Video aliasing/moire solution

<p>Hi all,</p>

<p>I was just doing a bit of research about moire/aliasing in dSLR videos and learned that this issue plagues most models, the 5DIII being a brilliant exception.</p>

<p>I also found that one company has come up with a solution. After locking up the mirror, you insert their more aggressive anti-aliasing filter into the mirror box beneath the mirror. A plastic box/frame holds the AA filter close to the shutter and sensor.</p>

<p>It's definitely not cheap, at $365 for the 6D version, but it's a solution that works. Here's a link to their website:</p>

<p>http://www.mosaicengineering.com/</p>

<p> </p>

<p>I really have to wonder whether there would be more economical solutions:</p>

<ul>

<li>Firmware revision: The sensor's AA filter is sufficient to eliminate aliasing at full resolution, and the firmware should theoretically be capable of resampling to a lower resolution for the video without aliasing. Is the problem processing speed? I'm guessing the 6D, 5DII, etc. use a strategy of only reading off a small fraction of the pixels, hence the drop in sampling and Nyquist rates without a similar drop in the AA's rolloff sampling rate.</li>

<li>Screw-on filter for the lens? It might be possible to lower the lens' resolution with an optically irregular filter.</li>

<li>Spherical aberration: Perhaps a soft focus lens, or shooting at large aperture.</li>

<li>Diffraction: Shooting beyond the diffraction limit, perhaps f/22 or even f/32.</li>

</ul>

<p>But anyway, it would appear $365 (or less for other models) solves the problem.</p>

<p>EDIT: Hmmmm... I probably should have posted this to Casual Photo Conversations, because there's also a version for the Nikon D800. Too late?</p>

<p>You are better off with a more video-oriented mirrorless or camcorder. Those don't drop pixels hence can use digital anti-aliasing filter.</p>

<p>I should also mention that there's one little quirk with this filter. Because it's inserted between the lens and focal plane, it effectively changes the lens register, so that the lens focuses just a bit beyond infinity, and so that the focus scale is somewhat inaccurate. I doubt this is a significant problem, though.</p>

<p>Here's one filter solution, but I don't know what it costs:</p>

<p>http://caprockdev.com/antimoire.htm</p>

<p>Discussion here:</p>

<p>http://www.dvxuser.com/V6/showthread.php?200980-Caprock-Anti-Moire-Filters-Tested-on-the-7D</p>

<p>Dehaun, your point is well taken, but a dSLR as a video cam offers lens choices and low light performance not available with camcorders. Besides, if you already have the dSLR, why not use it also for video? ;-)</p>

<blockquote>

<p>low light performance</p>

</blockquote>

<p>I really doubt it. Dropping pixels also drops effective sensor area.</p>

<p>I've been running into this problem lately. In developing techniques for aerial work from a camera drone, my larger rig flies a Sony NEX-7. The problem: I'll use it for both stills and video on the same flight. So, no resolution-lowering filters for me! Stills look great, and video is usually very satisfactory ... but once in a while a tile roof or a row of fence boards or a walk paved with a pattern of stones can produce some pretty spectacular moire. When I know I'm up against that sort of subject matter, I just try to do several takes from slightly different distances, so I have my choice of shots in post.<br /><br />SOME of this can be worked on in post, but not well. The real solution is indeed a camera body that has enough horsepower to down-sample to something like 4K video on the fly, so that I can then downsample again in post as I put the video to work. Cameras like that are currently large and heavy (like the 5D III), and require a heavier gimbal, a bigger drone (like a one-meter octacopter, with bigger, heavier batteries, etc) ... things get large and a lot more expensive very quickly. I'm sort of hoping that Moore's law still has some miles left in it, and that smaller, lighter cameras can get around to processing on the fly. But being able to kick in and out of video and stills remotely means that, for now, I can't rely on an optical solution.</p>

<p>Matt, check out that second link of mine -- the filters that go in front of the lens. Perhaps you could rig a remote-control actuator to flip down one of those filters in front of the lens when you want to go into video mode. Just a thought.</p>

<p>Thanks, Sarah. Too fiddly, though! The gimbal is a precision-balanced device. Even a few grams of more widgetry out front of the lens would require delicate mods. And more to the point, this particular rig involves a completely 3-D gimbal that can operate in all dimensions. There are only so many slip rings in the moving parts, and they're all busy powering the yaw, pitch, roll motors, talking to the camera's IR input, and getting HD video back out of the HDMI port. Obviously this calls for a bigger drone! Let's see ... octacopter, 5Diii, new gimbal ... probably still less than $10k. We'll have to skip dinner out this week.</p>

<p>Perhaps the AA filter wouldn't need to be remote operated. Could you do your stills and video on separate flights? If your gimbal rig could tolerate just a tiny bit more weight, you could attach a small counterbalance arm/weight to balance the AA filter you attach to the lens. Just a thought. It seems far less scary than an octacopter with a 5DIII! :-)</p>

<p>Maybe switch to Panasonic? I think the GH3 has less in the way of moire and aliasing issues and is lighter, too. The announced GH4 will deliver clean 4K video a well.</p>