Real world examples: moire and rolling shutter
 

I shot my first paying job yesterday on the 5D mkII, with mixed results. Overall I am pleased with the footage, but I thought I'd share with you two examples of where the 5D mkII can go horribly wrong.

We shot a ten minute interview with a park ranger. He was wearing a simple cotton shirt, which led to moire throughout the footage.

Worse is the pronounced rolling shutter on footage shot from hilltop vantage points. Why? It was a windy day. Even on a fairly robust manfrotto tripod, vibrations caused by the winds buffeting the camera led to.... well, see for yourself :-(

So, the moral of the story is, be wary of men in uniform, and stay inside on windy days.

Canon 5d mkII Moire & rolling shutter on Vimeo

Interesting results, Josh,

Frankly, the moire isn't too bad downsampled to 720p and h.264 re-encoded. I'd imagine that it's a lot stronger on the original 1080 footage.

Man, that was windy! It was strong enough to blow the mountains around!

It would be interesting to put this through After Effects and lock a couple of points of the mountains. It would let us decouple the shake from the tilt and accordion effects. I wonder if After Effects would be able to lock enough points and do it's own morphing to take the squish and skew out of the equation. I've never tried any such tricks...

Probably easier to sandbag the tripod or wait for a calm day.

Hi Jon, while the moire isn't "too bad", it does pop out a lot more in 1080. Given that this was my first talking heads shot, and the subject was wearing a fairly innocuous looking shirt, it's definitely an artifact I'm going to be wary of in future.

The winds are pretty strong where I live, but I've never had any trouble with camcorder (DV/HDV) footage in the same situation. Unfortunately, as with the client's shirt, sometimes we can't choose our shooting days. The tripod legs were actually quite secure and the whole rig was barely moving; the 5D sensor seems extremely sensitive to rapid vibrations - the resulting rolling shutter exaggerates the severity of movement.

That said, I'm not complaining. I've already captured a lot of beautiful footage with the 5d, and in 80% of situations I prefer it to my former cameras, the XH-A1 and XL-H1. It's just important to be aware of its limitations.

So true. My hammer is great with nails, but really makes a mess with screws. And my sledge hammer is a bit much for both. ;)

So, what tripod/head were you using? We've been too cheap to buy a good tripod so far, and have rented when it was important. I'm curious to know how beefy we need to go for non-ideal conditions. Clearly, the 5DM2 needs more stability than an HVX, but at least it's not near as rubbery as the D90.

I encountered a similar situation here in Seattle recently. Luckily I was using the 5D Mark II as a backup camera, allowing the Sony PMW-EX1 to do the real work. It's weight made it more stable on the Manfrotto tripod since there wasn't enough mass for the 5DMkII to fight the physics. My solution was to order the RedRockMicro DSLR kit which added about 12 pounds fully kitted out. The wind bouncing around the camera is no longer an issue.

We did tons of tests in the studio and on location comparing the two systems. At the end of the day we finally figured out how to do most of our typical shooting needs with the Canon. Obviously there are some nice things such as 24p, timing, exposure, etc. that the Sony offers out of the gate. Luckily what we couldn't do on the Canon we could do in post via Adobe AE CS4 and a few other programs.

My goal is to shoot this entire month using a few of the 5DMkIIs for multi-cam shoots and nail down our workflow. It's been a hell of a ride so far with great results.


Contents:
RedRockMicro DSLR Kit (Cinema w/Indie)
VM8000HDMI Monitor
Marantz 660
Rode NTG-2
Tekkeon MP3450
Canon 5D Mark II

Canon 50mm f1.2 L
Canon 85mm f1.8
Canon 17-40 f4 USM L
Canon 70-200 f2.8 IS USM L
Canon 100-400 f4.8-5.6 IS USM L
Canon TS-E 24mm f3.5
Canon TS-E 90mm f2.8

From earlier testing, I'd already figured out a monopod is no good beyond 50mm focal lengths, even indoors! With my 105mm lens on a monopod, any small movement sends a shiver through the image.

