[Steve Roberts]

The built-in Lens Flare effect?

[Steve Roberts]

…or just follow the link in my post. 🙂

[Steve Roberts]

If you follow the link, then cut the COW stuff out of the link and re-paste it, you’ll see the movie.
https://www.allbetsareoff.com/Cow_Stuff/Cow_Light_CD.mov

Nifty, Aharon!

[Steve Roberts]

[beenyweenies] “Motion and Maya both will take better advantage of the card… “

Along those lines, Apple’s Color app is very dependent on the graphics card. But beware, it’s a finicky app, not for the timid.

[Steve Roberts]

[beenyweenies] “Just be aware that the files you render from AE will only be 8-bit even when your project is 16-bit…”

… unless you render to a 10-bit codec. But unless you output that 10-bit file to something like Digibeta through SDI, it’s quite likely that your final output will eventually be 8-bit, as beenyweenies wrote.

Now sometimes, working in 16-bit then rendering to a 10-bit intermediate file anyway can preserve quality even when the job will end up in 8-bit. Hey, do some tests. 🙂

[Steve Roberts]

Stroke applies a line to a mask, making it visible.

There is no information contained in a font that will allow us to write it on. We have to do it manually.

[Steve Roberts]

Hey, no problem. 🙂

[Steve Roberts]

Layers 2 and 3 should not overlap if each anchor point is on the edge of each layer.

For Layers 1 and 4, if you parent them to 2 and 3, you can keep them from “rotating” if you rotate them in the opposite direction from their parents. If 3 rotates from 0 to 45 degrees, rotate 4 from 0 to -45 degrees. Don’t animate their position.

To make everything push back, parent 2 and 3 to a null and move the null back.

Are you following my instructions from the beginning? It sounds as if you’re trying to shoehorn them into something you’re already doing, and the two methods aren’t mixing well.

[Steve Roberts]

All right …

Imagine layers 1,2,3,4 from left to right, seen face on. Move the camera (make one explicitly) for a different POV)
Each layer will pivot around its anchor point.
The anchor point for layer 1 should be at its right edge. Use the Pan Behind tool.
For layer 2: at its right edge.
Layer 3: its left edge.
Layer 4: its left edge.
Parent layer 4 to layer 3, layer 1 to layer 2.

Rotate layer 2 and 3 on their Y-axes, then rotate layer 1 and 2 to follow.

Does that make sense?

[Steve Roberts]

One way ($$$) is to use motion control to shoot two identical passes: one with, one without.

Now unless you shoot it twice identically that way, one thing (talent or room) has to be CG.

You could shoot the talent on greenscreen with markers, matchmove it (syntheyes, etc.) and create a 3D room around the person using the 3D camera created from the matchmove data.

Another way is to shoot it without, then matchmove (syntheyes, etc.), then using a 3D camera created from the matchmove data you model a 3D person with texture maps from reality, and place that person in the scene. That’s more difficult, but great if the talent has to perform superhuman feats.

All those might be out of your budget. You could try staging the scene so it’s tracking, and the talent is walking with the camera. You shoot the scene without, then reshoot with a greenscreen on wheels behind the talent as he/she walks. Or do that in the studio so you don’t have to roto the feet. The idea is: the talent is essentially static to camera, and his/her position can be calculated by just measuring to set her position … as long as he/she can keep pace with the tracking camera. Poor man’s motion control?

Anyway, I’d look into ways of staging it so you have the talent in a medium shot (no feet), and the camera is tracking steadily, or you’re spinning around a fixed point where the talent is (Scorsese, Color of Money) not moving. It might work …

Anybody else?