Isn’t it completely insane we get this technology available to individuals and indie, wild.
Loved the stream, very informative, very clear and educational!
Haven’t checked the code yet, but I wonder if you tried machine learning instead of your distance based computation and interrupt system to find the optimal frame/database to transition to ? Maybe to find some subtle correlations or to filter out noisy input ?
Wondering if MotionBuilder is “still” (to my knowledge) the best solution for retargeting out there, used to play with it with educational licenses but it’s 2K$ licensing per year <.<…
I did a true first/third person replicated distance matching test 4 years ago inspired from Paragon.
Had to indeed edit the charactermovement component code back then to add acceleration replication, bake the distance to the various runtime generated projected data points, create a custom animation node and choose the right animation frame etc…
It was an interesting project but required way too much finetuning of tedious parameters, wasn’t very practical, looking forward to test the motion matching plugin and its optimal frame selection system, think it should be much more flexible !
