Figure 1: Equal-time comparison of two-bounce path tracing with our approach. Images are rendered at 1080p resolution with an NVIDIA 3090 RTX GPU without denoising. (Left) Path tracing with one sample per pixel in 8.0 ms. (Middle) ReSTIR GI using spatial and temporal resampling and one sample per pixel in 8.9 ms. Mean squared error is improved by a factor of 15.1. (Right) Path traced reference image.This is a challenging scene for path tracing, as direct lighting is concentrated in small regions, making it difficult to find indirect lighting paths. ReSTIR GI is much more effective thanks to sample reuse in both space and time.

Figure 1: Composing parts, possibly with sharp features and non-overlapping boundaries, presents challenges to both part alignment and blending. Our field-guided approach (see middle for a visualization of the fields) leads to alignment of parts away from each other and feature-conforming surface blending. The bridging surfaces generated (colored yellow on the right) are piecewise smooth. AbstractWe present an automatic shape composition method to fuse two shape parts which may not overlap and possibly contain sharp features, a scenario often encountered when modeling man-made objects. At the core of our method is a novel field-guided approach to automatically align two input parts in a feature-conforming manner.The key to our field-guided shape registration is a natural continuation of one part into the ambient field as a means to introduce an overlap with the distant part, which then allows a surface-tofield registration. The ambient vector field we compute is featureconforming; it characterizes a piecewise smooth field which respects and naturally extrapolates the surface features. Once the two parts are aligned, gap filling is carried out by spline interpolation between matching feature curves followed by piecewise smooth least-squares surface reconstruction. We apply our algorithm to obtain feature-conforming shape composition on a variety of models and demonstrate generality of the method with results on parts with or without overlap and with or without salient features.

No abstract

No abstract

No abstract