of elements which communicate with each other, it is to be expected to get a mayor speedup when performing the cloth simulation on the GPU side, too. This is analyzed in [SKBK13], based on the idea to limit the simulation model mainly to the communication problem itself. The results show promising speedups due to a novel hybrid multi-grid simulation approach with low resolution CPU computations and high resolution GPU simulation. The two com- putation units target different complexities, which adds to the performance of the approach. Again this still needs to be combined with my work presented in this document.
6.4 Other Applications Within Garment Design
Figure 6.2: The designers can work as a team, observed by a supervisor, who guides the creativity towards specified directions, for example certain product lines of a company.
The methods presented in this work are not limited to be used together with garment simulation. They can be used stand-alone as shown in the following application example.
What needs to be done in order to let a designer (artist) do art, without endangering the producibility of the product as mentioned in the introduction? Or to be described in a more technical manner: how can I transform the things the designer does into parameter structures necessary for production and give him a direct visual feedback of what he does?
Figure 6.3: The kind of operation a designer performs can be depicted as a classical model- view-control pattern. The model is in this case the arrangement of appliqués on a 2D sewing patterns. The view is the 3D presentation of the appliqués on the virtual garment and the control is the interaction and operations the designer performs, changing the initial arrangements, closing the loop.
One method to ensure this is to let all data he generates stay in the 2D world of the sewing patterns. For appliqués it is now possible to build a design tool, which can be used to support design teams in their initial creativity process. This process can be performed by small design teams optionally supported by a supervisor, which guides the design processes into certain directions (see Figure6.2). Assuming new designs are often influenced on existing designs, it is possible to modify appliqués on the existing virtual prototype’s surface using digitizers on tablet computers.
Here, my presented techniques can be used to ensure the generated data to stay in the 2D world of the sewing pattern: When using a digitizer on a virtual prototype on a tablet computer it is trivial to get the corresponding 2D coordinates within sewing pattern space, since the sewing pattern space is basically a linear transformed version of the texture space on a pattern in 3D. Therefore, appliqués can be directly specified in the 2D pattern space using a digitizer. For visualization feedback my presented techniques can now directly re- project the appliqués into the cloth surface of the virtual prototype. This closes the feedback loop necessary to allow interactive editing (see Figure6.3). Since no simulation is involved, current tablet computers are powerful enough to allow this kind of operation in real time.
6.4 Other Applications Within Garment Design
Figure 6.4: In a creative team anyone can communicate with everyone. This needs to be reflected within the creativity tool process.
The data footprint of the necessary operation is small. It mainly consists of information regarding what appliqué has to be placed where in a 2D world. Since tablet computers usually work in a network environment and the generated data has a small footprint it is possible to quickly exchange designs between single members of the design team or present them to the whole team (see Figure6.4).
This would create a versatile tool for the creative design process. However, its details and its real usefulness is another topic and open for future research.
In the overall summary, my presented method can handle changes of the scene within geometry level, photo-realistic rendering and surface detailing in a 2D CAD friendly way at the same time. However, parts of this work can be used as stand-alone, too, for example to provide stand-alone editing capabilities.
A Publications and Talks
The thesis is partially based on the following publications and talks:
A.1 Publications
1. Arnulph Fuhrmann, Clemens Groß, Martin Knuth and Jörn Kohlhammer Virtual Prototyping of Garments.
ProSTEP iViP Science Days 2005, September 2005. 2. Martin Knuth and Arnulph Fuhrmann
Self-Shadowing of dynamic scenes with environment maps using the GPU. WSCG FULL papers proceedings, pages 31–38, February 2005.
3. Martin Knuth and Jörn Kohlhammer
A hybrid ambient occlusion technique for dynamic scenes.
WSCG 2009. Communication Papers Proceedings, pages 1–8, February 2009. 4. Martin Knuth and Jörn Kohlhammer
Embedding Hierarchical Deformation within a Real-time Scene Graph.
VISIGRAPP 2010, International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications. Proceedings, pages 246–253, May 2010.
5. Martin Knuth and Jörn Kohlhammer
A geometry-shader-based adaptive mesh refinement scheme using semiuniform quad/triangle patches and warping.
VRIPHYS 2010, 7th Workshop on Virtual Reality Interaction and Physical Simulation, pages 21–29, November 2010.
6. Fabian Bauer, Martin Knuth, Jan Bender and Arjan Kuijper Screen-space ambient occlusion using A-buffer techniques.
13th International Conference on Computer-Aided Design and Computer Graphics, pages 140–147, November 2013.
7. Nikolas Schmitt, Martin Knuth ,Jan Bender and Arjan Kuijper Multilevel cloth simulation using GPU surface sampling.
VRIPHYS 2013, 10th Workshop on Virtual Reality Interaction and Physical Simula- tion, pages 1–10, November 2013.
8. Martin Knuth, Christian Altenhofen, Jan Bender and Arjan Kuijper
Efficient Self-Shadowing Using Image-Based Lighting on Glossy Surfaces. VMV 2014, pages 159–166, October 2014.
9. Martin Knuth, Jan Bender, Michael Goesele and Arjan Kuijper Deferred Warping.
B Supervising Activities
The following list summarizes the student bachelor, diploma and master theses supervised by the author. The results of these works were partially used as an input into the thesis.
B.1 Diploma and Master Thesis
1. Michael Benz, Cluster based Rendering with Distributed Frame Buffers, TU- Darmstadt, FG Graphisch interaktive Systeme (GRIS), FB Informatik, 2005
2. Carsten Schoger, Globale Beleuchtung interaktiver Szenen, TU-Darmstadt, FG Graphisch interaktive Systeme (GRIS), FB Informatik, 2007
3. Jens Tinz, GPU-basierte Bekleidungssimulation, TU-Darmstadt, FG Graphisch inter- aktive Systeme (GRIS), FB Informatik, 2007
4. Christian Altenhofen, TU-Darmstadt, FG Graphisch interaktive Systeme (GRIS), FB Informatik, 2010
5. Fabian Bauer, Realistic Realtime Rendeirng of Garment with Transparency and Am- bient Occlusion, TU-Darmstadt, FG Graphisch interaktive Systeme (GRIS), FB Infor- matik, 2013
6. Nikolas Schmitt, Multilevel cloth simulation through GPU surface sampling, TU- Darmstadt, FG Graphisch interaktive Systeme (GRIS), FB Informatik, 2013