So this week in VFX was interesting. Peter announced we’d be looking at camera tracking and then ran through some of the technical details of tracking. However all the way through until the last part of the exercise I had no idea what the end result was going to be, in my head I just couldn’t assemble what we were setting up. Needless to say I was impressed. So those technical details I mentioned, there are five ways you can track footage:
- 1-point, uses an X and Y to track position (okay still sticking with X, Y, Z, I can figure this out)
- 2-point, as above but with Z rotation and scale.
- 4-point (Peter mentioned theres no such thing as 3 point) uses a 4 corner pinning process to track a rectangle through the scene/footage. To make a comparison I guess this is like perspective warp in Photoshop, using 4 points to track a perspective. Same theory.
- 3D Camera tracking, basically computer wizardry that tries to track the camera throughout the entire shot with no input from the artist.
- Planar tracking, takes a region of the image (the example was the side of a train carriage) and attempts to follow it through the scene.
Peter explained there will be times when a computer will just refuse to track your footage effectively and you will be required to track the manual way using methods other than 3D Camera. However for now (and simplicity) he chose to demonstrate the automated 3D option to us.
Usual composition setup, 1080p, 25fps but eight seconds long this time. The first step was to make a text object the size of the canvas, put the font size up to 300pt and write our name. Next we dived into composition settings and set the renderer to raytraced 3D. We were instructed the final product would not render out correctly unless we remembered this step. The worst part is that feels like an easy thing to forget, hence I’m writing it here to remember!
The 3D box for the text layer needed to be checked and this opens up geometry options in the drop down for the layer. Using these new options we cranked the value up on extrusion depth, giving the text a Z value, pushing it out into 3D space. Sadly I don’t have any screen captures of this stage, I only remembered to start documenting the process once we began tracking. My bad! Lastly a light was added to give the text some shading rather than flat colour, see my previous posts talking about lights in AE.
Now we imported the camera footage into AE, dragged it onto the timeline and jumped to Effects > Perspective > 3D Camera Tracker, adding new options to the layers drop down. The footage begins to process immediately, however before the process ran its full course we dived into the settings and enabled detailed analysis. Peter suggesting there is no reason to ever not use this option for better tracking.
Once the detailed analysis had finished the scene looked like this.
For someone with no clue whats happening, I’m sitting there thinking okay, thats a nice field of flowers! Every ‘flower’ is actually a tracking point the software is using to figure out how the camera is moving. Next we would choose an area of points to designate the ground, this was a simple matter of shift clicking between 3 points on the ground. It then looks like this.
Using these points you designate a bullseye the software uses to determine where the ground is during tracking. For this to take effect you need to right click the bullseye and select ‘Set Ground Plane’. All that was left to do was to turn the text layer back on, and I WAS NOT expecting what happened next.
Yeah, THAT! For how little work I’d actually done to get that point I was very impressed with the result. I thought I was impressed by the particle systems last week but this totally blew that out of the water. We’ll be continuing tracking next week, we didn’t get into using shadow catchers to get the text to project a shadow even, so plenty to cover next week.
Check out my other VFX post for a progress update on the logo project.