multiplane technique - almost there, need some debug help.

I’m working on an multiplane effect and I’ve got it working almost perfectly except some strange movement in the layers when the camera comes too close.

In this example you see that the red and blue dots goes out in wrong directions after halfway through. They should go in rougly the same path as the green dot.
Also their speed is accelerating too fast at the end.

Can anyone spot what is wrong with my calculations?

The whole thing works with a main layer which acts a a camera. The planes below moves in relation to this layers translation and tracking. All layers is sorted in space based on their Z-value. Camera starts at Z 0 and the layers is placed at different heights below the camera using z-values like 0.2, 0.5 etc. z-value of 1 is quite far away. Far limit is about 1.7 after which no movement of the layer is visible.

Zoom and blur of layers is working as they should. zoom is just based on distance from “camera” while you got some options to set for the blur: range of where you got full focus and amount of falloff for the blur - how far/fast from this range the layers goes blurry.

The issue with movement is somewhere in the translation layers of the planes. Here I need help to debug.

I think this info should be enough to set you going, ask if you need anything else.

Here’s the synfig-file:
rylanderanimation.se/temp/mu … ample.sifz

Thank you,
-David

Hi rylleman,
although it would be more work for you but can you please write down the expanded formulations for a general plane in the standard math language? If all the planes have the same synfig’s math combination all they should react the same if the original math formulation is fine.
Just looking to the way of perform the operations (zoom first, translate after and then blur) I think that the zoom must be calculated with the center of the zoom at the center of the subject, if not, you produce a translation too if the zoom origin is not centered. So, something is missing in the center of the zoom layer, I think.
-G

Hi rylleman,

I do not think it’s a bug; I see this as a natural acceleration by the proximity with the camera. Closer to the camera, faster speed. We have too movements, the objects motion and the camera motion getting closer to the objects, so when the camera get’s closer…the object seems to accelerate. Am I wrong? Think about relativity! Isn’t that?

I do not see any mistake, nice scene, congratulations.

You might be right. I’ve been testing different motions and it feels like the motion goes too fast closer to the camera but it might be right there. Perhaps the exponential movement-amount curve goes a bit too steep.
Feel like I have to tweak a little.
But it’s working now if I disregard the feeling that it’s not quite right movement close to camera. I’m doing a tutorial for the technique right now.