M. Ioannides et al. (Eds.): EuroMed 2012, LNCS 7616, pp. 505–512, 2012. © Springer-Verlag Berlin Heidelberg 2012

From Real to Virtual Rapid Architectural Prototyping

Mike Tato1, Petros Papanikolaou2, and George Papagiannakis2

1 3DTouch S.A. 59 Route des Jeunes, 1227 Carouge, Switzerland

mt@3dtouch.ch 2 University of Crete, Computer Science Department, Leoforos Knossou,

71409, Heraklion and Foundation for Research and Technology Hellas, 100 N. Plastira Str.,

70013, Heraklion, Greece {ppapanik,papagian}@csd.uoc.gr

Abstract. Can greater visual realism of a real-time architectural virtual walkthrough achieve similar high sensory impact, or qualia, as a fabricated 3D printed scale model of an urban landscape? The aim of this project is to answer that question by allowing a real existing city heritage landscape during a large urban planning project to be 3D modeled and subsequently be studied via a dual output: a fabricated real, physical scale model based on a latest high quality color 3D printer and an equivalent 3D virtual walkthrough of enhanced real-time visual realism based on a recent game engine. Conclusions of this experiment and user study suggest that a virtual, interactive simulation based on specific latest real-time rendering algorithms can indeed convey a similar user experience and feeling of “presence” that an equivalent architectural scale model offers, regarding fast appreciation of both space and structure.

Keywords: 3D Printing, Scale models, presence, virtual environment, serious games, architectural walkthrough, rapid prototyping.

1 Introduction and Main Concept

Recent studies [1], [2] have shown that visual realism enhances realistic response and the feeling of “presence” while interacting with an immersive virtual environment. Recent advances in 3D printing technology [3] [4] have allowed for a transformation of the process on how architectural and figure scale models are built; based on a 3D model of a CAD system or skinned mesh, a complete, colored small-scale model or scale articulated character can be built in an efficient and fast manner. Our work aims to extend the intrinsic properties of real scale architectural models and compare them with latest real-time game engine powered virtual walkthroughs of the equivalent architectural scene (Fig. 1), for enhanced, large-scale urban planning and experimentation.

506 M. Tato, P. Papanikolaou, and G. Papagiannakis

Fig. 1. Real architectural scene; corresponding virtual scene; larger-area scale model; equivalent larger-area virtual model

2 Novelty

Our main novelty is the research on whether rapid prototyping based on advanced, serious-game 3D rendering is more suitable than rapid prototyping via 3D printing, given the same 3D architectural models as starting point. This targets the application of future architectural urban planning and visualization, in a low-cost, efficient methodology. Our final goal is to overcome the need for scale models in the future via incorporating the unique features that they offer in a virtual, serious game environment: low cost, portability of experience, fast examination of alternative structures via material and object editing, agent simulation and finally heightened presence and user experience (UX).

3 Implementation and Vision

The same, existing real architectural scene was substantiated with rapid prototyping using CAD software and a) the ZPrinter™ 650 color 3D printer for the real scale model and b) the Unigine™ [5] game engine and their programming environment for real-time simulation of the same scene (as depicted in the scale model) as shown in Fig. 2 below. Advanced 3D printing features such as multi-part coloring were employed as well as latest 3D rendering techniques for HDR image based day-night sun-path simulation, spherical harmonics-based cube map interpolation, screen-space ambient occlusion, dual-quaternion camera path animation and cloud, particle systems for natural, event-controlled atmospheric effects and crowd-agent simulation.

The user study experiment involved 15 participants, asked to search for a specific object (a model of a car) hidden inside the real scale model and subsequently to perform the same task in the equivalent real-time virtual scene. After the completion of each task they were given a questionnaire to complete, adapted from [2]. The results of our experiment suggest that enhanced visual realism as provided by latest game engines can enhance realistic behavioral response, despite the different interaction metaphors and scaling. Thus we can envisage that in the near future we can avoid large scale-models and aim instead for similar real-time experiences.

Our system process (Fig. 2) is based on photogrammetric models, textures, objects combined with architectural plans that are provided to our architects in order to reconstruct fully the urban heritage area in a modern 3D modeling tool. Then we can

From Real to Virtual Rapid Architectural Prototyping 507

Fig. 2. The overall project system workflow: from the architectural 3D model to the dual output: fabricated from a 3D printer scale model and equivalent virtual scale model with advanced editing capabilities

export the static scene in a dual output directly via a 3D printer as a 3D scale model and via game-engine plugins and tools as a real-time 3D walkthrough. The first output we cannot enhance it further but the second real-time one, we can further take advantage of the scripting language capabilities of the game-engine and instill modern real-time rendering serious-game elements for enhanced presence and interaction.

In the following sections we describe how the process to export the 3D elements for real-time simulation was carried-out, how the real-time editing of architectural elements was introduced and how a relatively fast system for crowd simulation was created. The reason for the description of these extra elements lies in the fact that the users from our study appreciated greatly these features, as they were not available in the traditional 3D printed scale model. In the final section we draw conclusions and describe future work.

3.1 Editing Architectural Objects for Space Hypotheses

Considering 3D buildings and other static elements have already been modeled properly, we needed an easy, efficient and fast way to add furniture and other user-hypothetical 3D elements in the inner room structures. Assuming that such static 3D objects are nodes in a scene graph, by default in order to place any of them in the world, we followed the standard following steps:

Architectural 3D modeling

Photogrammetric models and textures

3D Printer authoring

Real-time engine authoring

Architectural plans

3D scale model

Real-time 3D walkthrough

508 M. Tato, P. Papanikolaou, and G. Papagiannakis

1. Load the engine WYSIWYP (what-you-see-is-what-you-play) editor 2. Browse through the scene graph of the world to find the furniture as node 3. Select the node and clone it 4. Place manually the cloned one in the appropriate position and orientation using

the GUI

There were at least several problems at this point. The first one was that it was possible for the scene graph to grow up enormously consisting of hundreds nodes. Users had to browse through the whole scene graph to find the desired node. Another problem was that if a user wanted to place a large number of static objects (such as numerous furniture) in a large building, the process took a significant amount of processing time. That was our motivation to develop the Object Editor as shown below in the figure below, Object Editor with a menu of 3D static objects.

