Search for: All records

The gradual nature of a diffusion process that synthesizes samples in small increments constitutes a key ingredient of Denoising Diffusion Probabilistic Models (DDPM), which have presented unprecedented quality in image synthesis and been recently explored in the motion domain. In this work, we propose to adapt the gradual diffusion concept (operating along a diffusion time-axis) into the temporal-axis of the motion sequence. Our key idea is to extend the DDPM framework to support temporally varying denoising, thereby entangling the two axes. Using our special formulation, we iteratively denoise a motion buffer that contains a set of increasingly-noised poses, which auto-regressively produces an arbitrarily long stream of frames. With a stationary diffusion time-axis, in each diffusion step we increment only the temporal-axis of the motion such that the framework produces a new, clean frame which is removed from the beginning of the buffer, followed by a newly drawn noise vector that is appended to it. This new mechanism paves the way towards a new framework for long-term motion synthesis with applications to character animation and other domains. 

We introduce HyperFields, a method for generating text-conditioned Neural Radiance Fields (NeRFs) with a single forward pass and (optionally) some fine-tuning. Key to our approach are: (i) a dynamic hypernetwork, which learns a smooth mapping from text token embeddings to the space of NeRFs; (ii) NeRF distillation training, which distills scenes encoded in individual NeRFs into one dynamic hypernetwork. These techniques enable a single network to fit over a hundred unique scenes. We further demonstrate that HyperFields learns a more general map between text and NeRFs, and consequently is capable of predicting novel in-distribution and out-of-distribution scenes--either zero-shot or with a few finetuning steps. Finetuning HyperFields benefits from accelerated convergence thanks to the learned general map, and is capable of synthesizing novel scenes 5 to 10 times faster than existing neural optimization-based methods. Our ablation experiments show that both the dynamic architecture and NeRF distillation are critical to the expressivity of HyperFields. 

We present a neural technique for learning to select a local sub-region around a point which can be used for mesh parameterization. The motivation for our framework is driven by interactive workflows used for decaling, texturing, or painting on surfaces. Our key idea is to incorporate segmentation probabilities as weights of a classical parameterization method, implemented as a novel differentiable parameterization layer within a neural network framework. We train a segmentation network to select 3D regions that are parameterized into 2D and penalized by the resulting distortion, giving rise to segmentations which are distortion-aware. Following training, a user can use our system to interactively select a point on the mesh and obtain a large, meaningful region around the selection which induces a low-distortion parameterization. 

We report the design conception, chemical synthesis, and microbiological evaluation of the bridged macrobicyclic antibiotic cresomycin (CRM), which overcomes evolutionarily diverse forms of antimicrobial resistance that render modern antibiotics ineffective. CRM exhibits in vitro and in vivo efficacy against both Gram-positive and Gram-negative bacteria, including multidrug-resistant strains ofStaphylococcus aureus,Escherichia coli, andPseudomonas aeruginosa. We show that CRM is highly preorganized for ribosomal binding by determining its density functional theory–calculated, solution-state, solid-state, and (wild-type) ribosome-bound structures, which all align identically within the macrobicyclic subunits. Lastly, we report two additional x-ray crystal structures of CRM in complex with bacterial ribosomes separately modified by the ribosomal RNA methylases, chloramphenicol-florfenicol resistance (Cfr) and erythromycin-resistance ribosomal RNA methylase (Erm), revealing concessive adjustments by the target and antibiotic that permit CRM to maintain binding where other antibiotics fail. 

Abstract In contemporary organic synthesis, substances that access strongly oxidizing and/or reducing states upon irradiation have been exploited to facilitate powerful and unprecedented transformations. However, the implementation of light-driven reactions in large-scale processes remains uncommon, limited by the lack of general technologies for the immobilization, separation, and reuse of these diverse catalysts. Here, we report a new class of photoactive organic polymers that combine the flexibility of small-molecule dyes with the operational advantages and recyclability of solid-phase catalysts. The solubility of these polymers in select non-polar organic solvents supports their facile processing into a wide range of heterogeneous modalities. The active sites, embedded within porous microstructures, display elevated reactivity, further enhanced by the mobility of excited states and charged species within the polymers. The independent tunability of the physical and photochemical properties of these materials affords a convenient, generalizable platform for the metamorphosis of modern photoredox catalysts into active heterogeneous equivalents.