Moire and UV Fluorescence with 5D Sensor

<p>I am not sure where this question goes, but am putting it here for two reasons. First, it has to do with the Canon 5D sensor and second, I am hoping Bob Atkins will have an answer for me. :^)</p>

<p>I recently asked a question in the Digital Darkroom forum about fixing really bad moire patterns on many photos I took at a client's bar mitzvah. I got great help, and am in the process of painstakingly fixing the moire stripes. However, this is the second time the issue has raised it's ugly head, and I am not going to go through this again. I need to find the cause and know how to avoid the problem in the future. Here is the thread.</p>

<p><a href="../digital-darkroom-forum/00W8gC">http://www.photo.net/digital-darkroom-forum/00W8gC</a></p>

<p>In trying to get to the root of the problem, I tried to find a sample bit of fabric that would produce moire with my gear. I apparently don't own anything that will do so. While looking for a sample, a friend said that it would have to be a little shiny. Her statement was drawn from the fact that the two instances were with young men's suits. Technically, this is not true--if going only on the moire theory, the pattern itself and not the sheen, would produce the moire. But thinking about this further, I remembered that some brighteners used by fabric manufacturers can fluoresce, making the material appear to be blue. Taking it one step further, since the fabrics in question are suiting material, it would stand to reason that sizing, and perhaps brighteners were used in the manufacture, and since most suits are not washable, the sizing and brighteners remain in the fabric.</p>

<p>I wondered if perhaps the moire patterns are made much worse due to fluorescing. The texture of the material itself must first be prone to produce moire with my 5D sensor, but are these patterns exacerbated by the fluorescence?</p>

<p>Looking at my gear, most of the worst moire was when I was using my Sunpak 120J. It does not have a UV coated tube. Some of the images that showed moire were shot using only a bounced 580EXII, however. I have a UV filter on my lens--a Hoya Pro 1 Digital UV filter.</p>

<p>I also notice, in my research to find answers about moire and the 5D sensor, that there aren't a lot of reported incidents of moire as bad as I got, which I found curious. I don't think I'm the only photographer using a 5D shooting bar mitzvahs a lot.</p>

<p>I guess what I'm asking is--can UV fluorescing be a factor (outside the texture causing the moire) in making moire appear worse?</p>

<p>Try <em>Photoshop LAB Color</em> by Dan Margulis. The book offers excellent solutions to deal with moire problems. Pgs 234 - 242<br>

g</p>

<p>Thanks, Gary, I'll look at it, but my question isn't about dealing with just moire, but whether UV fluorescence is a factor. I've researched and had excellent suggestions for dealing with moire in the image already. Are you saying that what is covered in the book goes beyond that?</p>

<p>Nadine,</p>

 

<p>The only sure way to figure it out would be to find something that exhibits the problem and

photograph it under lights with differing amounts of UV. Sunlight and incandescents should

be sufficient to get a quick idea, and using a fluorescent blacklight would be a clincher.</p>

 

<p>But my gut says that this isn’t likely to be a problem at all. Whiteners work by

absorbing ultraviolet light and reemitting it as visible light. The camera will generally see

exactly what you see with regards to fluorescence.</p>

 

<p>More likely to be a problem would be the other end of the spectrum — infrared.

Cameras are generally more sensitive to IR than humans. If you’ve ever tried to

photograph a red flower and had the red channel blow out without realizing it, you have first-

hand experience. (Red flowers tend to be very bright in the infrared part of the spectrum we

can’t see but cameras can.)</p>

 

<p>And, more likely still, I should think, would simply be whether or not the pattern of the

fabric happens to be at a very close scale to the pattern of the sensor. If you can compare a

5D and a 5DII, you could probably find a way to frame the scene such that the 5D has

moiré and the 5DII doesn’t. Then, magnify the scene (by zooming, either with

the lens or your feet) and the problem should switch to the 5DII and no longer be present on

the 5D.</p>

 

<p>Do please let us know what you find?</p>

 

<p>Cheers,</p>

 

