Abstract. Game generation systems perform automated, intelligent design of games (i.e. videogames, boardgames), reasoning about both the abstract rule system of the game and the visual realization of these rules. Although, as an instance of the problem of creative design, game generation shares some common research themes with other creative AI systems such as story and art generators, game generation extends such work by having to reason about dynamic, playable artifacts. Like AI work on creativity in other domains, work on game generation sheds light on the human game design process, offering opportunities to make explicit the tacit knowledge involved in game design and test game design theories. Finally, game generation enables new game genres which are radically customized to specific players or situations; notable examples are cell phone games customized for particular users and newsgames providing commentary on current events. We describe an approach to formalizing game mechanics and generating games using those mechanics, using WordNet and ConceptNet to assist in performing common-sense reasoning about game verbs and nouns. Finally, we demonstrate and describe in detail a prototype that designs micro-games in the style of Nintendo’s WarioWare series. 1

The ability to create and edit a model of a large-scale city is necessary for a variety of applications. Although the layout of the urban space is captured as images, it consists of a complex collection of man-made structures arranged in parcels, city blocks, and neighborhoods. Editing the content as unstructured images yields undesirable results. However, most GIs maintain and provide digital records of metadata such as road network, land use, parcel boundaries, building type, waterlsewage pipes and power lines that can be used as a starting point to infer and manipulate higher-level structure. We describe an editor for interactive reconfiguration of city layouts, which provides tools to expand, scale, replace and move parcels and blocks, while efficiently exploiting their connectivity and zoning. Our results include applying the system on several cities with different urban layout by sequentially applying transformations.

In the digital entertainment industry, cities are one of the largest artifacts modeled by artists. One alternative to modeling an entire city by hand is to use an urban simulation. Often, those simulations use a gridded terrain representation. Translating gridded simu-lation results into a more continuous, realistic representation can often be difficult. Our vectorization process transforms gridded urban land use data into a representation that mimics what might be seen in GIS or online mapping tools. Our method consists of three major phases. In the first phase, the raster data is is analyzed and the transportation layer is abstracted and filtered. Next, the city blocks are constructed from the raster data. Third, the blocks are subdivided and land use and density are assigned to each constructed parcel. The result is scored and compared with the output from the simulation to match location, sizes, and proportions.

Abstract. Generating realistic-looking but interesting terrains quickly is a great challenge. We present RiverLand, an efficient system for terrain synthesis. RiverLand creates a realistic-looking terrain by first generating river networks over the land. Then, the terrain is created to be consistent with the river networks. In this way, the terrains created have a proper drainage basin, an important feature lacking in many existing procedural terrain methods. The terrains generated by RiverLand are also widely varied, with rolling hills, river valleys, alpine mountains, and rocky cliffs, all seamlessly connected in the same terrain. Since RiverLand does not use complex fluid simulations, it is fast, and yet is able to produce many of the erosion features generated by the simulation methods. 1

I particularly acknowledge my supervisor George Drettakis for the hundreds (thousands?) of hours spent with me during my PhD, his great ideas and the cool supervision work he did. I also want to acknowledge my main collaborators: Michiel van de Panne, who gave great ideas for the second part of the thesis, and who is very cool as well, and Frédo Durand, who hosted me at MIT-CSAIL for a month, initiating the successful ‘hair project ’ (Part 2, Chapter 6). I also thank my co-authors, in particular Sylvain Paris, Sylvain Lefebvre, Clara Suied, Nicolas Tsingos and Isabelle Viaud-Delmon, as well as our modelers. In particular, Fernanda Andrade-Cabral who did a great job modeling most scenes in a rush during deadlines. A big thank you to Monique who put up with me for almost 3 years in the same office, and managed to drive our European project CROSSMOD at the same time, as a project assistant. By the way, I also thank all the CROSSMOD team, involving ISTI CNR-Pisa, UNIBRIS-Bristol, CNRS/IRCAM-Paris, VUT-Vienna, FAU-Erlangen, for this fruitful collaboration. I acknowledge the reviewers and members of the jury, for their time spent on my thesis and for their interesting feedback.

