Someone who influenced my own interest in multimedia innovations over the last 15 years is Irwin Sobel, a senior researcher at HP Labs. Irwin should enjoy retirement since a few years but he carries-on working on his passions, attempting to be what he calls the good conscience of research projects he contributes to at HP Labs or follows at Stanford University. He has always been kind enough to give me small tutorials and I also learnt from discussions with him an approach of research, mixing detailed analysis of problems over several years with a pinch of stubbornness or rebellion. What's constant with these projects is that they seems luxurious and far-out initially, but given Moore's law and progress in algorithms and peripheral, they appear reasonable 5 years later and commonplace 5 years later.
From 3D medical imaging
On my last visit, we strolled on the grounds of Stanford Medical University and recalled our first project in 1992. I had initiated an eclectic stream of research covering tele-radiology, web-based medical records and the emerging field of image management (or picture archiving systems ; PACS). Irwin and my friend Dr. Davide Caramella at the University of Pisa convinced me to add a 3D project. This included a 1-year sabbatical of Irwin at HP’s center in Pisa, to investigate visualization system to facilitate surgical-planning using multi-modal magnetic resonance imaging (MRI). MRI measurements generate 3 images (T1, T2 and Proton Density) and false-color technique allows the visualization of a composite image. The results were shown in 1993 at RSNA and allowed holographic-like virtual explorations of a 3D thorax with cardiac MRI scans. Students from Francesco Beltrame at the University of Genoa helped develop the software on HP's most powerful RISC-processor. RSNA 93, was also important as we participated to the launch of the 1st medical imaging standard, DICOM with interoperability with 20 medical imaging vendors . But from that year I remember mostly project reviews in Pisa under the natural colors of Tuscany.To telepresence and collaboration
More recently I followed his work on immersive visualization. In 2002 he showed me Coliseum, a teleconferencing system, involving 5 cameras, that rendered arbitrary-perspective views of participants giving the illusion of interacting in a common space. In 2005 I met him at a meeting of the AMI project, investigating multi-party interaction. (I commented on later results of AMI in February). At my next visit to Palo Alto, I sat in an early prototype of the Halo collaboration studio that HP was developing with DreamWorks. I looked at the video processing equipment, mostly professional TV equipment, probably the influence of DreamWorks. The challenges of telepresence ranged from accurate calibration of screens and cameras to improving eye contact feeling. .
Irwin is still working on this last problem using 3D video cameras that allow perform 3D facial extraction that can then be oriented towards the video conference participant.Having fun with immersive game boys
In 2007, Irwin worked on photometric and geometric calibration for the Panoply project, an immersive visualization using stereo projection on a concave screen that covers the field of view. The demo featured a F1 racing simulation game played from a car seat, steering wheel, stick shift, and pedals with major force feedback on the steering wheel and acceleration vibrations on the seat. It's so immersive I was quite stressed and, like most beginners I experienced the off-track flying experience serveral times !
