|
|
 |
|
Seminar Computer Graphics
|
Seminar Computer Graphics
|
| Time, Place: |
Thu 14:00 - 15:30, MI 02.13.010 |
| Begin: |
14.04.2005 |
| Prerequisites: |
none
|
| Registration: |
Click here.
|
Registration
Registration has been closed.
Contents
The main purpose of this seminar is to present and explore state of the art
terrain modelling, processing and rendering techniques. The main focus will be the interactive processing
and rendering of large-scale terrain data.
With the advent of new technology such as GPS and satellite measurement, large-scale, ultra-high resolution
terrain data is commonly available. Current state of the art scans include for instance an elevation map of the entire
USA with a grid-spacing of 10m and an elevation accuracy of less than 1.0m. Hence the resulting amount of data is
tremendous (about 40GB for the USA). As a logical consequence the interest in algorithms capable of
interactively managing and rendering these datasets for both civil and military purposes comes to no surprise.
In this seminar attendants should become familiar with the basic concepts of dealing with such data. Topics to be
discussed will include algorithms for modelling, remeshing, compressing and rendering large terrains, as well as some
processing algorithms such as filling holes in the elevation data or deriving height data from contour lines.
Organization
To sucessfully participate in this seminar, the attendants will have to prepare a talk, which should
last about 60 minutes. Supplementary, a short workout (approximately 4 pages) should be prepared.
During the preparation for the talks, independing investigation for further reading is desired.
For more details, please see Seminar FAQ (PDF).
Topics and Talks
Further Reading & Links
Related Files
By today the resolution of non-classified elevation data has reached 10m gridspacing and 0.1m vertical tolerance.
This high precision was made possible using satellite-based Earth Observing Systems that scan the Earth’s surface
using multiband imaging devices.On December 18, 1999 NASA launched TERRA (EOS AM-1) as its Earth Observing System
flagship. The satellite carries an ASTER (Advanced SPaceborne Thermal Emission and Reflection Radiometer) imaging
device, that is capable of obtaining detailed maps of land surface temperature, emissivity, reflectance, and
elevation.
The attendee should give an overview into the process of data acquisition using such an ASTER device.
NN.
Supervisor: NN
Usually terrain data is given with respect to a local parameterization of the Earth’s surface. Where these
local coordinate systems meet, data has to be merged along the boundary. A similar problem is the projection
of the Earth to a 2D plane. Since this is the very content of cartography, a vast amount of projections trying
to achieve different goals has evolved during the last centuries or even millennia. Some of these projections
have proven themselves to be so adequate for visualization purposes, and hence become so popular, that we
immediately “feel at home” when looking at them.
The seminar attendee should give an introduction to these local coordinate systems, along with their advantages
and disadvantages. Over that, she or he should present the evolution of projection systems during the history
of cartography abroad.
NN.
Supervisor: NN
The world we confront every day is visually complex. The pursuit of realism in computer graphics is largely a
problem of reproducing that complexity in synthetic images. Fractal geometry is our first cogent language of visual
complexity; its lexicon, fractals, provides a potent vocabularity for complex form, particularly the kinds of forms
found in Nature. Fractal geometry can map chaotic complexity into the terse, deterministic idion of mathematics in a
way that, as we shall see, is uniquely suited to the capabilities of the computer. Computer graphics, on the other hand,
can map complex synthetic fractal constructions into the form that is best suited to human cognition: images.
Tayfur Coskun
Supervisor: Prof. Dr. R. Westermann
The seminar attendee is supposed to give the audience a broad and very basic introduction to several aspects
of terrain rendering. Questions to be answered include, but are not necessarily limited to:
- how can Terrain be naively triangulated and rendered
- what are the problems / bottlenecks of naďve approaches
- how does texturing work
- how are normals obtained and what to do with them
- why is Level-of-Detail needed and what happens if one ignores it
- what is anti-aliasing
- how does Frustum and Occlusion Culling work
- how can shadows be rendered for terrain
Manuel Hampel
Supervisor: Prof. Dr. R. Westermann
To illustrate […] just how simply the procedural approach can generate piles and piles of visual detail, I designed […]
an algorithm. Again, the goal was maximal simplicity in the algorithm, period. Accordingly, I expected it to be really
slow. It came as a considerable surprise when it turned out to be only very slow, not glacial. (That is, it took on the
order of a minute to create an image, when I was expecting days.) I gave this algorithm the wonderfully turgid name
‘quasi-analytic error-bounded ray tracing’, or QAEB tracing for short. To balance the scales of pretense, I pronounce
the acronym QAEB whimsically ‘kweeb’ )to rhyme with ‘dweed’, of course).
