Comments

Coursedesign wrote on 7/16/2005, 7:51 AM
It is a lot of work to do this ("rotoscoping"), but that's what's done for many feature films for example.

Vegas isn't really a suitable tool for this, other than for very short sequences.

Combustion and several other programs have a kind of automated tracking of mask points from frame to frame, this helps tremendously.

It is still a significant effort to make it perfect though, can't be helped.

If you want to do it in Vegas, try adding a LOT of feathering on the mask covering the person's body, to make it less critical that the mask fits perfectly in every frame. This will change the look slightly but may be OK depending on exactly what you want to do.

Remember that if the text passes only behind the person's chest say, you only need to mask the chest (assuming no flailing arms...).
epirb wrote on 7/16/2005, 8:21 AM
Sony Chaboud had a tutorial on this I dont remember where, but the title went behind trees and buildings and all over the place .... really cool, but the tutorial is a bit complicated, walk in the park for him, advanced studies for me : )
Ill try and do a search......


ahhh here it is this is his web page and at the bottom click on the "shadowed text" links.

http://matthew.chaboud.com/vegas5/
Throw togethers...that kills me it'd take me forever to concieve and do this

Very Cool, I gotta learn that when I have alot of time to kill.
arem wrote on 7/16/2005, 3:02 PM
Throw together? Yeah I don't think I could do that in less than... 4 days. He even had reflections and shadows. Wow. Well, thanks for that link.

I'm going to stick with frame by frame masking for now. Just short sequences though. :)

-Dan
Edward wrote on 7/16/2005, 10:25 PM
I found out that in bezier masking, you gotta preview your footage in 'Best' mode. If you use draft, preview or good, your rotoscoping won't be accurate as you worked so hard to achieve (especially since I render out to NTSC DV Best)..
With interlacing, it really helps. Go every other, or every three frames. You don't really have to use feather. The interlacing will kinda fill in the gaps between keyframes if your footage is 'lower field'.
Did some stuff for broadcast with it, and it looked sharp!
Veggie_Dave wrote on 7/18/2005, 2:53 AM
>I found out that in bezier masking, you gotta preview your footage in 'Best' mode.

I've also only recently found out that the definition of the image in either pan/crop or track motion is controlled by the standard video preview setting.

I wish I'd known this a long time ago...
arem wrote on 7/18/2005, 10:21 AM
Thanks for the heads up. It would have been horrible to go through 30 seconds worth of footage and then have to redo it.

Thanks again!
-Dan
GregFlowers wrote on 7/18/2005, 10:57 AM
Some compositing programs allow you to make "difference mattes." I'm not sure if Vegas can do it or not. You will need 2 shots. One of the plate (background) footage by itself with only the things you want behind the text. Then you have the actors, props, or whatever things you want in front of the text placed in front of the background stuff you just shot and film that. The compositing program can analyze the pixels from the two shots and make a matte with only the "difference" between the two. In this case all of the foreground elements would be different and placed on a matte of their own, kind of like what chromakeyed footage looks like. You would place the background shot on the bottom track, followed by the text, and place the "difference" matte on top. This way the text moves behind all of the foreground stuff and in front of the background. The difference matte usually requires a lot of clean up and only works well with a stationary background. If there are moving elements in the background like moving clouds or trees waving in the wind then it won't work well or at all. Hope this helps.
GregFlowers wrote on 7/18/2005, 11:02 AM
I just looked at the link supplied by "epirb" that discusses difference matte creation in Vegas. Sorry for the redundancy.
Greg