Back in 2008, we covered what looked to be an interesting development: a rewritable holographic display with a reasonable refresh rate and decent durability. Well, the people behind that work are back, and they've clearly been pushing their technology closer to commercial applications. This time around, they've built a display that can accept input that's streamed over an ethernet connection and reassembled into a hologram in near real time. The refresh rate is pretty slow, but it's a significant step forward from the static images they were displaying the last time around.
The paper itself reads (at least in part), like it was written with Ars readers in mind; the second sentence notes, "The concept of 3D telepresence, a real-time dynamic hologram depicting a scene occurring in a different location, has attracted considerable public interest since it was depicted in the original Star Wars film in 1977." In the intervening time, there have been a few stabs at what the authors call "holographic cinema," but these are just like regular films, in that they require heavy processing and can't be used to display live images.
The display technology described in 2008, which was rewritable, could potentially provide the opportunity to display dynamic content. But the authors spend most of a paragraph disparaging that tech (which we were impressed by at the time), and explaining why it doesn't make for a good telepresence medium:
The recording time was about one second per hogel [holographic pixel], and the scanning system (used to shift from one hogel position to the next) needed to be stopped and damped each time to avoid vibration, resulting in a time-consuming process. As a result, the overall recording time for a 4 inch hologram consisting of 120 hogels was of the order of 3 min. An erase time of 1 min was needed before refreshing the image. The system suffered from high sensitivity to ambient noise (such as vibration and air turbulence) and to thermal expansion, requiring a fully enclosed air damped optical table.
Not good. Most of these problems could be solved simply by increasing the recording rate; once it's fast enough, things like vibrations and air turbulence simply don't have enough time to cause a significant impact on the image. The first step in doing so was to simply speed up the laser. The new paper uses a laser with pulses in the nanosecond range, and each pulse is split into enough separate beams to record 100 hogels with each exposure. The laser can also cycle at 50Hz, which increases the number of hogels that can be written per second.
The recording media relies on the same principles as the original one: the laser creates areas with a charge within an organic polymer, and these areas refract light differently, allowing the formation of the sort of complex interference pattern that creates a hologram. The original medium, however, was optimized for longer durability; the holograms would last for several hours without refresh. That's simply not needed when the image is going to be overwritten in a second or two anyway, so the new material was optimized for sensitive exposure to short laser pulses.
The authors first demonstrate that they're able to do what they call "angular multiplexing," to create multicolor holograms. In this case, different exposures are done simultaneously using lasers at slightly different angles and with separate polarizations, which ensured that they did not interfere with each other.
But the killer demo is the telepresence. Obviously, there are safety issues involved with exposing subjects to lasers, so the imaging was done with regular cameras—16 of them, all using Firewire to provide something close to real-time performance. These were sent across ethernet to the display computer; the authors indicate that even 100Mbps has plenty of capacity to spare. At the receiving end, a desktop class computer reconstructed these into a 3D image, and used that to control the laser that encoded the image into the display media.
We're not quite ready to see Princess Leia emotionally plead for help, given that the refresh rates are still a couple of seconds between frames. But the work at least demonstrates that the general approach is flexible enough to handle both long-lived displays and relatively rapid refresh, so it's possible that a bit of further tweaking would improve its performance for one or the other of these, or get it to do something entirely new.
Source: ars technica
