Print this page

Degree Dissertations

Last modified: December'18

PhD Thesis (2018)

Scalable Exploration of 3D Massive Models
  

[12.4 MB, english, PDF]
 

This thesis introduces scalable techniques that advance the state-of-the-art in massive model creation and exploration.

Concerning model creation, we present methods for improving reality-based scene acquisition and processing, introducing an efficient implementation of scalable out-of-core point clouds and a data-fusion approach for creating detailed colored models from cluttered scene acquisitions. The core of this thesis concerns enabling technology for the exploration of general large datasets.

Two novel solutions are introduced. The first is an adaptive out-of-core technique exploiting the GPU rasterization pipeline and hardware occlusion queries in order to create coherent batches of work for localized shader-based ray tracing kernels, opening the door to out-of-core ray tracing with shadowing and global illumination. The second is an aggressive compression method that exploits redundancy in large models to compress data so that it fits, in fully renderable format, in GPU memory. The method is targeted to voxelized representations of 3D scenes, which are widely used to accelerate visibility queries on the GPU. Compression is achieved by merging subtrees that are identical through a similarity transform and by exploiting the skewed distribution of references to shared nodes to store child pointers using a variable bit-rate encoding

The capability and performance of all methods are evaluated on many very massive real-world scenes from several domains, including cultural heritage, engineering, and gaming.

Keywords: Computer Graphics, Scalable Rendering, Out-of-core algorithms, Raytracing, Voxels, Level-of-detail.

Master Thesis (2012)

Point Cloud Manager
A multiresolution system for managing masive point cloud datasets


[3.3 MB, spanish, PDF]

With the recent evolution of LIDAR (Light Detection and Ranging) and 3D scanning equipment, many professional fields such as Topology, Architecture, Civil and Industrial Engineering, VFX, etc. are incorporating this technology to their standard workflow. They are able to capture the surrounding reality (objects, environment) and its features with a great precision and reliability. These huge datasets, called point clouds, are composed of even billions of 3D samples without topological structure. Each one of these points contains its space coordinates and different data representing its properties, such as color, reflection index, temperature, etc.

Professionals normally use point clouds to detailed visualization of the subject of study, and also to apply complex algorithms (such as filtering, measurement, etc.) for analysis.

Working with these huge datasets (in the order of terabytes) is very expensive as they usually overweight the memory capacities of current systems (in the order of gigabytes), even high-end workstations. Moreover the structure of point clouds (3D fixed single elements massively repeated) makes it perfect candidates for parallel processing and rendering. Nowadays GPUs offer this possibility in low-end machines, both for visualization (OpenGL) and computing (OpenCL or CUDA).

This Master Thesis presents Point Cloud Manager, a complete solution for out-of-core visualization and processing of arbitrary-sized point clouds in commodity hardware. To achieve this goals, its core library implements two new concepts:

  • Multiresolution spatial hierarchical structures that optimize data organization using levels-of-detail.
  • Dual-level software cache that efficiently transfer data-chunks across different levels of the memory hierarchy (HDD – RAM – VRAM) exploiting its spatial nature.

In addition, the whole PCM package offers a set of tools for end-users as well as a complete framework for developers that allows to extend the functionality and to implement user-custom processes, both for CPUs and GPUs.

Keywords: point cloud, multiresolution system, big-data, parallel processing, GPU computing, out-of-core rendering, hierarchical data structure, multilevel software cache, levels-of-detail.

Computer Science graduation thesis (2011) 

Multitouch interactive surfaces:
development of prototype, control system and application paradigms for 3D visualization


[5.2 MB, spanish, PDF]

I presented my end of degree project at the University of A Coruña as one of the works I made working with VideaLab. It is about multitouch surfaces and its application to 3D visualization.

There are three main objectives or subjets in it:

  • Software: design and development of a cross-platform, multithread, client-server system to control any kind of multitouch device (even home-made ones), with a easy interface to be integrated with arbitrary applications.
  • Hardware: the construction of a full usable prototype, following the optics principle of FTIR, that was broadly used in different presentations, exhibitions, etc.
  • Research: the creation and development of two new paradigms of 3D movement in 3D that responses to the gestures of the user, as a natural interface. Mainly to be used in architecture and terrain visualization. They are called PRotaZoom and Walk-through.