colorcurvature opened this issue on Apr 27, 2010 · 117 posts
colorcurvature posted Wed, 05 May 2010 at 11:21 AM
I think it is more than just the magnets.
I have written a new toolset just to debug the effect on morphs in the world coordinate space. I found that actually the layout of the whole virtual space is changed in a non linear fashion. Both the dialing itself as well as the axes of the (hyper)space appear to be unstable (from a linear point of view). This is a fundamental issue and might be not the fault of poor v4. I hunted the issues down to m4 as well.
Poser seems to calculate in a way that I am not able to find out. It might be sums of rounding errors or be the cause of the core of their algorithms.
However I have a small hope. By repeatedly applying my algorithm and tracking the movement of the vertices in poser I seem to be able to reduce the error from about 10^-2 down to about 10^-6. At least for the worst example it worked.
I morphed M4 posed with AdamTwaithes Pose 72 into a V4 with Adam's Pose 84 and the error was estimated at only 10^-7.
If this would work in general than it would get as close to mathematical perfection as possible :).
However I have to redesign the script from scratch now, and noone can say if it works good on any other poser than mine. @Testers: You will have mail soon ;)