-
Faking 3D with 2D tracking data
Alright, I have a question. I’ve looked around quite a bit, so if the answer’s obvious, be gentle! Here’s what I’m needing to do:
I’ve recorded a subject on a greenscreen with tracking markers. I’ve used AE’s tracker to get tracking data. I used Mocha, too, but it seems that for this, I’m not needing a planar tracker. But if Mocha’s the tool to use for this, someone let me know. Since I’m wanting to parent to a Null, I’m just using X, Y, and Scale.
What I’d like to do is place my subject in a new 3D environment. And I get that my tracking software is 2D tracking, but it seems to me that using the Scale data, one could use that in place of Z depth for a 3D camera’s position in After Effects. Could you fake it this way? (The camera doesn’t zoom, but is hand-held, and has some in and out motion.) That way my original footage would stay 2D, but I could add a 3D camera and some 3D layers that would follow the movements of my real-world camera.
Or, here’s another approach. We’ll forget about moving a 3D camera and move layers instead. Currently using just my greenscreen replacement and tracking data, I can easily insert a new BG and make it look integrated into the hand-held footage. But if I insert multiple layers, they’re all moving the same way and at the same rate, and looking like they’re the same distance away from my subject as the greenscreen was, which is about 15 feet. If it were real-world 3D, the layer furthest back wouldn’t move as quickly as layers closer to the camera, right? So I’m thinking (and I suck at math, so Expressions folks, help me out here) I could use an expression to link the Z position of a layer to the Scale data from the tracker (probably copied onto its own Null). The scale value change is too pronounced, but like I said, I suck at math, and it seems like there would be a way to take the scale values and make them more compressed for layers that I want to appear further back. I would want to see some movement, but not as pronounced. Or maybe forget the Z-position, keep it all in 2D, but just use an expression on position and scale values to change all the tracking data to a lesser degree for layers I want to seem further away.
Does any of this make sense? Am I making things too hard? I’m just trying to see if there’s a way to fake 3D tracking from 2D data if scale is assumed to accurately represent distance, and can be extrapolated as such for different layers at different depths.
-
6 Replies
-
What are you actually trying to accomplish?
Is there a reason your green screen footage was handheld rather than on a tripod?
Could you just stablilize your green screen footage then create all the camera moves in AE?
What kind of camera moves are we talking? Is it gentle handheld rocking/shifting or are we talking swooping all over the place?I would also like to point out some advice from Dave LaRonde:
Dave’s Stock Answer #2:
When you’re out on a shoot, and you say, “we’ll fix this in post” without knowing PRECISELY HOW you’re going to fix it in post, don’t shoot it!
You’ll only end up shooting it over again.
Since post typically costs three times the cost of production, fixing something in post is not a way to save money, but rather a way to spend
more of it.
And, before you say “well fix it in post,” always consider who’s doing the work, especially if you’re the one doing the editing.
– The Great Szalam
(The ‘Great’ stands for ‘Not So Great, in fact, Extremely Humble’)No trees were harmed in the creation of this message, but several thousand electrons were mildly inconvenienced.
-
Excellent questions. First, the reason I’m shooting hand-held is because I want to shoot hand-held. 🙂 Seriously. That’s the look I need. So of course it would be easier with a tripod. And I could absolutely stabilize the shot and then create the camera moves in AE, but that’s never going to quite have the hand-held look that I want for this shot. Think Cloverfield, though not nearly as frenetic or complex. Just a little bobbing and weaving, and stepping towards and away from the talent. I need someone outdoors in a field, and as they’re speaking, a circus tent rises behind them at a fairly far off distance, and then I need text floating in space between them and the circus tent. I appreciate you sharing Dave’s Stock Answer #2. The good news is, except for a minute of test footage I’ve been experimenting with, nothing’s been shot yet. That’s why I’m writing this. So I’ll know precisely how to do this in post. I’m in pre-production now, not production. And if it’s not possible, I’ll move on to another option.
So, if anyone knows if there’s a solution, that’d be great. Like I said, I think if one assumes a camera isn’t zooming but is physically moving towards and away from an object, couldn’t scale data be used to extrapolate Z position? (Wow. That’s my whole question in a much more concise format than my original post!) Thanks for the thoughts, Michael – I appreciate you taking the time to respond.
-
Wow. I’m so happy! Someone’s actually planning ahead and shooting test shots!
There probably is a way to extrapolate some z-axis data, but I don’t know the math right now and I’m in a hair salon waiting room right now typing this on my iPhone so I can’t really play with anything.
If you want a hand held feel, I still think it would be better to shoot it on a tripod and do the camera move in AE. However, you could shoot some handheld of just some tracking markers and try your z-axis translation and apply it to a null then parent your AE camera to the null. That way you will get your true handheld look without any problems of a mismatched move.– The Great Szalam
(The ‘Great’ stands for ‘Not So Great, in fact, Extremely Humble’)No trees were harmed in the creation of this message, but several thousand electrons were mildly inconvenienced.
-
Yeah, I’m sure Syntheyes, PFHoe or any of those programs could get me where I want to go. (Boujou would be just a touch outside my budget, though!) I’m trying to save $400 by using what I have now. If it proves impossible, then maybe we go the 3D tracker route, or re-think the necessity of such a shot. I’m just a little surprised it hasn’t come up before, but I’ve searched and searched and can’t find anyone trying to do this. But it seems entirely plausible that this could be done, especially by someone with some good scripting or expressions skills. It’s just one of those things that I think, “You know, there should be a way to do this…”
And yes, I could just key out the talent and build a 3D camera rig and do all the handheld movement in post, but then I’m still dealing with a flat 2D layer, and if I’m moving in tighter on a face and need to move the camera down and look up someone’s nose, it just doesn’t look right with Flat Stanley. Not that I’m going to be doing that exactly, but my point is, doing the camera motion in post makes some things easier, but pushed to its limits reveals… well, its limits. But it’s a good suggestion, and probably the route I’ll end up taking if needed. The longer I look into this, the more academic the question becomes for me. I’d like the ability to do what I’m asking about, but it’s not going to make or break the project. But it would sure be nice.
-
Were you able to figure this out? I’m trying to do the same thing.
-
You know, I can’t even remember the project I was doing. My guess is, I never did quite master this, because I know I never completed anything like I described. I’m not sure what I did instead, but it wasn’t this. So, short answer, no. But I do a lot less with After Effects these days, and I imagine the capabilities have increased, so it may be much easier. Sorry I can’t be of more help!
Reply to this Discussion! Login or Sign Up