Handle-driven image warping based on linear blending is widely used in many applications because of its merits on intuitiveness, efficiency, and ease of implementation. In this paper, we develop a method to compute high-quality weights within a closed domain for image warping. The property of C₂ continuity in weights is guaranteed by the carefully formulated basis functions. The efficiency of our algorithm is ensured by a closed-form formulation of the computation for weights. The cost of inserting a new handle is only the time to evaluate the distances from the new handle to all other sample points in the domain. A virtual handle insertion algorithm is developed to allow users to freely place handles within the domain while preserving the satisfaction of all expected criteria on weights for linear blending. Experimental examples for real-time applications are shown to demonstrate the effectiveness of this method.

@en

Existing techniques for motion imitation often suffer a certain level of latency due to their computational overhead or a large set of correspondence samples to search. To achieve real-time imitation with small latency, we present a framework in this paper to reconstruct motion on humanoids based on sparsely sampled correspondence. The imitation problem is formulated as finding the projection of a point from the configuration space of a human's poses into the configuration space of a humanoid. An optimal projection is defined as the one that minimizes a back-projected deviation among a group of candidates, which can be determined in a very efficient way. Benefited from this formulation, effective projections can be obtained by using sparsely sampled correspondence, whose generation scheme is also introduced in this paper. Our method is evaluated by applying the human's motion captured by an RGB-depth (RGB-D) sensor to a humanoid in real time. Continuous motion can be realized and used in the example application of teleoperation.

@en

Handle driven character skinning is widely favored in animation applications due to its advantages of intuitiveness, effectiveness and simplicity. A research thread to realize this is to compute a proper weighting distribution on points associated with specified handles like points, bars or skeleton, which is critical to the quality of manipulation and can be utilized to produce animation by controlling user handles. In this work, we introduce a new skinning method that is compatible with different model representations, with a key idea of evaluating the local influence of each handle by decomposing the shape domain into small overlapped regions. Thanks to its well-designed formulation, the computation of weights and update of models can be conducted in par-allel on GPU, leading to high efficiency and good visual quality supported by the provided experimental and statistical results in this paper.@en