Fig. 3. Object Editor menu

In order for the users to add new 3D static objects, all they have to do is to simply click the desired furniture and then click the position in world. Two actions take place by clicking the icon. The first one is to clone the node and add it to the scene graph. The second one is to activate the "Place Manually" procedure for the cloned object (figure below).

The only drawback in this method is that the user has to predefine the path of each furniture node by filling the following array (an xml file can be used for this purpose too) based on 3D meshes exported from a 3D Modeling tool via the game-engine export plugins.

Fig. 4. Place manually proceposition)

3.2 Editing Architectur

Another task that real-time control the overall materialby game engines, the userAlmost similar as before, th

1. Load editor mode 2. Select a surface 3. Browse through materi4. Press assign material b

By creating the Material Eclicking the material icon an

From Real to Virtual Rapid Architectural Prototyping

edure (object’s position is depended on the user mouse poi

ral Materials

architectures usually perform in the 3D environment il and color management. In a virtual world, which is br simply has to assign a material on the desired surfahe user has to:

ials (which can be hundreds) utton

Editor (figure below), we can assign materials by simnd then clicking the surface.

Fig. 5. Material Editor menu

509

inter

s to built ace.

mply

510 M. Tato, P. Papaniko

The main drawback in tuser's viewport. Again, sam"Ok" in the Material Editmaterial has been applied to

Fig. 6. The same

3.3 Instilling “Live” El

Object Editor and Materiexisting procedures, which equally important as visuamoving around the 3D virtu

This is why we have crwhich handles population aCurrently we only use age

Fig

olaou, and G. Papagiannakis

this method is that the surface has to be visible from me as before, the changes are registered only after clicktor Window. Fig. 6 illustrates the object after the no it.

e object after the selected material has been applied

lements in the Urban Environment

ial Editor are concrete enhancements from the alreaim for faster and simpler 3D world management. Wha

al realism is also ‘live’ elements such as people & cual world. reated the Crowd Manager as a new system componand their actions in the 3D environment via simple scrients (i.e. virtual humans) and virtual cars. We need

g. 7. Crowd manager dependency graph

the king new

eady at is cars

ent, ipts. one

From Real to Virtual Rapid Architectural Prototyping 511

different crowd object manager for each such entity. Subsequently we have to attach to each one a path manager, which takes responsibility for the movement of all moving entities of the same type. For example, agents choose randomly a path to follow, while cars follow a circular repeated direction with collision detection supplied by the 3D engine. The way each type of entity moves is totally depended on the path manager it chooses to apply. Each crowd object (agents, cars) is following its own path route, which has been assigned to them by the path manager. In case we want to change the movement behavior of a specific crowd object type, we create a different path manager and assign it to the corresponding crowd object manager. Fig. 7 depicts briefly the dependency graph of the crowd manager and in Fig. 8 we can see the impact of car and agent in the virtual environment of the scale model.

Fig. 8. 3D Car and Agent managers in action

4 Results, Conclusions and Future Work

The main system novelty in this urban heritage industrial system project was to answer the question on whether rapid prototyping based on advanced, serious-game 3D rendering is more suitable than rapid prototyping via 3D printing, given the same 3D architectural models as starting point. Our initial experiments and user studies point towards this direction, especially with the added value that it is relatively straightforward to export a 3D scene for real-time game-based simulation. The serious-game experience has then all the advantages of modern real-time simulation: environment modifications, 3D objects manipulation with several hypotheses testing, crowd-agents depicted in the real scene for added ‘live’ elements etc.

In this project the Unigine™ [5] game-engine was employed, their Autodesk 3DSmax™ export plugins and their scripting language to build the real-time demo interactive controls.

However, for a fast, non-accurate appreciation of a relatively small general area-space without any modifications of the environment or hypotheses testing, the scale model is still the most familiar to all user age groups according to our user study (mainly due to camera control in the virtual environment and the level of familiarization of users with virtual environments). Hence in the future we aim to bridge this “fast appreciation gap” by continuing UX tests with AR or VR HMD with

512 M. Tato, P. Papanikolaou, and G. Papagiannakis

head tracking in the real-time scene simulation. Finally using techniques as depicted in [4] we aim to also fabricate static 3D human figures from the existing rigged articulated characters that were created for the virtual 3D scale model.

References

1. Egges, A., Papagiannakis, G., Magnenat-Thalmann, N.: Presence and Interaction in Mixed Reality Environments. The Visual Computer 23(5), 317–333 (2007)

2. Slater, M., Khanna, P., Mortensen, J., Yu, I.: Visual realism enhances realistic response in an immersive virtual environment. IEEE Computer Graphics and Applications 29(3), 76–84 (2009)

3. Gibson, I., Kvan, T., Ming, L.W.: Rapid prototyping for architectural models. Rapid Prototyping Journal 8(2), 91–95 (2002)

4. Bächer, M., Bickel, B., James, D.L., Pfister, H.: Fabricating articulated characters from skinned meshes. Transactions on Graphics (TOG) 31, 4 (2012)

5. UNIGINE real-time 3D engine (game, simulation, visualization and VR) (2012), http://unigine.com (retrieved at June 25, 2012)