The goal of my initial tests was to confirm whether or not i could manipulate geometry based on camera movement alone. The first setup was a combination of blend shapes, set driven keys, and mel scripts. I used a locator (null object) to grab data from a selected camera's location, and then fed that data into a custom attribute. I then created a small library of blend shapes to swap between based on the number in the attribute. The result was an indentation in a cup that followed the camera as you rotated around it.
- In theory the test was successful, however I will need to do some tweaking to the procedure in order to iron out some issues I ran into. For example, the low resolution of the cup geometry prevents the desired smooth transition between blend shapes, resulting in the dip appearing to bounce up and down as it interpolates between shapes.
Another test I attempted was parent constraining the facing side of a cube to a camera's position using clusters. The result is what appears to be the lack of movement in the facing side, and a deformation of the rest of the cube. I'm thinking that a level of offset while moving the camera might serve me better, and look a bit more interesting.
I'll be using a combination of both techniques to achieve a look I want on a per-shot basis. More tests to follow.
No comments:
Post a Comment