Another parameter that can be changed is the Video Edge Treatment. Essentially, SteadyHand works by shifting, rotating and zooming in the opposite direction as the motion on in the video. As a result, the video often ends up shifted off the screen, leaving blank spaces on the video. You’re given the choice to either leave these black spots, or to let SteadyHand zoom the footage sufficiently, so that it fills the screen completely again. This zooming does reduce the sharpness of the picture somewhat, but seems less distracting than having black space on the edge of the footage. However, if you own a plug-in that is specifically made for enlarging footage (such as Digital Anarchy’s Resizer), you will probably get a sharper image if you leave the black spaces and use your plug-in to do the resizing instead of letting SteadyHand do it automatically.
You can dive deeper into the parameters if you want in order to tweak the exact strength of the four different types of motion stabilization.
One limitation of SteadyHand is that it is only able to process AVI files; and while it only works consistently on DV-AVI encoded footage, it is nevertheless able export to any AVI format available, including uncompressed. If you have the drive space, it’s a good idea to export uncompressed in order to avoid a generation of lossy DV encoding.
Another option that I think needs to be included is the ability to process only a certain section of a video clip. For example, say I capture a five-minute clip, but only thirty seconds of it was handheld. Right now, there is no way to just process that thirty seconds, while leaving the rest alone.
SteadyHand only allows export to AVI, but you can choose from any of the installed codecs on your computer, or can export uncompressed for the highest quality.
Performance
SteadyHand actually took a lot less time to render than I expected. On my Athlon 64 3200+ PC with 2GB of RAM, it took around five minutes to render a one-minute clip. Not too shabby, especially compared to most After Effects plug-ins I’ve used.
As far as the quality of the video produced by the software, performance varies. I found that the “Normal” setting wasn’t strong enough for most footage, I really didn’t like what the “Zoom” parameter did to the shots (it seemed to randomly zoom in and out, and not in a pleasant manner). Nevertheless, the video looked great when the horizontal and vertical stabilization were upped to the maximum limit, the zoom was turned off, and the rotation parameter was turned way down (to limit a weird judder side effect). I did notice that it tends to glitch at the beginning and end of footage, perhaps because it doesn’t have a reference point to lock onto at either end. Also, while processing some handheld video of my band playing a show, I noticed that flash photography completely threw SteadyHand off, causing the picture to jump significantly. I’m assuming this is because there was basically a frame of pure white, so there’s no tracking information for the software to use, essentially causing it to reset the tracking and start from zero again.
