Bridge Comp Test 1 from Dane Kirkland on Vimeo.
Bridge Comp Test 2 from Dane Kirkland on Vimeo.
AE Motion tracking WIP.
Exploring motion tracking in After Effects. The base unaltered track that it gave is the top video. As far as a stable track it was rubbish, but that's to be expected when you haven't tweaked the settings what-so-ever. Mind you, it kept it in the general area of the tracking point reasonably ok-ish so that was enough for me to work from. I linked the position of the comped character through the use of an expression to link it with the tracking data. That way when I changed the track layer the character's position would be automatically updated. Sure I could have copy and pasted the relevant data but this was just far more efficient.
The second video is a more refined track, but as you can see it still slips here and there, particularly with the larger camera movements.
The final video is that same track again, tweaked further - with touch up's done by hand. It was a bit painstaking to do it this way but still worked out pretty well. It's still not perfect, I'm going to continue to play with it to see if I can get it perfect - but this isn't too bad as far as it goes I don't think.
I have done a few other tests which I will upload as well, but this was the best one for this particular example. So after this little exercise I've realised that the challenge of tracking into unstable footage for the length of time planned and the complexity of what I will be doing would be... well suicide. I'm pretty sure it would be doable, but would not look as good as it would if it were stable footage. There will still be some handheld footage in the film, but the rest will be shot on a tripod and manipulated into appear as though it is handheld. I've thought of a few different ideas on how to approach this (keying, plugins etc) I think the route I will explore further is using handheld footage as a direct base for simluating the movement. Using the same technique that I did in the examples above and applying the tracked movement to a stable piece of footage. I'm yet to shoot any stable test footage but I've quickly mocked up an example using the movement of the footage above on a still image. I think as long as the footage which is to be used is of a higher res than the final export it should work. Tests will confirm this of course and it may mean double shooting everything that I want as handheld, but that's alot less time-consuming than the alternative. If not, I'll use a mixture of the techniques. (Video eg follows, ran out of my daily vimeo quota =p)
No comments:
Post a Comment