I've had my first 5D encounter of the moire kind. I have to say I was a little surprised given that the offending item of clothing was a plain dark-blue t-shirt. I've attached a series of stills. Note the psychedelic pattern on the shoulder on the left of the image. I was shooting with the L 24-70mm 2.8 set at the 24mm end with an aperture of 5.6 and ISO at 320. The shutter speed was 1/50 (to cancel 50 hertz flicker). Any ideas on what went wrong?

Even though its a solid color, the shirt has a pattern to it as you can see if you look close enough. If your focus is bang on the shirt... this will happen.
Here is what I do...
Just make sure your focus is further or closer then the shirt by a very small amount and your F stop is around 4 or less.

Hello Brad,
Thanks for your reply and the tips. Yes I guess if you call the weave of the cotton a pattern then the t-shirt has a pattern. Anyone who reads this forum regularly knows that EOS for video has its flaws and limitations. I posted this partly because I am yet to see examples of moire on this type of material and I wanted to draw attention to the possibility of this type of moire occurring.

Does anyone else have examples of moire on surfaces that don't have lots of lines or grid patterns?

I guess the key here is being able to predict when you might have problems with moire so that you can take action (whether that be a change of clothes or focal point) to avert any issues before they occur.

Yikes... that's definitely worse than I'd expect it to be. But I think Brad's right — I haven't encountered anything quite like this probably because my aperture is always more open than 5.6, and I usually set the focus a bit off the clothes (focusing on the subject's eyes in those screencaps probably wouldn't have resulted in the moire being as extreme, or even there at all).

You can see moire on most anything with fine details, though. It could be a shirt that's too focused, it could be a row of fences far in the distance. There's ways around it, but I think that often involves a shallower DOF, which isn't everyone's cup of tea.

Hello Brad,


I guess the key here is being able to predict when you might have problems with moire so that you can take action (whether that be a change of clothes or focal point) to avert any issues before they occur.

Wouldn't a key also be to have first rate monitoring? The built in LCD is too small to catch something like this I think.

Yes Brian, good point. I think good monitoring is something that will definitely help to avoid this type of moire. I just got my mini HDMI cable and will be using it with a 24 inch LCD on all future studio shoots. Location monitoring is something I'll look into in the future when I have a budget for it.

From now on I'll be sticking to lower apertures with the 5D and using my HDV cam if I need alot of DOF. While I'm sure there are some subjects that you might shoot at higher apertures without running into moire, I do wonder if it's worth attempting to shoot anything above 4.0, particularly at the wider end of the lens scale, (where the lens angle already gives you greater DOF).

The built in LCD is too small to catch something like this I think.

Agreed, although I think it's unusable for the opposite reason: the built-in LCD adds so much of its own moire that it gives me false positives. I see moire all the time on the LCD when there is none in the MOV file. (Zooming in 10X is no help either, of course, since that rarely ever shows moire when it does in fact exist.)

Well, I hate to toss a bucket of cold water here.

But I don't see any moire. I know what that looks like, I battled it in the early generations of digital cameras we got when I worked at an air force photo lab. It evidenced itself worst on blue and that meant air force blue class A uniforms.

From what I've seen these days it looks worse on CRT monitors where you have a distinct raster pattern. I'm looking at these images on a Samsung 21.6" LCD monitor driven by an Nvidia 8800GT 512Mb graphics card.

I don't see a bit of moire in these images.

That's odd Bruce, I click on the pics and I can see it quite clearly on my LCD monitor.
Avey

While I appreciate the need for precise language my main concern is not whether or not this is technically moire, but rather the fact that there is an undesirable artefact here. The question is how to learn from this and avoid it in the future.

I'm interested in the idea that the colour of the shirt might have something to do with the creation of this type of artefact. I've shot basically the same set-up (same F-stop etc.) but with a white T-shirt and there didn't seem to be any problems. Although it was far from being a scientific test I would definitely be interested if anyone had an explanation of why navy blue might cause more problems than white and whether or not this observation coincides with what other people have experienced.

While I appreciate the need for precise language my main concern is not whether or not this is technically moire, but rather the fact that there is an undesirable artefact here. The question is how to learn from this and avoid it in the future.



