-
All-day Depth Completion
Authors:
Vadim Ezhov,
Hyoungseob Park,
Zhaoyang Zhang,
Rishi Upadhyay,
Howard Zhang,
Chethan Chinder Chandrappa,
Achuta Kadambi,
Yunhao Ba,
Julie Dorsey,
Alex Wong
Abstract:
We propose a method for depth estimation under different illumination conditions, i.e., day and night time. As photometry is uninformative in regions under low-illumination, we tackle the problem through a multi-sensor fusion approach, where we take as input an additional synchronized sparse point cloud (i.e., from a LiDAR) projected onto the image plane as a sparse depth map, along with a camera…
▽ More
We propose a method for depth estimation under different illumination conditions, i.e., day and night time. As photometry is uninformative in regions under low-illumination, we tackle the problem through a multi-sensor fusion approach, where we take as input an additional synchronized sparse point cloud (i.e., from a LiDAR) projected onto the image plane as a sparse depth map, along with a camera image. The crux of our method lies in the use of the abundantly available synthetic data to first approximate the 3D scene structure by learning a map** from sparse to (coarse) dense depth maps along with their predictive uncertainty - we term this, SpaDe. In poorly illuminated regions where photometric intensities do not afford the inference of local shape, the coarse approximation of scene depth serves as a prior; the uncertainty map is then used with the image to guide refinement through an uncertainty-driven residual learning (URL) scheme. The resulting depth completion network leverages complementary strengths from both modalities - depth is sparse but insensitive to illumination and in metric scale, and image is dense but sensitive with scale ambiguity. SpaDe can be used in a plug-and-play fashion, which allows for 25% improvement when augmented onto existing methods to preprocess sparse depth. We demonstrate URL on the nuScenes dataset where we improve over all baselines by an average 11.65% in all-day scenarios, 11.23% when tested specifically for daytime, and 13.12% for nighttime scenes.
△ Less
Submitted 27 May, 2024;
originally announced May 2024.
-
An Inverse Procedural Modeling Pipeline for SVBRDF Maps
Authors:
Yiwei Hu,
Chengan He,
Valentin Deschaintre,
Julie Dorsey,
Holly Rushmeier
Abstract:
Procedural modeling is now the de facto standard of material modeling in industry. Procedural models can be edited and are easily extended, unlike pixel-based representations of captured materials. In this paper, we present a semi-automatic pipeline for general material proceduralization. Given Spatially-Varying Bidirectional Reflectance Distribution Functions (SVBRDFs) represented as sets of pixe…
▽ More
Procedural modeling is now the de facto standard of material modeling in industry. Procedural models can be edited and are easily extended, unlike pixel-based representations of captured materials. In this paper, we present a semi-automatic pipeline for general material proceduralization. Given Spatially-Varying Bidirectional Reflectance Distribution Functions (SVBRDFs) represented as sets of pixel maps, our pipeline decomposes them into a tree of sub-materials whose spatial distributions are encoded by their associated mask maps. This semi-automatic decomposition of material maps progresses hierarchically, driven by our new spectrum-aware material matting and instance-based decomposition methods. Each decomposed sub-material is proceduralized by a novel multi-layer noise model to capture local variations at different scales. Spatial distributions of these sub-materials are modeled either by a by-example inverse synthesis method recovering Point Process Texture Basis Functions (PPTBF) or via random sampling. To reconstruct procedural material maps, we propose a differentiable rendering-based optimization that recomposes all generated procedures together to maximize the similarity between our procedural models and the input material pixel maps. We evaluate our pipeline on a variety of synthetic and real materials. We demonstrate our method's capacity to process a wide range of material types, eliminating the need for artist designed material graphs required in previous work. As fully procedural models, our results expand to arbitrary resolution and enable high level user control of appearance.
△ Less
Submitted 27 September, 2021; v1 submitted 13 September, 2021;
originally announced September 2021.
-
Countering Racial Bias in Computer Graphics Research
Authors:
Theodore Kim,
Holly Rushmeier,
Julie Dorsey,
Derek Nowrouzezahrai,
Raqi Syed,
Wojciech Jarosz,
A. M. Darke
Abstract:
Current computer graphics research practices contain racial biases that have resulted in investigations into "skin" and "hair" that focus on the hegemonic visual features of Europeans and East Asians. To broaden our research horizons to encompass all of humanity, we propose a variety of improvements to quantitative measures and qualitative practices, and pose novel, open research problems.