<p>b&</p>

<p>Well, they may be made more visible by fluorescence I guess, but I doubt it.</p>

<p>Moire patterns typically result from detail in the subject which is at a higher spatial frequency than the sensor is capable of resolving but only just higher so they can get through the anti-aliasing filter.The detail in question here will be detail in the fabric. I suppose fluorescence could increase contrast in that detail, but again I sort of doubt it.</p>

<p>Such patterns are more likely to show up with a really good, high contrast, high resolution lens than with a cheaper "consumer" grade lens. They're also usually worse with cameras that have no or weak anti-aliasing filters (but this doesn't apply to any EOS cameras).</p>

<p>You can mess with various RAW converters to see if any show less moire than others but the "moire data" is present in the RAW file so there's only so much that you can do post-exposure.</p>

<p>When taking the images you can minimize moire by stopping the lens down to the point where the resolution is degraded by diffraction. That's like adding a stronger anti-aliasing filter since it limits high frequency information hitting the sensor. You could also use a diffusion filter or slightly (very slightly!) shift focus. Getting closer to the subject (so the detail is actually resolved) or getting further away to the detail shifts to higher frequencies and gets blocked by the anti-aliasing filter would also work.</p>

<p>Nope, it offers solutions to aspects of moire patterns, in particular the suit texture issue.<br>

If you are familiar with the LAB color concept it offs unique opportunities to control color independent of contrast, both these issues are pivotal in rendering delicate detail particularly sharped raw or jpgs. Regrettably, it makes no reference to the source of the moire origins which appears to be you major concern. Anyway, it's an excellent way to deal with the problem. Sorry I couldn't be more help. </p>

 

<p>Nadine, I have a thought.</p>

 

<p>See if you can find out where the suit was rented from, and then see if you can rent it (and a

mannequin, if you don’t have a spare teenager handy to model for you).</p>

 

<p>And then spend an afternoon shooting it, under all sorts of lighting conditions, to try to both

reproduce the moiré and eliminate it. Be sure to test with different lenses and different bodies.

For bonus points, try two different suits….</p>

 

<p>Bob’s list makes me think of another option: rather than stopping down to the point of

diffraction (probably at least f/16 to do enough on the 5D, but that’s what the rented suit would

be to figure out), open up all the way and go for razor-thin DoF, with the problem fabric out of the plane of focus.

It’d be harder technically, but perhaps preferable aesthetically for certain scenes.</p>

 

<p>Cheers,</p>

 

<p>b&</p>

<blockquote>

<p>I also notice, in my research to find answers about moire and the 5D sensor, that there aren't a lot of reported incidents of moire as bad as I got, which I found curious.</p>

</blockquote>

<p>Well, in this case perhaps Canon would be interested. If you are CPS talk to your tech, or call 1-800-OKCANON and ask them to put you in touch with the proper technical people.</p>

<p>If you have access to CaptureOne (the newer the better) see if it does a better job of getting rid of it.</p>

<p>Gary--thank you for clarifying. I am still going to try and find the book and read the pertinent parts. The thing about the kind of moire pattern I got was the actual 'stripes' themselves. While the color moire is easy to get rid of, the stripes persist, even in all the channels. So if the LAB method can address the stripes, I'm all ears. It is lightening the stripes by hand in the green channel or B&W layer, that is most time consuming. Looking at the stripes close up, though, I notice the actual fabric pattern is broken up, so I don't have a lot of hope of finding a way to easily deal with the stripes.</p>

<p>Bob--thanks for the thoughts about this. I didn't know if the idea of UV adding to the problem was even possible. Sounds like it probably isn't a factor. My thinking was that if fabric brighteners can fluoresce (sizing is also made of the same 'plastic' stuff), causing a blue shift, it might cause the moire stripes or pattern to go more bluish, contributing to the resulting contrast or darkness of the pattern, making it both more prominent and more difficult to get rid of. If my images had only color moire, I wouldn't be so concerned since color moire is relatively easy to get rid of. Also, I forgot to mention above, the shots I took of the boy with natural light (shaded) had no moire, although there were some shots taken with flash where I took a couple of the basically the same set up--one showed moire, the other, not, or very little.</p>