For this shoot we were out in forests and hilly terrain. Part of the appeal with the 5dm2 is its size, and I thought I could get away with downsizing tripods too. So I left my 516 with beefy sticks at home and trekked up the hills with lighter 755 legs and 701 head. It was fine out of the wind, but in exposed conditions the 5dm2 must be *absolutely* locked down. When those shots were taken, the small movements weren't apparent, the tripod was stationery and ground level, and on the viewfinder it didn't seem to be a problem. I've shot in the same conditions (handheld with OIS!!) with an XH-A1 without incident. It's my first experience with a CMOS camera so the results in the editing room were an unpleasant revelation for me.

Mark's solution sounds right, keep a ton of weight on the rig in exposed conditions.

Josh: what lens were you using? Was IS used or not?

TIA,

Ron

Hi Ron, it was actually a 28-105mm Nikkor zoom, non-IS. Yes it would definitely be interesting to see what a good IS lens could do in a similar situation.

Did you have the sharpness turned down or was this the default setting?

I've attached a pic at 100% resolution. The bad news is, sharpness was turned down to 1 (in an eight step range from 0-7). I've encountered the effect to a much more dramatic extent shooting in the city (tiled railway platforms for example). What disturbed me here is that the shirt wasn't the kind we'd normally advise against (no frilly patterns, stripes etc).

Two things...

1 I am fully prepared to defer to higher powers on this matter, but it doesn’t look to my fairly well-trained eye that the camera is causing the moiré effect. Rather, it seems to be due to the fabric of the shirt (are you sure it wasn’t partly polyester?) and the way the light is reflecting off of it. If not, why are only parts of the shirt banding, and others not, and why is the moiré following the folds of the shirt?

2 The rolling shutter: I don’t see it being the main problem here. This is sped up footage, right? (If not, that’s some serious wind.) What I mainly see is a very unsteady tripod under windy conditions that would look unacceptable, rolling shutter or not. The solution to this obviously is not to "stay indoors" but to get a sturdy tripod.

Hi Mark, I posted the footage simply to assist others with their choices. I'm not a tech-head by any means, but I have owned half a dozen HD cameras without ever experiencing these issues, which I routinely encounter with the 5dm2.

The moire is real, irrespective of this example. Others have posted far worse instances (some quite comical). Whatever the technical explanation for this particular artifact, it's not something I have encountered shooting with the DVX, Z1, XH-A1 etc.

As for the rolling shutter, I can only restate that the perceived tripod movement is exacerbated by the rolling shutter itself, and that I have had no issues shooting in the same conditions (for the same client on the same terrain) with CCD cameras and OIS. I agree a heavy duty tripod would mitigate the issue, and that was partly the point of posting, to help point out what you need with the 5dm2.

The footage was not sped up at all, that was a typical Wellington day :-) I'm used to shooting in the wind; with the same tripod and an XH-A1, wind is not an issue. But I know for next time.

What you're describing is pretty typical. The effect will follow the folds of the shirt, because that's where the critical frequencies are. Where the shirt is flat to the camera, the frequencies will be lower, and where the shirt is perpendicular to the camera, the frequencies will be higher. When the shirt fold is at that perfect angle, the frequency is high enough to alias, but not too high to make it through the camera's optical filter.

The problem with aliasing is that you will get large bands of lighter and/or darker areas. And, yeah, it can look like polyester shimmering. It doesn't matter how much you filter the image or scale it down, because the high frequencies have already been misinterpreted as low frequencies.

With the right optical filter, you'd lose detail on the shirt folds, and the aliasing will be reduced to more acceptable levels. Downscale to 720p, and the result should look great.

Thanks for your feedback, Josh. I assumed that IS was not used in the rolling shutter example - in my experience, Canon EF IS eliminates such movements fairly competently. I'll be off on a shoot soon, using largely EF super telephotos (300 - 1000mm in various lens/TC-combinations) from within a vehicle. It will be interesting to see how the 5DMkII will respond to the slight movements common in such shooting situations, and what effect the IS of the different lenses will have on video (no problem with stills in my experience)...

Cheers,

Ron

Ron it would be great to hear how that goes. I don't have any IS lenses for my 5dm2, having chosen to go mainly with Nikkors primes with adaptors. Telephoto primes (even 100-135mm) on the 5dm2 are hopeless on a monopod - even tiny vibrations bring rolling shutter into play. I'd be interested to hear if short telephotos with IS can be used on a monopod.

It's all very well to cart a hefty tripod around to counter rolling shutter, but sometimes it's nice (or necessary) to travel light.