Current computer graphics research practices contain racial biases that have resulted in investigations into "skin" and "hair" that focus on the hegemonic visual features of Europeans and East Asians. To broaden our research horizons to encompass all of humanity, we propose a variety of improvements to quantitative measures and qualitative practices, and pose novel, open research problems.
△ Less
Submitted 2 June, 2022; v1 submitted 28 March, 2021;
originally announced March 2021.
-
Bidirectional Self-Folding with Atomic Layer Deposition Nanofilms for Microscale Origami
Authors:
Baris Bircan,
Marc Z. Miskin,
Robert J. Lang,
Michael C. Cao,
Kyle J. Dorsey,
Muhammad G. Salim,
Wei Wang,
David A. Muller,
Paul L. McEuen,
Itai Cohen
Abstract:
Origami design principles are scale invariant and enable direct miniaturization of origami structures provided the sheets used for folding have equal thickness to length ratios. Recently, seminal steps have been taken to fabricate microscale origami using unidirectionally actuated sheets with nanoscale thickness. Here, we extend the full power of origami-inspired fabrication to nanoscale sheets by…
▽ More
Origami design principles are scale invariant and enable direct miniaturization of origami structures provided the sheets used for folding have equal thickness to length ratios. Recently, seminal steps have been taken to fabricate microscale origami using unidirectionally actuated sheets with nanoscale thickness. Here, we extend the full power of origami-inspired fabrication to nanoscale sheets by engineering bidirectional folding with 4 nm thick atomic layer deposition (ALD) SiNx-SiO2 bilayer films. Strain differentials within these bilayers result in bending, producing microscopic radii of curvature. We lithographically pattern these bilayers and localize the bending using rigid panels to fabricate a variety of complex micro-origami devices. Upon release, these devices self-fold according to prescribed patterns. Our approach combines planar semiconductor microfabrication methods with computerized origami design, making it easy to fabricate and deploy such microstructures en masse. These devices represent an important step forward in the fabrication and assembly of deployable micromechanical systems that can interact with and manipulate micro- and nanoscale environments.
△ Less
Submitted 18 June, 2020;
originally announced June 2020.
-
AniCode: Authoring Coded Artifacts for Network-Free Personalized Animations
Authors:
Zeyu Wang,
Shiyu Qiu,
Qingyang Chen,
Alexander Ringlein,
Julie Dorsey,
Holly Rushmeier
Abstract:
Time-based media (videos, synthetic animations, and virtual reality experiences) are used for communication, in applications such as manufacturers explaining the operation of a new appliance to consumers and scientists illustrating the basis of a new conclusion. However, authoring time-based media that are effective and personalized for the viewer remains a challenge. We introduce AniCode, a novel…
▽ More
Time-based media (videos, synthetic animations, and virtual reality experiences) are used for communication, in applications such as manufacturers explaining the operation of a new appliance to consumers and scientists illustrating the basis of a new conclusion. However, authoring time-based media that are effective and personalized for the viewer remains a challenge. We introduce AniCode, a novel framework for authoring and consuming time-based media. An author encodes a video animation in a printed code, and affixes the code to an object. A consumer uses a mobile application to capture an image of the object and code, and to generate a video presentation on the fly. Importantly, AniCode presents the video personalized in the consumer's visual context. Our system is designed to be low cost and easy to use. By not requiring an internet connection, and through animations that decode correctly only in the intended context, AniCode enhances privacy of communication using time-based media. Animation schemes in the system include a series of 2D and 3D geometric transformations, color transformation, and annotation. We demonstrate the AniCode framework with sample applications from a wide range of domains, including product "how to" examples, cultural heritage, education, creative art, and design. We evaluate the ease of use and effectiveness of our system with a user study.
△ Less
Submitted 30 July, 2018;
originally announced July 2018.
-
Ergonomic-driven Geometric Exploration and Resha**
Authors:
Youyi Zheng,
Julie Dorsey,
Niloy Mitra
Abstract:
The paper addresses the following problem: given a set of man-made shapes, e.g., chairs, can we quickly rank and explore the set of shapes with respect to a given avatar pose? Answering this question requires identifying which shapes are more suitable for the defined avatar and pose; and moreover, to provide fast preview of how to alter the input geometry to better fit the deformed shapes to the g…
▽ More
The paper addresses the following problem: given a set of man-made shapes, e.g., chairs, can we quickly rank and explore the set of shapes with respect to a given avatar pose? Answering this question requires identifying which shapes are more suitable for the defined avatar and pose; and moreover, to provide fast preview of how to alter the input geometry to better fit the deformed shapes to the given avatar pose? The problem naturally links physical proportions of human body and its interaction with object shapes in an attempt to connect ergonomics with shape geometry. We designed an interaction system that allows users to explore shape collections using the deformation of human characters while at the same time providing interactive previews of how to alter the shapes to better fit the user-specified character. We achieve this by first map** ergonomics guidelines into a set of simultaneous multi-part constraints based on target contacts; and then, proposing a novel contact-based deformation model to realize multi-contact constraints. We evaluate our framework on various chair models and validate the results via a small user study.
