[Steve Renard]

I don’t work in PAL much, but in my experience the best quality I’ve gotten with NTSC when dealing with a 16×9 anamorphic project has been to scale the horizontal value and lock the vertical value. ie – for a 720×486 anamorphic timeline in FCP, I would use an 864×486 comp in AE. Trying to have Final Cut scale down a 1920×1080 into an anamorphic timeline will be a render hog and the quality will not be very good, IMHO.

Cheers,
Steve

[Steve Renard]

Can you post some screen shots? I’m not sure I understand the various issues you’re running into.

To guide you on what to post, here are the things I’d like to know that might help me answer your question:
1. How are you achieving the panning effect? Are you moving a 3D camera over the layer? Is it possible you’re in too tight on the layer, causing it to break down? Or that the layer size itself is too small or has been scaled in a weird or unexpected way giving you pixelation?
2. If I understand correctly, you have multiple layers with mask lines of them, and you are trying to unify those by using a Stroke on an adjustment layer. Is that correct? (I’ve never done this myself but it sounds like what you’re describing) If so, why not copy all the masks to one layer and the put the effect on that layer?

Whenever I get frustrated with AE, it invariably turns out to be something ridiculously simple that I did not notice at the time. Take 5 minutes, get yourself your beverage of choice, sit back down, and retrace the pieces of the comp to figure out where something is going wrong.

Good luck,
Steve

[Steve Renard]

That has been my experience as well. Does that mean the only way to access the control points of a path is through a separate .js script file? Did Adobe do something to kill that functionality in the expressions language, or is it a question of being limited by the interface?

FWIW, I’ve already put in a “Feature Request” with Adobe for “animated mask points” that would more or less eliminate my problem – is there a known reason why they only allow you to animate the entire path, and not the separate points? I’m not a hacker by any means, but it seems like a fairly simple proposition to write the program so that if you have a mask, you can twirl it open and get a bunch of mask points with keyframe stopwatches, and each of those can be twirled open to give access to the bezier control points, also keyframable.

I’m sure I’m revealing my ignorance here, at the very least…

[Steve Renard]

That’s perfect – thanks Dan!

[Steve Renard]

I’ve tried this on my end and the best I’ve got is that maybe if you adjust the brightness/contrast settings of the Fractal Noise effect (or put a Curves effect on that layer below the Fractal Noise) you will get different results. It sounds like you need to raise the contrast and/or lower the brightness so that the dark areas of the noise are truly black (R0, G0, B0), and therefore will not interact at all with the layers underneath. Even R1 G1 B1 will interact with the layers underneath in Add mode.

It’s not clear to me why the rendered video would be different than the noise effect on a layer, unless you’re rendering to a color space that is crushing the blacks on the way out, giving you something that might look almost the same, but actually has more “true” black pixels than your noise layer. For confirmation, you could also render your noise to the Animation codec and try using that rendered video to replace your noise layer – if what I think is happening is the case, then you will see the same results.

[Steve Renard]

Rendered premultiplied. The settings are correct in terms of the alpha in FCP, etc… but it doesn’t seem like that’s the problem, because even if there’s no alpha (AE generated text over an AE generated image, with the QT file, for example) I still get the aliasing. The alpha only comes up because that’s where I can’t escape the problem by re-rendering in ProRes.

I thought it might be a frame size or scaling issue, but I’ve tried many different permutations of sequence frame size vs. rendered file frame size and scaling on AE output, etc. and none of them makes any difference. For whatever reason, anything rendered out of AE with the animation codec looks great in QuickTime, etc., but comes through ugly in FCP.

That gave me another thought, though, which seems to have fixed it – I was using the 1440×1080 frame size for the sequence, but changed it to 1920×1080 (which is the size of the rendered files from AE, but not the size of the footage) and that seems to have cleared it up. Still working on the logic behind that, but at least I’ve solved the problem for the moment.

[Steve Renard]

I did render.

I’m aware that the new ProRes codec has alpha support, but that system/software upgrade is still in the future, somewhere, unfortunately.

[Steve Renard]

Dave – I completely agree with you on the “limited support” factor. From my own recent research into CS4 (still running CS3 here as well, though with some possibility of an upgrade soon), the process to bring a 3D object into CS4 looks not only completely asinine, but also bordering on useless. As Jeremy said you have to make your model, bring it into Photoshop, then save it as a PSD with a 3D layer, then bring it into AE, and from there it looks like there are still several more steps to make it all work.

Has anyone out there with CS4 tried this out? Is it more useful than I have been able to describe above?

[Steve Renard]

Not sure if this is your problem, but when I installed CS3 the first time I had to go back and install the CC plugins separately. I can’t recall what the process was, but try running the installer again and see if there’s an option to install the plugins separately. As I’m thinking about it, too, there might also have been a separate installer for the CC plugins, or I might have simply copied them from the install disc into the AE CS4 plugins folder.

I do remember having that problem with the CS3 install though, and got it sorted out eventually, so hopefully you will have the same luck!

Steve

[Steve Renard]

You will have to tweak it a bit, but that tutorial should give you the tools in your toolbox to do what you want to do. The issue will be that that tutorial is based on a stationary camera on a tripod panning across a scene. What you’re doing here is more complicated in that the camera perspective itself it changing, but you should be able to avoid any true roto work by making a separate painting for each of the 5 or so perspectives you have in the shot you posted.