It'd be nice to have a sticky of guidelines with these DSLR's. So far we have,
-Avoid Apertures above F4
-Focus on faces not clothing when possible
-Avoid wardrobe with tight print patterns.
-Cityscapes in general are problematic
-Use shallow focus to defeat background artifacts.

feel free to add more

Also, where appropriate, use a softening or diffusion filter. I wouldn't use one on a gritty prison film, but a subtle one can work for romance or gothic horror.

A softening filter (assuming that it enhances the look) has a second benefit. It removes complexity in the image, so there is less stress on the codec. Even if you don't get rid of 100% of the moire, you will likely end up with a better image in the end.

Also, where appropriate, use a softening or diffusion filter... Even if you don't get rid of 100% of the moire, you will likely end up with a better image in the end.

Yes Jon I think this is important. Sometimes it's not so much about eliminating moire but rather getting it below a certain threshold of noticeability.

Since my original post I've had a chance to do a few more tests. Fortunately for the purposes of these tests I tend to only buy one brand of T-shirt, so the weave was identical on most of the colours I tested. Also for the purposes of these tests I tried to recreate similar conditions to those in the stills I posted above. I kept the aperture at 5.6 shutter speed 1/50 with ISO varied to test correct, under and over exposure. I also deliberately created the slight folding of the shirt that you see in the above images. Here's what I've found so far

-Well or slightly over-exposed whites render any moiré that may be present invisible

-Navy blue displays noticeably more moiré than other colours tried, with perhaps a slight drop in moiré when it is underexposed

-Red seems moiré free although a different textile was used so this needs to be confirmed

-Mottled light grey showed the second most moiré after navy blue although a plain grey will probably perform better.

-Lighter blue had less visible moiré than navy blue but it was still noticeable

-black was virtually moiré-free but once again this might depend on exposure (from what I've seen slightly under-exposed blacks will do better)

So these test were mainly dealing with the variable of colour and tone and to a certain degree exposure. I plan to test the moiré-threshold for navy blue with regards to F-stops in the next couple of days. Light position might be another variable worth considering, particularly as it relates to the way that shadows are cast, or not cast, by the clothing.

So from what I've seen so far solid whites and blacks seem to be the best tones for avoiding moire on clothing. But I should say that this is totally preliminary and that I would be interested if anyone can confirm or contradict these claims based on their own experience.

It'd be nice to have a sticky of guidelines with these DSLR's. So far we have,
-Avoid Apertures above F4
-Focus on faces not clothing when possible
-Avoid wardrobe with tight print patterns.
-Cityscapes in general are problematic
-Use shallow focus to defeat background artifacts.

feel free to add more

I like this idea. The "all things audio" section of this forum has three great stickies that cover both FAQs and general guidelines for getting great audio. Maybe there could be an "Issues and Fixes" (or something along those lines) sticky for the 5D and possibly other EOS cameras.

Your list on Moire is a great start (a list that we might soon add to if we can confirm some of the things I mentioned in the previous post). The jello effect and audio are two other potential issues that immediately come to mind and for which there is plenty of great information (if not fixes) on this forum that might be appropriate for such a sticky. I'm sure there are other issues that might be added to the list.

Well, I hate to toss a bucket of cold water here.

But I don't see any moire. I know what that looks like, I battled it in the early generations of digital cameras we got when I worked at an air force photo lab. It evidenced itself worst on blue and that meant air force blue class A uniforms.

From what I've seen these days it looks worse on CRT monitors where you have a distinct raster pattern. I'm looking at these images on a Samsung 21.6" LCD monitor driven by an Nvidia 8800GT 512Mb graphics card.

I don't see a bit of moire in these images.I was just about to post agreeing with you as I couldn't see any moire pattern while viewing these images on my 17" MacBook Pro. However as others seemed so convinced that it was there I look really hard up close & can now see it very clearly. However if I change the angle of the screen just a small amount the moire totally disappears while the photos look almost identical in colour, contrast, brightness etc So it appears that the display device has an influence on the apparent severity of this defect.

It would be interesting to see how bad the effect is on the original video as these screen shots are not even full 1920x1080 resolution.