Interactive Headight Simuation P. Lecocq, J-M. Keada, A. Kemeny Renaut Direction de a Recherche - Research Department Technocentre Renaut TCR AVA 2 12 1, av du Gof 78288 Guyancourt Cedex France Phone : +33 1 34 95 19 85 Fax : +33 1 34 95 27 30 E-mai : Andras.Kemeny@renaut.fr Abstract Headight simuation has been recenty studied for the conception of a new car projector by automotive car makers and suppiers. It aows the study of a suppier's numerica soution without making a physica prototype. Unfortunatey, simuations are difficut to impement in rea-time, computed by ray-tracing agorithms [1,2] or radiosity techniques [3], and are often visuaized on a sma screen. Consequenty, the operator can sedom test his simuated projector in driving conditions. In order to have an interactive simuation, computer generated images need to be rendered in rea-time. The rea-time constraint impies the use of graphics hardware engine and optimized ighting modes. However the choice of the ighting mode is critica because of the necessity of high-eve image quaity in rendering the ight distribution of the projector. Renaut, Direction de a Recherche, has recenty deveoped such interactive headight simuation software. Severa parameters can be set in rea-time, such as the position and the orientation of the headamp. It is aso possibe to provide photometric simuated measurement data on the road during the simuation. The ight distribution of the projector is taken into account according to its photometric description. More precisey, a projector is assimiated to a positiona ight source that deivers correct attenuation. The photometric description (in Lux) of the headamp is encoded in a texture [6] which moduates the effect of the ight source on a poygona surface. Thus, the precision of the simuation directy depends on the eve of precision of the photometric description and the eve of detai of the scene in term of poygons. The headight simuation software was integrated in Renaut s driving simuators in 1998 and vaidated by Renaut s photometric aboratory. Résumé La simuation d écairage est utiisée depuis peu par es compagnies automobies et équipementiers pour étudier et concevoir de nouveaux phares. Ee permet d étudier des propositions de phares sans a fabrication de prototypes physiques. Maheureusement, a pupart des simuations, utiisent des techniques de ancé de rayon [1,2] ou de radiosité [3] difficies à impémenter en temps rée. D autre part, es simuations étant souvent visuaisées sur de simpes moniteurs, opérateur ne se trouve pas immergé dans e poste de conduite rendant pus déicate évauation du projecteur. La contrainte temps rée impose utiisation de matérie dédié à affichage graphique et utiisation de modèes d écairage appropriés. Cependant, e choix du modèe d écairage est critique en raison du degré de précision nécessaire pour rendre a distribution photométrique d un phare. Renaut, Direction de a Recherche, a récemment déveoppé un ogicie de simuation de phare interactif. Pusieurs paramètres peuvent être modifiés tes que a position et orientation des projecteurs. I est égaement possibe de prendre des mesures photométriques sur a route durant a simuation. La distribution umineuse du projecteur est prise en compte à partir de sa description photométrique. Pus précisément, un phare est assimié à une source de umière ponctuee déivrant atténuation sur a route. La description photométrique du projecteur est ensuite codée dans une texture [6] moduant effet de a source ponctuee sur es surfaces. Ainsi, a précision de a simuation dépend directement du niveau de précision de a description photométrique du phare et du niveau de détai de a scène en terme de poygones. Le ogicie de simuation de phares a été intégré dans es simuateurs de conduite de Renaut en 1998 et vaidé par es experts en écairage chez Renaut.
