Reply to John B. Photogrammetric vs. laser:
Okay, I guess I can't escape this thread. It has become a curious hobby-- but only until July 30 when trial membership to SI expires.
Funny that you ask because I was also curious and I downloaded the research paper that Virtuoso is based on from their web site.
The short answer is that Virtuoso appears to be a mixture of laser, photogrammetry, and a very novel stereo disparity algorithm.
Actually, in a some sense stereo disparity is a form of photogrammetry because it involves determining 3D position by triangularization from features on two photos. However, in photogrammetry the features are marked deterministically (usuallty a laborous operation); whereas, in stereo the features are located along the photogrammetric epipolar line (line intersection on photo with plane defined by camera focal points and marked point) by attempting to match colors automatically within some distance along epipolar line. Stereo disparity has traditionally been an ill-posed problem that did not provide consistent results on real world objects. The Virtuoso stereo method is one of two recent ones (1996) I have seen that are extremely powerful and robust. Virtuoso finds the set of all possible 3D coordinates and then chooses the 3D surfaces with the highest area of continuity. In effect, it turns an ill-posed problem in a well-posed one where either the correct 3D surface is obtained or no data is obtained.
Virtuoso uses 6 sensor and a strip of laser light on the object to determine the relative orientation of the camera for each photo captured. In other words, they use laser method to automated the 3-pt manual orientation of Synthonics. Thus, their solution is highly automated and flexible (and expensive for now, $25K) compared to Synthonics.
Succintly, photogrammetry is much more laborous than stereo or laser. Virtuoso combination provides nearly the best characteristics of all solutions: freedom of object location and size, quality of photo-texture (photo appearance of resultant 3D model), speed and automation of capture and processing, quality and detail of geometry.
Whereas, Synthonics' technology apeears to be weak in: freedom of camera lens, speed and automation, and detail of geometry. The later two being common to all photogrammetric products-- not just Synthonics.
Besides, as I said, Synthonics' technology appears to be the weakest of the photogrammetric offerings in terms of camera lens flexibility, quality of software tools implementation (Wireframe Express can't hold candle to 3D Builder or PhotoModeler for me), and accuracy. However, apparently Synthonics has a rendering package (viewer) and is willing to do the labor, so they were able to win the Smithsonian evaluation contract. I think over time, it will become clear which photogrammetric product is the best and the other elements such as viewer are easily sourced from other parties. In fact, with VRML export and web browser is a viewer. I would think if Smithsonian was doing the labor themselves, they would look harder at which photogrammetric tool would save the most time, provide the highest quality, and most flexibility. They might even toss photogrammetry in favor of Virtuoso. Well, just my **guesses** based on some tid bits of info in grapevine. I have no overwhelming hard facts on the Smithsonian deal, except the analysis of Synthonics' technologies and software tools.
Thanks for your question. It is very rewarding to see that someone investigates to this level and asks a very relevant question. I think now Synthonics will either have to put up or shut up by the end of the year. I genuinely hope they have found my statements challenging and will be even more inspired to do it. It would be nice to see them compete in action instead of words and all you investors do reasonably well with this investment. I am not a Synthonics hater per se, just hate the past misinformation campaign.
Good luck in your other investments as well.
Henry |
