Color information on 3D models
Although 3D scanning and the other geometry acquisition
techniques have been incredibly improved in the last few years, there
is probably still some space to work on the possibility to add highly
detailed color information to highly detailed geometric information.
The broad issue is: starting from a 3D model and a set of uncalibrated
images, can we produce a realistic colored mesh? And (more important
for me) can we reach this without the need of expensive devices,
complex lights setup or time-consuming post-processing? And the answer
is... "Well, we'll see!"
During the last few years I've been working quite a lot in this field: one of the most useful things I collaborated in creating was TexAlign, which is a semi-automatic tool for the registration of lots of images on a 3D model. Texalign was used in several Cultural Heritage related activities, like the Cenobium project, the restoration of Michelangelo's David or the peculiar small work on the Bronze Horse.
Finally our idea about fully exploiting the potential of built-in digital camera flash has been accepted (it was presented at Mirage 2009, Paris, and at VAST 2009, Malta).
A very promising work in the context of automatic image registration has been accepted at Pacific Graphics 2009. Currently I'm working at a method to correct the small misalignments of calibrated images, in order to obtained more detailed colored models. If everything goes well, we will know something about it in the first half of 2010...
Crossmodal effects and HRTFs
In the period between the end of 2005 and the end of 2008 I've
been involved in the
EU project Crossmod.
aim of Crossmod was to study and exploit in CG applications the
crossmodal effects, which are the effects of visual perception on
auditory perception and viceversa.
The idea was to study the main effects, find the best ones for
implementation in CG applications, and build a "crossmodal manager" able
to use these effects to improve realism and performances. This project
was brought on with partners from several countries in Europe, and it
involved both CG and neuroscience experts. The whole thing proved to be
very hard to address and exploit, but in the end I think we had quite
good results, although we would need at least other 3 years of work to
exploit all the initial
Anyway, in the end the contribution of our group was mainly in the field of HRTF calculation. Head Related Transfer Function mathematichally describes the perception of a sound by a human ear, in relation to its frequency and position in space. This function can change a lot between subjects, it's associated to the shape of the ears and (partly) of the head features; it's very important to have accurate HRTFs to best exploit the potential of 3D sound rendering. We collaborated with the colleagues at Reves Lab, INRIA Sophia Antipolis: the first work was on an algorithm for fast sound scatering calculation, then my colleague Nico Pietroni and I worked hard to find an automatic way to build 3D heads from a few photos (plus very simple input by the user). The work was accepted and presented at Pacific Graphics 2008. Now we'll see if the large scale application of this stuff will be possible in the near future...
Last but not least, in the last few years I has the chance to
work a lot in the field of 3D scanning. Although nowadays 3D scanning
is not a very active field of research, I really like this kind of
activity, and I find it very useful and stimulating.
I especially think that the scanning campaign are always a very useful
exercise of practical problem solving: every one is different, and the
objects need always
different strategies and approaches. Moreover, I find that 3D scanning
has a natural application in Cultural Heritage, and very
interesting and useful things can be done in this field. I had the
chance to visit incredible places, like
Delos and the
Most of the 3D scanning papers I co-authored are related to Art archiving and preservation: there are very few places like Tuscany with such density of interesting stuff! Unfortunately, working on them covers two fields of work which are among the least funded by italian state: research and cultural heritage!
Nevertheless, I think that working with people who had different education paths can be hard but very useful and stimulating. Several of Visual Computing Lab works come from scanning activity: for example, the Cenobium project on Monreale cloister, the impressive Portalada scanning and the promising project about the Luni temple fronton.
Last year I was in Cartagena for a short Scanning campaign. In march there was a very interesting experience in Cefalù, then the double mission to Rome. What's next? We'll see, maybe at the other side of the world...