INTRODUCTION The design of a new car headamp impies the conception of severa prototypes and night tests according to the evoution of car stying. These prototypes and night tests are required for testing the performance vaidation of the amp in driving conditions. Severa criteria are studied during these tests: the ight distribution on the road, ighting homogeneity, far visibiity. These tests are carried out a high number of times with the contribution of professiona testers. Renaut s Research Division has deveoped a rea-time ighting simuation too integrated in a driving simuator. This too aims at reducing design cost and deay of new headamps with esser number of prototypes and night tests. In this paper, the headight simuation software is presented. First an overview of computer graphics techniques for ighting simuation is given, foowed by the description of Renaut s interactive ighting simuation software. Techniques empoyed to simuate in rea-time headamp s ighting distribution on a road are detaied and the vaidation process is described. OVERVIEW OF COMPUTER GRAPHICS TECHNIQUES Computer graphics techniques for ighting simuation have evoved together with computer technoogy. The progress has aowed to provide ighting iumination modes more and more compex and physicay correct. Now the ight transport and interaction physic aws are we known and are appied in rendering methods ike ray-tracing[1,2], radiosity[3] or Monte- Caro[2] techniques. Unfortunatey the impementation of these agorithms into rea-time simuation appications is difficut. In parae, the evoution of dedicated graphics hardware workstations aows the rea-time rendering with a high number of poygons, hardware impemented ighting modes and texture mapping. Therefore muti-pass agorithms has been deveoped to take advantage of these capabiities in order to provide rea-time compex ighting rendering. The headight simuation software uses these agorithms.
THE INTERACTIVE HEADLIGHT SIMULATION SOFTWARE The Renaut s interactive headight simuation software a part of the driving simuation visuaization software. Overview of the driving simuator configuration Renaut s driving simuator [4,5], for ighting simuation is composed of a rea vehice cockpit. Different modues communicate together to compute and render driver s actions in the cockpit as iustrated in figure 1: Dynamics mode Graphics Generator Figure 1: Driving simuator configuration The visuaization generates images at 30 to 60 Hz for a forward fied of view of 150 and rear view mirrors. Experimentation using ighting simuation today is imited to a mono-channe configuration due to rendering compexity and the imits of the used one pipeine SGI InfiniteReaity hardware. During the simuation session, the driver has the possibiity to switch from ow beam to high beam from the cockpit in the same way he or she woud do in a rea car. In addition, unike in a rea car, the driver has aso the possibiity to switch from one ight beam design to another. The iumination mode For rea-time simuation the foowing constraints have been introduced in iumination mode: - Headamp is assimiated to a white point ight source with non-homogenous uminous intensity distribution. - Light reemission of surfaces is not taken into account. - Surfaces are diffuse. Headamp ϖ r θ Observer Road surface Figure 2: Non-homogenous point ight source
Thus the iuminance coming from the headamp reaching a surface can be written according to the Lambert s aw as foows: I( ϖ r ).cosθ E = (1) 2 r with: E : iuminance (Lux), I( ω ) : Intensity of the ight source in direction ϖ (Candea), θ : the ange between the surface normae and ϖ, direction of incident ight. r : distance between source and surface point (m). Therefore, the observed uminance is: ρ. E L = (2) π with: L : uminance (Candea/m²), ρ : surface refectance (constant for diffuse surfaces), E : iuminance (Lux). The main task is the modeing of the ight source uminous distribution. Due to the compexity of a headamp, it is difficut to provide a mathematica mode of the uminous distribution. Photometric headamp description A car amp is generay composed of severa mirrors and striated gasses producing a given ight distribution. For the sake of convenience, ighting simuation characteristics of the amp are taken from measurements of the exit iuminance of the amp. There are two manners to obtain measures of the exit headamp iuminance. A physica headamp is paced onto a goniometer, and a photometric ce paced at a given distance measures the iuminance coming from different directions given by the goniometer. Limitations: Measurement of the outgoing iuminance is not sufficient to address the entire characteristics of a amp. Due to the structure of the headamp, mutipe refexions can occur inside the amp. The amp therefore acts as mutipe sma ight sources that produce ight emission with different directions as iustrated in Figure 3. However, the uminous distribution is considered to be totay formed at the considered observation distance.
