4 Conclusion
The results of our perceptual experiment have implications for both
motion compression and data-driven parametrization models. They
suggest that, in motion compression, the underlying databases do
not have to contain densely sampled low speed locomotions, as they
can be reconstructed from the higher speeds without producing perceptual
artifacts. Similarly, the indications are that motion synthesis
methods should avoid time warping a source motion to create a
faster movement. Rather, interpolation between slowed down locomotion
clips would be preferable to the traditional combination
of speeded up and slowed down animations. However, our conclusions
are directly applicable to locomotion only and further studies
are now needed to confirm and generalize our results
4 ConclusionThe results of our perceptual experiment have implications for bothmotion compression and data-driven parametrization models. Theysuggest that, in motion compression, the underlying databases donot have to contain densely sampled low speed locomotions, as theycan be reconstructed from the higher speeds without producing perceptualartifacts. Similarly, the indications are that motion synthesismethods should avoid time warping a source motion to create afaster movement. Rather, interpolation between slowed down locomotionclips would be preferable to the traditional combinationof speeded up and slowed down animations. However, our conclusionsare directly applicable to locomotion only and further studiesare now needed to confirm and generalize our results
การแปล กรุณารอสักครู่..