-- F. Kenton Musgrave on QAEB-Tracing
Manuel Reinhardt
Supervisor: Prof. Dr. R. Westermann
We present a technique to perform occlusion culling for hierarchical terrains at run-time. The algorithm
is simple to implement and requires minimal pre-processing and additional storage, yet leads to 4-6 times
improvement in framerate for views with high degrees of occlusion. Our method is based on the well-known
occlusion horizon algorithm. We show how to adapt the algorithm for use with hierarchical terrains. The occlusion
horizon is constructed as the terrain is traversed in an approximate front to back ordering. Regions of the terrain
are compared to the horizon to determine when they are completely occluded from the viewpoint. Culling these regions
leads to significant savings in rendering.
N.N.
Supervisor: N.N.
Terrain visualization is a difficult problem for applications requiring accurate images of large datasets at high
frame rates, such as flight simulation and ground-based aircraft testing using synthetic sensor stimulation. On
current graphics hardware, the problem is to maintain dynamic, view-dependent triangle meshes and texture maps that
produce good images at the required frame rate. We present an algorithm for constructing triangle meshes that optimizes
flexible view-dependent error metrics, produces guaranteed error bounds, achieves specified triangle counts directly,
and uses frame-to-frame coherence to operate at high frame rates for thousands of triangles per frame. Our method,
dubbed Real-time Optimally Adapting Meshes (ROAM), uses two priority queues to drive split and merge operations that
maintain continuous triangulations built from preprocessed bintree triangles. We introduce two additional performance
optimizations: incremental triangle stripping and prioritycomputation deferral lists. ROAM execution time is proportionate
to the number of triangle changes per frame, which is typically a few percent of the output mesh size, hence ROAM
performance is insensitive to the resolution and extent of the input terrain. Dynamic terrain and simple vertex morphing
are supported.
Matous Sedlacek
Supervisor: J. Schneider
Height fields play an important role in the fast growing domain of Geographic Information Systems (GIS). For
exploring different kinds of geographic-based data sets on screen it is necessary to display height fields at
interactive frame rates. Because of the inherent geometric complexity, this goal is often unachievable even with
new generations of powerful graphics computers, unless the original height field data is approximated in order to
reduce the number of geometric primitives that need to be rendered without compromising visual quality.
So far most algorithms have focused on global reduction or multi-resolution techniques, which reduce resolution on
the basis of surface roughness. A recent new approach called Continuous Levels of Detail introduced a hierarchical
quadtree technique. In order to reduce the projected pixel error, the height field is dynamically triangulated in a
bottom up fashion according to the distance to the point of view. Since resolution is allowed to change smoothly, the
result is a much better image quality. However, this algorithm still has a major disadvantage. With the viewpoint moving,
the triangulation is continuously changing, resulting in a phenomenon called vertex popping. As the observer approaches
an area with detail information, this detail will suddenly appear at a certain distance. To eliminate these artifacts
we introduce a new, rapid geomorphing algorithm, which operates top down on a quadtree data structure.
Wolfgang Kirchler
Supervisor: J. Schneider
This paper describes a general framework for out-of-core rendering and management of massive terrain surfaces.
The two key components of this framework are: view-dependent refinement of the terrain mesh; and a simple scheme
for organizing the terrain data to improve coherence and reduce the number of paging events from external storage
to main memory. Similar to several previously proposed methods for viewdependent refinement, we recursively subdivide
a triangle mesh defined over regularly gridded data using longest-edge bisection. As part of this single, per-frame
refinement pass, we perform triangle stripping, view frustum culling, and smooth blending of geometry using geomorphing.
Meanwhile, our refinement framework supports a large class of error metrics, is highly competitive in terms of
rendering performance, and is surprisingly simple to implement.