<p>Ben--also thanks for chiming in. I fully intend to get a sample fabric piece that will produce moire with the 5D. Initially, I was going to test the 5D and my 40D, as I could just use the latter. However, it might also have it's own problems. I doubt the boy's suit was rented. So I'll have to scrounge around at the remnant section of the fabric store, maybe.</p>

<p>When I do find a piece, I will test it in all kinds of light, with and without UV filters on the flash tube, different distances, cameras, etc. I understand the concept of stopping down for the diffraction and opening up for shallow DOF, but I'd rather find a solution where I wouldn't have to sacrifice sharpness or gamble with razor thin DOF. The latter is really not an option for group shots, though.</p>

<p>Just to be thorough--the lens was a Tamron 28-75mm f2.8 at about f8 (because I was also going to do group shots) and f5.6.</p>

<p>Nadine,</p>

 

<p>I’ve got a 5D and a 28-75. Offhand, I can’t think of any fabric I have that would be likely

to induce moiré, but I’ll wander around the house with my eyes open this weekend and

report my findings.</p>

 

<p>Any chance you remember what subject distance and focal length were you at?</p>

 

<p>Cheers,</p>

 

<p>b&</p>

<p>Regarding UV, I have one point to emphasize, and one to add. The emphasis is on what someone else already said about brighteners: they absorb UV and emit (at least mostly) visible light, because their purpose is to make things like clothes look brighter to the human eye. Since by definition UV is invisible to humans, a brightener which emitted mostly or entirely UV would be useless as a brightener, and would therefore likely never make it to market. Therefore, I believe that "brighteners may be causing this" and "this is related to UV reaching the sensor" are not consistent with each other.</p>

 

<p>The point to add is regarding the practical effects of UV, with and without filters. <a href="http://www.bobatkins.com/photography/technical/filters.html" target="_blank">Bob wrote an article about UV filters</a> and makes a couple of points in it that are relevant to this discussion. One is that most UV is blocked by the lens itself:</p>

 

<blockquote>I've looked at the range between 350nm and 400nm for UV blocking since the glass used in almost all lenses will itself block any light with a wavelength shorter than 350nm</blockquote>

 

<p>The other is that many UV filters are pretty much useless at blocking the wavelengths of UV that aren't already blocked by the lens. Whether that applies to your filter, I don't know; he tested a filter of the same brand name you use, and it was in fact one of the few which actually did what they claimed, but it may or may not be the same model you use.</p>

 

<p>The bottom line, for me, is that I think that when you do your test, you will find that UV is not the issue. But your idea of doing a comprehensive test is a good one; it's the only way you'll know for sure.</p>

<p>Michael--sadly I am not CPS. I toyed with getting Capture 1, but upon reading the Capture 1 info, the plug in they have is for color moire. Not sure I want to spend the $$$ only to find it kills color moire but not the stripes. Do you know anything about it? I tried processing my RAW file with DPP and Camera RAW, playing around with sharpening or, actually, minus sharpening, but the results are not very much different from my Lightroom results.</p>

<p>Ben--I will check now, what the focal lengths were. I can guess at the subject distances. If you look at the thread I linked to, you'll see the patterns involved. I'd say the fabric made up of small checks is about the size that will interfere with the 5D sensor. You can see the boy's arm for scale. These are 13 year old boys, to give you an idea.</p>

<p>In my search for suitable 'bad' fabric, I did find a fabric (part of one of my suits) that is smooth textured, but did produce a tiny bit of moire--not heavily striped. There is only the weave of the fabric for texture, which is probably a bit too small for the sensor to go crazy with. This is why I began to wonder about the sizing or brighteners and UV. All of my soft clothes with textures or patterns did not produce moire. The just mentioned fabric is part of a suit, hence probably has the sizing, etc. in it.</p>

