Chroma versus Luma keying for LW animations.

[Tom Wood]

Hi All,

I'm reading through the manuals and trying to think ahead for my project. I'll be creating animations in LW that are mostly talking heads, which will then be used in VT to mimic a news program. I was thinking that I'd use a black background in LW so I can add the virtual set later in VT. This way I only have to render the backgrounds once and I can fiddle with the set without affecting the animations.

So I'll end up with the edges of the animated heads against a keyable background. The question then is - which keying method will look better in the final image - luma, chroma, or does it matter? Does my choice of file type matter, other than making sure to use 32 bit so I get alpha?

Related to luma keying, I assume that I'll need to make the blacks in the eyes one RGB unit off of pure black if I don't want the key to come through the eyes? Is that the preferred method, or there a more typical way of doing this?

Thanks,

TW

[Lightwolf]

Hi Tom.
If you use 32bit images, there is no nead to worry about chroma keying, the key is embedded right in your file.
Kame sure you don't have a 3D object in the background when you render, and set it to black for better edges (I assume the toaster works with pre-multiplied alphas, it should) and you're done.
I'd recommend 32bit Targa (.tga), since it is the simplest format and can be read by just about any application.
Cheers,
Mike

[Tom Wood]

Thanks Lightwolf,

When I place the single frame of the background image in the TED timeline, and copy it for the length of the scene, is it reloaded from the hard disk every time? If so, is there a more elegant way to do this in terms of computer resource usage?

Thanks again,

TW

[Lightwolf]

Hi Tom,
while I haven't used our toaster yet... Isn't there a way to specifiy the duration of a still? That way it would only be loaded once...

[Paul Lara]

Originally posted by Tom Wood
When I place the single frame of the background image in the TED timeline, and copy it for the length of the scene...

Tom,
No need to copy it. When you drag in a still, the default length is 5 seconds. Grab the right edge of the still and drag it out as long as you want!

[Tom Wood]

(rubbing hands together)

Eeeexcellent....

All those new SCSI drives are loafing along now....so let's see if I can add some sweat.

I'd like to make use of Digital Juice JumpBacks on parts of the virtual set behind my animated character. The virtual set will include panels on which additional video, like the JumpBacks could be projected. I could use the PIP DVE to position them on the panels in the virtual set, but it seems to me that it might make editing easier if I could just make different parts of the virtual set "active" and tell VT to place selected video on selected active areas. (Since the set is curved, this would be especially true.)

I realize the background is probably no longer a still, but I would go back and forth between the "active" set and the still, depending on the need. Is there a feature in VT2-3 that can do that?

Thanks,

Tom

[dch]

Originally posted by Tom Wood
I could use the PIP DVE to position them on the panels in the virtual set, but it seems to me that it might make editing easier if I could just make different parts of the virtual set "active" and tell VT to place selected video on selected active areas. (Since the set is curved, this would be especially true.)


If you need to project your Jump Backs onto a curved surface you will need to do this prior to bringing it into VT. If all you need is flat projections then this can easily be done in VT using either PIP DVE or using the Positioning function in the Edit panel. If you use shots that have the same camera position then you can set up the projection once then use the "paste(inherit)" into copies of the same setup, to put new or different Jump Backs into the same area on the screen at different points on the timeline.

David Hunt