Independent of our refinement algorithm, we also describe several data layout techniques for providing coherent
access to the terrain data. By reordering the data in a manner that is more consistent with our recursive access
pattern, we show that visualization of gigabyte-size data sets can be realized even on low-end, commodity PCs
without the need for complicated and explicit data paging techniques. Rather, by virtue of dramatic improvements
in multilevel cache coherence, we rely on the built-in paging mechanisms of the operating system to perform this
task. The end result is a straightforward, simple-to-implement, pointerless indexing scheme that dramatically
improves the data locality and paging performance over conventional matrix-based layouts.
N.N.
Supervisor: N.N.
We describe an efficient technique for out-of-core management and interactive rendering of planet sized
textured terrain surfaces. The technique, called P-Batched Dynamic Adaptive Meshes (PBDAM), extends the
BDAM approach by using as basic primitive a general triangulation of points on a displaced triangle. The
proposed framework introduces several advances with respect to the state of the art: thanks to a batched
host-to-graphics communication model, we outperform current adaptive tessellation solutions in terms of
rendering speed; we guarantee overall geometric continuity, exploiting programmable graphics hardware to cope
with the accuracy issues introduced by single precision floating points; we exploit a compressed out of core
representation and speculative prefetching for hiding disk latency during rendering of out-of-core data; we
efficiently construct high quality simplified representations with a novel distributed out of core simplification
algorithm working on a standard PC network.
Martin Schreiber
Supervisor: J. Schneider
Rendering throughput has reached a level that enables a novel approach to level-of-detail (LOD) control
in terrain rendering. We introduce the geometry clipmap, which caches the terrain in a set of nested regular
grids centered about the viewer. The grids are stored as vertex buffers in fast video memory, and are incrementally
refilled as the viewpoint moves. This simple framework provides visual continuity, uniform frame rate, complexity
throttling, and graceful degradation. Moreover it allows two new exciting real-time functionalities: decompression
and synthesis. Our main dataset is a 40GB height map of the United States. A compressed image pyramid reduces the
size by a remarkable factor of 100, so that it fits entirely in memory. This compressed data also contributes normal
maps for shading. As the viewer approaches the surface, we synthesize grid levels finer than the stored terrain
using fractal noise displacement. Decompression, synthesis, and normal-map computations are incremental, thereby
allowing interactive flight at 60 frames/sec.
Simon Bolek
Supervisor: J. Schneider
We describe an approach for rendering large terrains in real-time. A digital elevation map defines the rough
shape of the terrain. During rendering, procedural geometric and texture detail is added by the graphics hardware.
We show, how quad meshes can be generated quickly that have a locally varying resolution that is optimized for the
inclusion of procedural detail.We obtain these distorted meshes by importance based warping of geometry images. The
resulting quad mesh can then be rendered very efficiently by graphics hardware, which also adds all visible procedural
detail using vertex and fragment programs.
N.N.
Supervisor: N.N.
Contour lines from topographic maps are still the most common form of elevation data for the
Earth’s surface and in the case of historical landscapes, they often are the only available source
of information. In this paper we present a new contour interpolation method that solves this bivariate
problem by considering univariate curve interpolation along the approximate gradient directions of the
unknown surface. For a point between two contours the height value is computed with Hermite interpolation
based on the shortest distances to the contours and height and derivative information at the contours.
The surfaces generated are C1 except at terrain characteristics such as ridges and valleys which are
reconstructed as sharp features. The method also faithfully reconstructs summits, pits, and saddles and is
especially well-suited for sparse sets of contours. The approach allows for an efficient numerical
implementation as we demonstrate with a number of examples.
Alexander Gafriller
Supervisor: J. Schneider
Sunlight and skylight are rarely rendered correctly in computer graphics. A major reason for this is
high computational expense. Another is that precise atmospheric data is rarely available. We present an
inexpensive analytic model that approximates full spectrum daylight for various atmospheric conditions. These
conditions are parameterized using terms that users can either measure or estimate. We also present an inexpensive
analytic model that approximates the effects of atmosphere (aerial perspective). These models are fielded in a number
of conditions and intermediate results verified against standard literature from atmospheric science. These models are
analytic in the sense that they are simple formulas based on fits to simulated data; no explicit simulation is
required to use them. Our goal is to achieve as much accuracy as possible without sacrificing usability.
N.N.
Supervisor: N.N.
|
|
|
|
