[Stuart Elith]

David, my advice is to listen to Dave!
I have recently been doing some work on something that sounds quite similar, with actors speaking to camera but moving around slightly because they were trying to get “into” their role… but it meant a lot of pain for us!

In my case, we had tracking dots on the nose and 2 on the temple/forehead, and even so, had mediocre results with stabilization – any facial movements seem to distort the face more that you might think… a smile makes the nose shift slightly, any eyebrow movements really affect the whole forehead, etc.
And I tried tracking the eyes but it was painful and still not a great result, even after I cleaned up the blinking frames etc.

If a reshoot is at all possible (and you can give instructions that will avoid your problems recurring) then push for that, I think.

[Stuart Elith]

Yup, that’s the other main way to do it… I find it slightly annoying, but it works fine, and you get used to it pretty quick 🙂

[Stuart Elith]

Yeah, also not sure really what you mean. Perhaps, though, you could pre-comp all your text layers. Then you can apply an effect which will affect all of the text, because it’s now being treated as one layer.

This *might* help… depends on what the ‘window opening’ means.

[Stuart Elith]

Sure there is 🙂

Motion Tile will do what you are looking for… it takes a source layer and gives you options to scale the original source and/or expand the output. A combination of the two with good timing will allow you to make it look like the camera is panning out or splitting.

Note that it is using the ONE piece of footage, so if you wanted it to be lots of different people at once, you’d need to do it very differently.

[Stuart Elith]

Alternatively, you could split the layer so the blurred bit is on its own layer. This is sometimes an easy way to do it and you can find the part easily when looking at the timeline. But it can complicate things if there is a lot of stuff happening.

[Stuart Elith]

Unfortunately, the general consensus seems to be that rotoscoping (masking) is your best option in these circumstances. There is an effect, Difference Matte, which tries to do what you are saying by comparing pixels, but it doesn’t provide good results usually, for many reasons. Most people don’t bother with it.

Many will suggest you go for a reshoot for best results, with a greenscreen background you can key out. It certainly makes things less painful… otherwise, I think you’re stuck with roto.

But you will find that 12 seconds actually doesn’t take that long… and you will learn through the process, roto does get a lot quicker once you get into it (though it is always a manual procedure).

[Stuart Elith]

I think there are a few ways to achieve what you want… I can’t see the problem with using Replace Footage in the Project window, like you mentioned.

Otherwise, you can also select the layer you want to swap, hold alt and drag out the new item you want to use and drop it anywhere in the comp window. It will swap it over without getting rid of effects, keyframes etc.

[Stuart Elith]

I don’t think you are reading the tutorial carefully enough… have you tried following along with it?

He DOESN’T parent the text to the image. When you are working in 3D space, any objects that are 3D (using the 3D layer switch) will automatically respond to the camera movements. So if you put an image in the scene, and have some text also positioned in 3D space, they will both move in a “realistic” way when you move the camera, which is what is happening in the tutorial.

Steps 121 – 132 are dealing with text and all the steps that have info about camera keyframes and movement (such as step 146) are to do with the camera and the movement.

If you don’t understand what is going on by reading hte tutorial, I suggest you take a step back and learn some more basics. Of course everyone wants to create cool stuff, but by trying to do that without good foundations, you will either have constant issues and get frustrated, or otherwise only be able to copy other people’s things…

[Stuart Elith]

Here’s a script that helps with this… not sure if it’s worth the effort when you’re just playing with one layer (you need to run a script, adjust, then run another script to finish the step) but it does what you want 🙂

[Stuart Elith]

File > Collect Files… is what you want, I think. You can choose from different options as to what to package, be it individual comps, your whole project, or whatever 🙂