Figure 1: As the user zooms in, terrain geometry is created to adaptively refine the planet. Realistic terrain models are required in many applications, especially in computer games. Commonly, procedural models are applied to generate the corresponding models and let users experience a wide variety of new environments. Existing algorithms generate landscapes immediately with view-dependent resolution and without preprocessing. Unfortunately, landscapes generated by such algorithms lack river networks and therefore appear unnatural. Algorithms that integrate realistic river networks are computationally expensive and cannot be used to generate a locally adaptive high resolution landscape during a fly-through. In this paper, we propose a novel algorithm to generate realistic river networks. Our procedural algorithm creates complete planets and landscapes with realistic river networks within seconds. It starts with a coarse base geometry of a planet without further preprocessing and user intervention. By exploiting current graphics hardware, the proposed algorithm is able to generate adaptively refined landscape geometry during fly-throughs.

Figure 1. Urban Layout Synthesis. First, our method extracts the street network and per-parcel aerial-view images from real-world urban layouts. Then, using an example-based approach new layouts are interactively created by synthesizing streets and images based on data from the example layout fragments. We present an interactive system for synthesizing urban layouts by example. Our method simultaneously performs both a structurebased synthesis and an image-based synthesis to generate a complete urban layout with a plausible street network and with aerial-view imagery. Our approach uses the structure and image data of real-world urban areas and a synthesis algorithm to provide several high-level operations to easily and interactively generate complex layouts by example. The user can create new urban layouts by a sequence of operations such as join, expand, and blend without being concerned about low-level structural details. Further, the ability to blend example urban layout fragments provides a powerful way to generate new synthetic content. We demonstrate our system by creating urban layouts using example fragments from several real-world cities, each ranging from hundreds to thousands of city blocks and parcels.

Figure 1: Terrain modeling example: as each stroke is drawn (a) or manipulated (c), the terrain is tessellated in the GPU to follow the stroke. In the CPU, the quadtree data structure (b) controls the quad patches sent to the GPU. Introduction and Related Work In many editing tools, especially sketch-based modeling, it is important to have real-time feedback to help improve the editing quality. This importance is emphasized particularly in sketch-based terrain modeling, being able to see the terrain morphing at the same time the drawing edition occurs

We present a step toward interactive physics-based modeling of terrains. A terrain, composed of layers of materials, is edited with interactive modeling tools built upon different physics-based erosion and deposition algorithms. First, two hydraulic erosion algorithms for running water are coupled. Areas where the motion is slow become more eroded by the dissolution erosion, whereas in the areas with faster motion, the force-based erosion prevails. Second, when the water under-erodes certain areas, slippage takes effect and the river banks fall into the water. A variety of local and global editing operation is provided. The user has a great level of control over the process and receives immediate feedback since the GPU-based erosion simulation runs at least at 20 fps on off-the-shelf computers for scenes with grid resolution of 2048 × 1024 and four layers of material. We also present a divide and conquer approach to handle large terrain erosion, where the terrain is tiled, and each tile calculated independently on the GPU. We show a wide variety of erosion-based modeling features such as forming rivers, drying

Terrain modeling is an important task in digital content creation and physics-based approaches have the potential to simplify it by introducing a higher level of realism. However, most of the existing simulations are hindered by a low level of user control, because they fail on large-scale phenomena, or because they are focused only on the modeling of limited effects. We introduce a new interactive, intuitive, and accessible physics-based framework for digital terrain editing. Our solution is suitable for users involved in digital content authoring (such as game designers, artists, 3D modelers, and digital content providers) and does not assume any in-depth knowledge about physics-based simulations. To address the scalability issues of previous algorithms, we provide an adaptive GPU-amenable solution. Large terrains can be loaded from external sources, generated procedurally, or created manually, and they are edited at interactive framerates on the GPU. We introduce two simplifications that allow us to perform large scale editing. First, the terrain is divided into tiles of different resolutions according to the complexity of the underlying terrain. Second, each tile is stored as a mip-map texture and different levels of detail are used during the physics-based simulation depending on the dynamics of terrain changes. We demonstrate our approach at several examples including the editing of terrains, brushing with hydraulic simulation, and blending terrain patches into an existing terrain. In comparison with nonadaptive computation, by using our approach we can achieve a 50 % speedup with a simultaneous 25 % savings of memory. Most important, we can process terrain sizes that were not possible to process with previous approaches.