uncompressed format or max quality JPEG. I am assuming you downsampled your image and not only will this often remove the Moiré, it can also set up some false Moiré patterns as well. The forum members could very easily be chasing a red herring</p>

<p> 2) You need to view the image at 100%. Observing at other magnifications can introduce resizing artifacts that are not actually in the image data. Your Moiré patterns certainly could be there, yet there is no way to tell about the content of your original image without these two conditions</p>

<p>If with the above conditions met, if what you see is a Moiré pattern with color striations, then there are a couple techniques that are worth trying</p>

<p>a) Use a variety of different raw converters. Some are much better than others in avoiding Moiré in the first place through the demosaicing process.</p>

<p>b) Second to "a", if you have not already, try this technique in the link and remember you must do the adjustments at 100% magnification or multiples higher and using the original Raw file in 16 bit is best:

The blog on the same topic gives several examples of removing Moiré with that technique: jkwphoto.blogspot.com It does not work in every case of color striations yet has worked about 90% of the time for me. Keep in mind that once you have repeating structures that go below the resolution of the pixels in the sensor, there is not way to recover the detail and you will see artifacts of some sort.</p>

<p>If the Moiré is there yet it is not color striation Moiré, the the described technique is not the one to use.</p>

<p>Hope this information is helpful.</p>

<p>Thanks everyone for your thoughtful responses.<br>

So, as some have experienced, I see dramatic changes in the degree to which the moire pattern appears depending upon the magnification I'm using. At 33%, it's horrendous. You see big circular waves on the right side. At 100%, it mostly goes away, but the quality is still poor (some moire remains). I've attached a crop at 100%. The fact that it almost goes away at 100% makes correcting it difficult. I think I'll give another RAW converter a try. <br>

Regarding the bowing lines in the center, which I believe someone referenced, I didn't attempt to correct that yet. However, I was wondering if anyone knew if there was a lens profile out there for the Nikon 24mm tilt-shift lens. My ACR is up to date since I'm on CC.<br>

<img src="C:\Photos\Architecture\LOTT%20Clean%20Water%20Alliance\LOTT-33perc" alt="" /></p>

<p> </p><div></div>

<p>I think that most profilers tend not to include tilt-shift lenses because they have so many variable. I does seem like they'd provide a base line, with no tilt or shift.</p>

<p>Andrew, in the 100% crop at the left hand side wall panel which is in the shade - there might be some very minor diagonal, sparse, almost vertical banding. That may be moire or just a structural thing. I would not worry either one. The building looks great.<br>

(If you have a panorama tool like PTGui, you are easilly able to create corrective lens profile (a,b,c values) for a known shift value to apply later. Only tested to a non shift lens - though.)</p>

<p>Hi Andrew<br>

Trying another Raw Converter may do a better job so that is a great first step. <br>

That the Moiré almost goes away at 100% viewing is an indication that you are seeing a false Moiré at lower magnifications causes by the display resizing algorithm and for the most part does not exist in the image at all. </p>

<p>Not sure you need to make any changes as the Moiré at first glance would not be seen by the vast majority. One can make it stand out by looking at the Luminosity channel and putting in a very high contrast curve to amplify this subtlety.</p>

<p>However, trying to fix the image using your compressed JPEG would not have a chance with the approach I mentioned before (if it would work at all in the first place) because the color components are highly compressed. Reverse engineering your 100% crop has a signature that you had a compression setting of 69 in either Photoshop Save for Web or on exporting from LR. Not enough color information left to attempt a recovery using the mentioned technique.</p>

<p>If you want to post a download link to a 100% quality JPEG compression or better yet the Raw file, I could give it my best shot at correction. </p>

<p> </p>

<p>Just downloaded the OP's original attached image and applied various settings to Vertical, Horizontal, Rotate & Distortion sliders in CS5 ACR's Transform panel and saw the banding Kari mentions change and almost disappear viewing at 100% zoom, so it appears it's a fixable artifact.</p>

<p>That image really does need Transform like adjustments to make those vertical walls less bowed.</p>

<p>Try converting in DxO Optics Pro, with the camera and lens modules loaded and those bowed lines will straighten up. It's likely that the moire will also disappear. </p>

<p>I don't see it on your cropped example at all Andrew. I'm on a MBP retina. I see no artifacts or moire.</p>

<p>Upon closer inspection and after serious deliberation I have come to the conclusion that the dress is black and blue. Oh, I <strong>don't</strong> see a moire pattern as well.</p>

<p>Are you thinking that the vertical lines at constant spacing is moire? These are the overlapped edges of the corrugated metal panels, spaced about a metre apart. I can see nothing else though.</p>