Measure direction Head amp Mutipass rendering agorithm Figure 3: ight path inside the amp For a satisfactory rea-time ighting providing the required ight distribution on the road, accurate ight and surface characterisation has to be done. Using hardware impementation of a muti-pass rendering technique of projected textures, the ight distribution is defined by I r r ) = I. F( ω ) with 0 F( ϖ ) 1 (3) ( ϖ max where F( ϖ ) is the normaized uminous intensity distribution. The normaized uminous intensity distribution is directy encoded in a texture. Therefore, the ighting on the road can be seen as the resut of the ighting of the homogenous point ight source, in a first rendering pass, moduated by the projected texture in a second rendering pass (figure 4). Projected texture Point ight source Texture generated from photometric measurements Figure 4: Muti-pass rendering with projected texture Constraints on the visua database description The ighting pass must be performed using a convenienty defined road samping eve-ofdetai. In fact, as ighting cacuations are done at each vertex in the database, the resuts depend on the database samping. On the other hand, the use of a too high eve-of-detai in database description impies a arge coection of poygons resut in sowing down image frequency.
A compromise soution reies on the definition of eve of detais on the road. Leves of detai provide more or ess precise poygon description of the road as iustrated in Figure 5. PERFORMANCES Figure 5: Definition of eve of detai for road description The headight simuation software runs on an Onyx InfiniteReaity with one pipe and two R10000 processors (250 Mhz). Two projected textures are empoyed during the simuation (one for each headamp). The software runs between 20 Hz and 30 Hz frame-rate according to the vehice position in the database. THE TOOLS Different user interfaces and periphera devices are integrated in Renaut s driving simuator to take advantage of the simuation. - The first set of too contros the headamp positioning through a Graphica User Interface. The foowing parameters can be changed in rea-time: Left-right and front-back head amp positioning Atitude of amps Pitch of amps Space between amps Consequenty it aows an easy positioning the amps anywhere on the car to simuate for exampe fog amps. - The second set of toos aows the use of different points of view through user defined camera positioning. New cameras can be added and manipuated during the simuation providing for exampe aeria point of view for the observation of the ight distribution on the road.
- The third set of toos provides simuated measurements. First a graduated grid in front of the vehice is appied on the road. It aows estimating ight distribution and range. Moreover the user has the possibiity to take photometric measures (in Lux) anywhere on the road with a mouse connected to the workstation. VALIDATION The vaidation of the headight simuation software was done through comparison with rea car headamp iumination. Severa headamps have been used to vaidate the simuator. Physica measurement and computed measurements for the photometric description have been compared. Renaut s driving headight simuator has been vaidated in 1998 by Renaut s photometric aboratory. CONCLUSION & FURTHER WORKS The headight simuation too provides the necessary information to compare severa headamp performances. It aows the reduction of the number of prototype and night tests. Further works on the ighting simuation software wi consist in the improvement and compet the ighting modes according the foowing objectives: - To provide precise characterization of refectance of surfaces - To take into account ight source coors - To simuate different weather conditions simuation (fog, rain, snow, ) - To generate booming effect simuation Moreover, the use of PC based soutions for rendering the ighting simuation is investigated.
REFERENCES [1] Kemeny, A., Synthèse d images fixes et animées, Techniques de ingénieur, Paris, 1987, 5530, pp. 1-21 [2] A. Gassner. Principes of Digita Image Synthesis, Vo 1 and 2, Morgan Kaufman Pubishers, Inc, 1995. [3] Michae F. Cohen and John R. Waace. Radiosity and Reaistic Image Synthesis. Academic Press Professiona, Inc., first edition, 1993. [4] Kemeny, A. A cooperative driving simuator Proceedings of the ITEC Conference London, 1993, pp. 67-71. [5] Kemeny, A., Automotive driving simuators: technoogy, appications, products, Proceedings of the RTS 94 Driving Simuation Conference, Paris, January 1994, pp. 61-67. [6] M. Sega, C. Korobkin, R. van Widenfet, J. Foran, P. Haeberi. Fast Shadows and Lighting Effects Using Texture Mapping. In Proceedings of SIGGRAPH 92, pages 249-252, 1992