Mylenium
Forum Replies Created
-
I don’t quite get what you mean. Your post sounds a bit paradox. Do you mean that your objects shall home in on the “leader” from their initial positions and only then form a trail/ swarm? That would be pretty tricky to do, since it’s close to impossible to get smooth movement just with expressions. How many objects are we talking about? Would it be sufficient to blend between different manually animated motions? Have you looked into using particle systems?
Mylenium
[Pour Myl
-
Build the “room” from 3D-layers with the Grid effect applied.
Mylenium
[Pour Myl
-
Have you considered re-creating the animation yourself? The Blue Marble maps are freely downloadable from NASA and can be used with CC Sphere. This would allow you to integrate your arrows easily. At least you should download some low-res version of the maps as a placeholder and align your arrows in a sub-comp. When using this on CC Sphere you should be able to align your stuff more easily.
Mylenium
[Pour Myl
-
I don’t think pre-rendering will solve your problems. AE’s filtering will still soften your pixels if you get too close to your stuff. The only option is to increase the actual pixel resolution.
Mylenium
[Pour Myl
-
Take a look at Jayse Hansen’s “Growing Vines” tutorial
Growing Design Elements in AE. While it uses vector art, it would work on other footage types as well using similar techniques.
Mylenium
[Pour Myl
-
Yeah, I think that would be a very sensible compromise and second that suggestion.
Mylenium
[Pour Myl
-
Dunno, maybe you are looking at this the wrong way (to copy your words from your other post) ;o). Since text layers have no real per-letter-Z-depth, you may have a hard time. The only way I can think of to really precisely control this is to have each letter on its own layer and that’s pure madness. A better solution may be using Card Dance (again) in combination with custom maps. You may need a few layers of Card Dance, but I think it’s still easier than animating everything manually. For chopping the text into pieces, you should look at using transition effects – Venetian blinds can be wonderfully used to isolate separate rows and columns and you can then get rid of excess letters by applying another instance of e.g. a linear wipe. The only downside I see is that you have to work with many, many sub-comps.
Mylenium
[Pour Myl
-
Dunno, maybe you are looking at this the wrong way (to copy your words from your other post) ;o). Since text layers have no real per-letter-Z-depth, you may have a hard time. The only way I can think of to really precisely control this is to have each letter on its own layer and that’s pure madness. A better solution may be using Card Dance (again) in combination with custom maps. You may need a few layers of Card Dance, but I think it’s still easier than animating everything manually. For chopping the text into pieces, you should look at using transition effects – Venetian blinds can be wonderfully used to isolate separate rows and columns and you can then get rid of excess letters by applying another instance of e.g. a linear wipe. The only downside I see is that you have to work with many, many sub-comps.
Mylenium
[Pour Myl
-
[karim] “Anyway, which 3D tracker did you work with and if I timelapse the clips (Material is already filmed with a stabilized camera-system on Betacam-SP) to 5 Minutes, would rendering then be in the range of possible?”
Well, I do not do the tracking myself (as we so rarely need it, we outsource it to anothe company), but I have had a chance to test pretty much all available tools and personally I think Matchmover is a good combo between acceptable speed and good results. PFTrack seems to be even better quality wise, but is pretty slow and requires a lot of manual intervention.
Speeding up your footage won’t do much good, it will only make the results less precise. Unless you intend to use such sped up clips in the final videos, I’d stay away from this method – you’d end up with data that equals a linear and thus jumpy camera track once you stretch your tracking data to the full length again.
Front Projection is a method to use photos or live video in 3D programs – you model simple geometry like cubes and project your footage on them as seen from the camera. This will make the geometry effectively “disappear” but since it gives you obscuration, you can e.g. still have an airplane fly behind a mountain and get the correct look.
Mylenium
[Pour Myl
-
[karim] “Anyhow… First I wonder if 30 Minute Sequences 3D tracked are heavy stuff considering file-size and render-times.”
Takes a week to setup, a week to calculate and all the time you have to keep your fingers crossed nothing crashes. This is pretty heavy and I would even dare to say way beyond anything feasible. Most shots of this kind are no longer than 5 minutes and even those take days to get a halfway usable match.
[karim] “Second… when I track and get the camera-data (Maja I guess)I sure can import it in AE, but how do I place a layer with canals and buildings in my comp with the clip+camera movement so they fit in the picture.”
Most likely you’d have to place your stuff on 3D-layers in a sub-comp and than rotoscope/ mask that sub-comp in the main comp to get out the overlap of buildings, trees and so on. Yet even more work.
The whole project seems a bit out of league not only in terms of what AE can do but also how much work one person or a small team can do. You should look into other options. Personally I’d probably only stabilize the footage and then use it as front projections on simplified stand-in geometry in a 3D program. I wouldn’t aim at ultimate perfect matchmoving but rather create some nice camera moves that capture the mood and try to avoid “swimming” of the overlays as much as possible.
Mylenium
[Pour Myl