[Ryan Hill]

Well, to expand on what Dan said, you can make a slider and then pick-whip your variable to that slider.

This’ll result in a longer piece of text in the expression, which you’ll still need to put in all of the expressions, but it means if you later decide, “Oh, I want to see what happens if this is 5 instead of 4,” you only need to change the slider.

[Ryan Hill]

I’d have to cut short the one row of pixels.

If plan everything out, I might be able to create specially-sized compositions.

Lemme think. For the 40×22.5 example, I pad the bottom with 16 extra pixels and make a composition that’s 1280×736 and then divide it into 40×23 using the mosaic filter. And so the bottom row doesn’t end up darker or translucent, I have to double-up the footage so the bottom 16 pixels are repeated twice.

That should work, it’ll just take time to set up each special size properly.

[Ryan Hill]

How about combining time displacement with displacement mapping?

Because you basically have a whole series that are displaying the same location but at different times. So play with a gradient with displacement mapping, until you get everything displaying the same column of pixels, and then you apply the time displacement gradient to that.

[Ryan Hill]

Well, it’s converting from 2D to 2D but the conversion depends upon how the two 2D layers are related to one another in 3D space.

I don’t really know what 3rd-party plugin would be suited for this, but maybe I can figure out the math:
https://en.wikipedia.org/wiki/Line-plane_intersection
I can use that math to convert assuming I know the 3D position of the four corners of the bounding rectangle. I might end up matching those corner positions by eye, but maybe that won’t be too hard, especially if I get rough measurements when we shoot.

[Ryan Hill]

I’m not sure I understand? I’ve used toWorld before, but always to convert from 3D layers that I’ve made.

In this case, say I have footage of something moving along the floor, and I’m tracking it. But that’s just tracking its position as the camera sees it, which is affected by the perspective of where the camera is. And I want to be able to position the point on a 2D map of that floor.

[Ryan Hill]

Does it just move in a straight line? If so, you should be able to make the y value depend upon the x value if you get the slope of how much both change.

For example, if X moves from 100 to 300 and Y moves from 201 to 251, then your formula would be:

Y=(X-100)/200*50+201;

[Ryan Hill]

I think you can also use .toFixed()

[Ryan Hill]

I’ve sometimes filled holes from keys by playing with extreme values of matte chokers. Use a big negative choker until the hole is filled in, then a big positive choker to remove it from all the edges.

[Ryan Hill]

I had to look up what a Vision Station is, so I can’t give you a definitive answer.

This is what I read, if it’s worth anything:
https://computer.howstuffworks.com/elumens2.htm

It’s not promising, since it only talked about producing footage for it using a rendering engine. And that methodology required rendering four versions and recombining it.

My first thought would be to shoot with a fisheye lens and test that out, but I worry if it was that simple, they would just render the 3D with a simulated fisheye lens instead of that setup with 4 renders.

[Ryan Hill]

I don’t think there is a “wind” filter in AE.

Maybe you could achieve something similar with a displacement map, filled with horizontally stretched noise? That’s the best idea I can come up with.