Right now the point looks at the alpha channel percentage of the specified pixel. If the alpha of the animation is too solid, or empty, the data being outputted will plateau and become static. So by duplicating the anim while scaling and blurring it, there's a better chance of the point hitting always changing pixel. This is also why I used the line layer, there's less of a chance of there being a solid constant alpha. The downside to this is also the negative amount of alpha, also causing it to be static.
ANIMATION PROVIDED BY TOM LAW (Thanks, Tom)
The duplicated and blurred anim analysis layer.
The green square was parented to the texture layer the line anim is being matted to. Its movement are correlated to the strength of the alpha in the blurred line analysis layer (above), that pass under the white square (analysis point).
The problems that still linger are that the "point" method is unreliable in all types of anim. If this was a walking panning scene, the point would constantly have to be moved and keyframed to make sure there was anim underneath it. That defeats the purpose of going key-less. I think theres a way to extract histogram data from the entire frame, so anytime a frame changes a small amount, the data could be used without any points or alphas. The anim could be whole animation, not just a blurry line layer, which would make the ease of applying and setup entirely effortless.
CHECK BACK FOR UPDATES!