△ Less
Submitted 21 February, 2014;
originally announced February 2014.
-
A note on completeness in the theory of strongly clean rings
Authors:
Alexander J. Diesl,
Thomas J. Dorsey
Abstract:
Many authors have investigated the behavior of strong cleanness under certain ring extensions. In this note, we prove that if $R$ is a ring which is complete with respect to an ideal $I$ and if $x$ is an element of $R$ whose image in $R/I$ is strongly $π$-regular, then $x$ is strongly clean in $R$.
Many authors have investigated the behavior of strong cleanness under certain ring extensions. In this note, we prove that if $R$ is a ring which is complete with respect to an ideal $I$ and if $x$ is an element of $R$ whose image in $R/I$ is strongly $π$-regular, then $x$ is strongly clean in $R$.
△ Less
Submitted 13 July, 2009;
originally announced July 2009.
-
A Characterization of Certain Morphic Trivial Extensions
Authors:
Alexander J. Diesl,
Thomas J. Dorsey,
Warren Wm. McGovern
Abstract:
Given a ring $R$, we study the bimodules $M$ for which the trivial extension $R\propto M$ is morphic. We obtain a complete characterization in the case where $R$ is left perfect, and we prove that $R\propto Q/R$ is morphic when $R$ is a commutative reduced ring with classical ring of quotients $Q$. We also extend some known results concerning the connection between morphic rings and unit regular…
▽ More
Given a ring $R$, we study the bimodules $M$ for which the trivial extension $R\propto M$ is morphic. We obtain a complete characterization in the case where $R$ is left perfect, and we prove that $R\propto Q/R$ is morphic when $R$ is a commutative reduced ring with classical ring of quotients $Q$. We also extend some known results concerning the connection between morphic rings and unit regular rings.
△ Less
Submitted 7 July, 2009;
originally announced July 2009.
-
On minimal extensions of rings
Authors:
Thomas J. Dorsey,
Zachary Mesyan
Abstract:
Given two rings $R \subseteq S$, $S$ is said to be a minimal ring extension of $R$ if $R$ is a maximal subring of $S$. In this article, we study minimal extensions of an arbitrary ring $R$, with particular focus on those possessing nonzero ideals that intersect $R$ trivially. We will also classify the minimal ring extensions of prime rings, generalizing results of Dobbs, Dobbs & Shapiro, and Fer…
▽ More
Given two rings $R \subseteq S$, $S$ is said to be a minimal ring extension of $R$ if $R$ is a maximal subring of $S$. In this article, we study minimal extensions of an arbitrary ring $R$, with particular focus on those possessing nonzero ideals that intersect $R$ trivially. We will also classify the minimal ring extensions of prime rings, generalizing results of Dobbs, Dobbs & Shapiro, and Ferrand & Olivier on commutative minimal extensions.
△ Less
Submitted 21 May, 2008;
originally announced May 2008.
-
Morphic and principal-ideal group rings
Authors:
Thomas J. Dorsey
Abstract:
We observe that the class of left and right artinian left and right morphic rings agrees with the class of artinian principal ideal rings. For $R$ an artinian principal ideal ring and $G$ a group, we characterize when $RG$ is a principal ideal ring; for finite groups $G$, this characterizes when $RG$ is a left and right morphic ring. This extends work of Passman, Sehgal and Fisher in the case wh…
▽ More
We observe that the class of left and right artinian left and right morphic rings agrees with the class of artinian principal ideal rings. For $R$ an artinian principal ideal ring and $G$ a group, we characterize when $RG$ is a principal ideal ring; for finite groups $G$, this characterizes when $RG$ is a left and right morphic ring. This extends work of Passman, Sehgal and Fisher in the case when $R$ is a field, and work of Chen, Li, and Zhou on morphic group rings.
△ Less
Submitted 1 December, 2006;
originally announced December 2006.
