A5102930574- Computer Graphics and Visualization Techniques
- Advanced Vision and Imaging
- Image Enhancement Techniques
- Scientific Computing and Data Management
- Interactive and Immersive Displays
- Geological Modeling and Analysis
- Computational Physics and Python Applications
- Geoscience and Mining Technology
- 3D Surveying and Cultural Heritage
- Distributed and Parallel Computing Systems
- Generative Adversarial Networks and Image Synthesis
- 3D Shape Modeling and Analysis
- Geomechanics and Mining Engineering
- Artificial Intelligence in Games
Zhejiang University
2023-2024
University of Leicester
2024
Microsoft Research Asia (China)
2023-2024
Zhejiang University of Science and Technology
2023
This paper presents a novel neural implicit radiance representation for free viewpoint relighting from small set of unstructured photographs an object lit by moving point light source different the view position. We express shape as signed distance function modeled multi layer perceptron. In contrast to prior relightable representations, we do not disentangle transport components, but model both local and global at each second perceptron that, in addition, density features, current position,...
This paper presents a novel method for exerting fine-grained lighting control during text-driven diffusion-based image generation. While existing diffusion models already have the ability to generate images under any condition, without additional guidance these tend correlate content and lighting. Moreover, text prompts lack necessary expressional power describe detailed setups. To provide creator with over generation, we augment text-prompt information in form of radiance hints, i.e.,...
We propose a unified structured light, consisting of an LED array and LCD mask, for high-quality acquisition both shape reflectance from single view. For geometry, one projects set learned mask patterns to accurately encode spatial information; the decoded results multiple LEDs are then aggregated produce final depth map. appearance, light cast through transparent efficiently probe angularly-varying reflectance. Per-point BRDF parameters differentiably optimized with respect corresponding...
We present a spatial and angular Gaussian based representation triple splatting process, for real-time, high-quality novel lighting-and-view synthesis from multi-view point-lit input images. To describe complex appearance, we employ Lambertian plus mixture of Gaussians as an effective reflectance function each Gaussian. generate self-shadow, splat all towards the light source to obtain shadow values, which are further refined by small multi-layer perceptron. compensate other effects like...
In SC'21, Alexandros Nikolaos Ziogas et al. proposed a Data-Centric Python workflow in their DaCe paper. provides high productivity, performance, and portability with language extensions automatic optimizations. We reproduce the performance evaluation results from paper on both CPU GPU Azure CycleCloud cluster. also scaling up to 32 nodes 64 processes. Our show that has outstanding scalability provided cluster, accordance SC