Dynamic Image Based Lighting for Highly Realistic Lighting in Building Design
| industrial collaborators: | Arup (Ove Arup & Partners Ltd) |
| academic collaborators: | University of Warwick |
| initiated : | 2009/10/03 |
| last updated: | 2010/04/26 |
 |
 |
Project staff and support
Francesco Banterle (Intern, University of Warwick)Steve Walker (Company Supervisor, Arup)
Alan Chalmers (Academic Mentor, University of Warwick)
Lorcan Mac Manus (Technology Translator, Industrial Mathematics KTN)
This Internship project was carried out at Arup, in conjunction with the University of Warwick. It was part of the KTN's Industrial Mathematics Internships Programme, co-funded by EPSRC. Start date: October 2009; duration: 4 months.
Overview
The natural world presents the human visual system (HVS) with an ever changing and wide range of light, colours and intensities. For example the difference in average luminance level between a starlit night and a day-lit scene spans eight orders of magnitude or 100,000,000:1. The HVS can perceive detail in regions where the light intensities vary dramatically - up to a contrast around four orders of magnitude or 10,000:1. For example, a human eye can see, at the same time, dark areas inside an office, and bright sky outside windows. However, adaptation mechanisms kick in when the view changes from very dark to very bright regions due to limits of the HVS. Existing video cameras are only capable of capturing a limited part of this wide range of light, colour and detail with sufficient resolution.
High Dynamic Range (HDR) imagery is a novel way to capture, store, process, and visualise pictures and videos. It allows the representation of all real world luminance levels without limits in terms of quantisation and range. For example, when a picture of an indoor scene with bright light coming in through a window is taken using conventional imaging, this picture will show only the interior of the room clearly with a blank window or, conversely, the scene through the window will be clearly depicted with a dark interior. HDR imaging allows the capture of all radiance information within the scene, from the interior room detail to the bright day-lit landscape outside the window, due to the fact that the range is not ‘clamped’ and quantisation of light levels is encoded using a representation of real values.
Until now the capture of HDR images with a linear response function has been limited to static scenes rather than ‘moving images’. In July 2009, the University of Warwick took delivery of the world’s first true HDR video system. The system comprises a unique HDR video camera capable of capturing 30 frames-per-second of full High Definition television (HDTV) video stream at a resolution of 1920 × 1080 pixels.
Image Based Lighting (IBL) is a technique for artificially re-lighting real world or synthetic objects. The project sought to develop a new mathematical approach to IBL to cope with the dynamic nature of the lighting captured by the HDR video camera. Standard IBL uses only static information from a 360 degrees panoramic HDR image, assuming that light sources are located at an infinite distance. In the new approach, we wanted to exploit dynamic information from film clips recorded by the HDR video camera to capture real world conditions, where light sources change over time. This allows the modeling of time-varying high-fidelity environments. Another area we wanted to research was how to exploit the spatial information in a video stream to generate spatial light sources not assumed to be placed at infinite distances (which causes no attenuation of the light and long shadows), but to have physical properties as in the real world.
Additionally, the project sought to integrate accurate dynamic real world lighting captured from this HDR video system into Arup’s existing visualisation system in order to significantly improve the fidelity of images used for building design. This integration included the production of a working ‘pipeline’ that connects Rhinoceros3D (a NURBS modelling package) and a physically based renderer (PAR) developed at the University of Warwick. This involved extending the PAR in order to be compatible with Rhinoceros3D. Furthermore, an input/output (I/O) mechanism to connect them was developed.
Outcomes and benefits of the Internship
An important achievement of the project was the production of a series of algorithms for re-lighting virtual objects using HDR videos.
“The main benefit of the internship was the opportunity to work on a rendering system in a real-world situation outside the academic software-developing world where important, practical details are not always considered. For example, interoperability between 3D modelling packages and a renderer is usually not seen as a high priority task during the development of a renderer in an academic context whereas in an commercial or industrial situation this is a crucial aspect.” said Francesco Banterle, the intern for the project from the University of Warwick.
"I felt we had set ourselves very ambitious goals for such a short timescale project. I was impressed with the way that diverse research and knowledge at the University of Warwick was quickly brought to bear to resolve challenges as they arose, and with their readiness to respond to our, more application driven, priorities for the overall project. This experience has given us excellent insights into the range of knowledge and capabilities at University of Warwick and we are excited about future research work and collaboration with them", said Steve Walker, the Industrial Supervisor at Arup.
“This Industrial Mathematics Internship has enabled leading edge academic research results to be transferred to solve a challenging real industrial problem. To successfully achieve the short-term focus of this internship has necessitated close collaboration between the academic and industrial partners. This has been highly beneficial for both. The University of Warwick has gained significantly from exposure to Arup’s internationally leading industrial research and development. We look forward very much to working with Arup on many other exciting projects in the future”, said Alan Chalmers, the Academic Mentor at the University of Warwick.