<p>I also read that beyond about 20 megapixels, moire shouldn't be a problem--why, I don't remember, but I guess that's one more reason to get a 5DII. :^)</p>

<p>Nadine,</p>

 

<p>More megapickles should help, but I don’t think moiré will be a complete non-issue

until the sensor is diffraction-limited at f/1.4 (or whatever your fastest lens is). According to <a href="http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm">one online

calculator</a>, that’s about 250 megapickles for the 135 format — quite some time in the

future, I should think. But, when we finally get there, there won’t be any more antialising filters in

front of the sensor any more because there won’t be any point in them.</p>

 

<p>(And, yes, a 100% pixel view will be noisy as all get-out, but the noise will be so fine-grained in

prints as to be truly invisible. Higher ISOs will mean fuzzier pictures, not blotchy, noisy pictures.)</p>

 

<p>Cheers,</p>

 

<p>b&</p>

<p>Ben--I've checked my focal lengths and subject distances (guessing). Most of the images are from about 45-70mm, with subject distances in the 7-8 feet range, more or less. I also looked at frames where I used 28mm-35mm, and they exhibit moire too. I don't think I ever got less than 5-6 feet from my subjects. I tend not to shoot on top of people. I also checked the natural light (only) ones again, and did find an instance of moire, although it looked to be mostly color moire, without the deep patterning.</p>

<p>Steve--thanks for your input. I read Bob's article too. As you say--best way to find out is to test. Just gotta find a suitable bad fabric...</p>

<p>Below is one of my worst cases of striping. Gonna spend a lot of time with it. As I said, the stripes are in all channels. I did NOT sharpen for the web, so it appears slightly soft.</p><div></div>

<p>Looks like classic moire patterns. Your lens is just too sharp. If you'd used an old plastic consumer zoom, it probably wouldn't have shown up!</p>

<p>Thanks a lot, Bob! :^) I will be testing a lot soon and will update you all.</p>

<p>Hi Nadine,</p>

<p>I don't understand why you're thinking UV fluorescence might be the culprit. Are you thinking that something about the light would be passing your AI filter without getting "fuzzed?" I think any wavelength will get blurred by the AI filter. Of course there's also an IR filter over the sensor, but that shouldn't have anything to do with the moire pattern. Anyway, I'll offer the following thoughts about brighteners:</p>

<p>(1) Their purpose is to brighten white fabrics that otherwise might have a slightly yellowish tinge. They fluoresce blue when hit with UV light, thus offsetting the yellowish coloration slightly. You can read more about these agents here:</p>

<p><a href="http://en.wikipedia.org/wiki/Bluing_(fabric">http://en.wikipedia.org/wiki/Bluing_(fabric</a>)</p>

<p>Anyway, I doubt they would be used for anything except white fabric. Men's shirts, maybe. Men's suits, no.</p>

<p>(2) Even when these brigteners fluoresce, the color of the light your camera images is still blue. It doesn't matter that the exciting waveling was in the UV spectrum.</p>

<p>So no, I don't think UV fluorescence is a problem.</p>

<p>I think the source of the problem has more to do with the weave than anything. You might try printing some test patterns on paper and photographing them at distances that would coincide with the layout of the Bayer array. Pull out a calculator, and make certain each square is approximately (but not exactly) 1/2912 of the frame height. I'd be very interested in what you find. I'd do some tests myself if I weren't on the road.</p>

<p> </p>

<p>Sarah--yeah, I was asking just to see if there could be any possibility that UV was connected. The culprit I was thinking of is sizing, not brighteners. Sizing is made of similar 'petroleum products', I think. And not that it would cause moire, but that it would make it worse. So the weave is the culprit, but it is made worse, somehow, with UV. I know I'm probably dead wrong, but had to ask.</p>

<p>I know I've got a lot of testing to do. I was just going to try to find actual fabric that would cause moire with my 5D, but I guess printing out a pattern might work, although it wouldn't be drape-able like fabric. Note that the bluish suit is not comprised of checks, oddly enough, which is another reason it caught me by surprise.</p>