Originally Posted by Wolfgang S.
The deinterlacer in Vegas is still not as good as it could be - that is one reason why it makes still sense to use special applications like Tempgenc, for stepls like downsizing of HD-footage to SD-footage for a SD-DVD. What has been improved is the capability to render 1080 60i to 1080 24p, what is remarkable that this has become better - a lot of people are satisfied with that now, even if I for my part do still not like the rendering from 1080 50i to 1080 24p.
For edge violations at the left or right side you can use floating windows. At the bottom the are less critical. But I think we still have to be carefull here about the stress that we put in the video - not everybody has become s3D stress resistant like us, trained by sitting at the editing computer.
Yes, if you're careful not to pan too fast, Vegas does a very respectable job of de-interlacing 60i to 24p now. At least that's my experience.
I have lots of shots with edge violations that floating windows won't fix. For instance, one thing I always liked to do when shooting 2D was use a tree or other plants to frame a distant object. It was a way of adding "dimension" to flat shots before real 3D. I still like to do that, even when shooting 3D. For instance, I place the camera so that I get part of the trunk of the tree, and an overhanging branch, to frame a bridge in the background. This works great in 2D, but depending on the distances, it can cause edge violations in 3D. If I'm close and I manually converge on the tree trunk, the bridge spacing (left/right views) in the background is too wide. I might not have a problem watching such a shot in 3D, but many people would. On a large screen, it might not be possible for some people to converge the bridge at all, and it would likely cause headaches in many. So, is that shot unusable? If you believe that edge violations shouldn't be allowed, yes. I'd have to frame the shot entirely differently, or not use it at all. However, the JVC's auto parallax control adjusts the 3D spacing so that the middle ground becomes the point of convergence for that shot. This reduces the left/right spacing of close and distant objects, but it creates an edge violation in which the tree trunk extends into space in front of the monitor, The edge of the screen bisects the trunk. The good thing is that neither the foreground object nor the background object is so widely spaced as to cause viewing discomfort.
Remember Cameron's first rule of 3D - there is no screen. For him, the fact that the tree protrudes out of the monitor is not a problem. He says that the only time this is a concern is when you do the 3D gag in which an object moves straight out of the background and toward the viewers' eyes. Then, the edge violation calls attention to itself and shouldn't be used. Consider "Avatar." Cameron moves his cameras all over the place, in and around trees and bushes, through doorways in the labs, etc. Edge violations are everywhere in Avatar, and if you watch what happens in any given scene, you see that the way he shoots makes eliminating them all but impossible. Most Imax films pay no attention to this "rule," either. And look at many current 3D releases, including Scorsese's "Hugo" or "Open Season" (the animated film voiced by Debra Messing and Ashton Kutcher). Even in animation, where avoiding edge violations is easier, many filmmakers just don't do it.
Generally speaking, edge violations in the periphery of my vision (at the frame edges) don't bother me. They don't cause my brain to reject the 3D-ness of the shot. I've decided that trying to eliminate them is not worth the cost. For moving shots, it's clear to me that you either use them or you scale back motion so dramatically that you cripple yourself creatively. And if you make that exception for moving shots, it's easy to make it for many static shots, too. If it works for one type, and doesn't confuse your brain or your viewers', it's not a "rule" you have to follow. That's my take on it.