Boffins render full HD million-point animated hologram
One step closer to astromech droids?
Scientists in Japan have built a computer capable of generating holographic images from three-dimensional models composed of up to a million points.
But don't expect holographic messages from alluring princesses any time soon: the machine can only render million-point holograms at a single frame every second.
Drop the complexity of the image to 100,000 points and the frame rate rises to 10f/s, "near video rate", the team from the Tokyo University of Technology, Chiba University and the Kisarazu National College of Technology said in a paper published by the Optical Society of America.
Not Princess Leia
The system pumps out images with a resolution of 1920 x 1080 - full HD, in other words. The image is dislayed as an interference pattern on a reflective LCD. Laser light is passed through a beam splitter to send half the light on to a viewing lens and the rest to the LCD. The light reflecting off the LCD interacts with the first split beam to generate the hologram.
Calculated and render successive interference patterns and the hologram becomes an animated one.
Creating the interference pattern on the LCD requires some hefty processing power. The team developed a board for the task, called Horn 6, and the rendering system contains 16 of these. Each Horn 6 contains four off-the-shelf Xilinx Field-Programmable Gate Array (FPGA) chips for image calculation.
The Horn 6 processing board
The 16 cards are fitted four at a time in four PCs linked on a Gigabit Ethernet network. A fifth PC, also networked, manages the whole operation.
The team said the next step in their research programme will be to up the system's processing power to allow them to run real-time million-point hologram animations. They also hope to be able to incorporate calculations to hide parts of the object that should be occluded by foreground elements.
Now make occluded parts disappear
You can download the full paper (PDF) from the Optical Society of Amerca here. ®
COMMENTS
Re: Proper Hologram from LCD
Ah, but if you're only aiming or 5 degrees of viewing angle, then your Nyquist wavenumber gets multiplied by sin(5 deg), I guess. With a factor of two in there for good measure, I make that something like 700nm / (2 sin(5 deg)), or about 4 microns per pixel. Still quite small, though. It probably doesn't work like that though, or there's be all sorts of diffraction problems.
This is probably why holographic television requires more than just a bunch of us semi-informed commenters to invent it.
Proper Hologram from LCD
The wave length of red light is around 700nm. A 1920 by 1080 LCD for interference patten would be 1.3mm by 0.76mm.
I would imagine you would have to ray trace the light from a virtual laser illuminating the model to the virtual screen pixel, measuring the length of the trace in wavelengths and adding to all the other traces hitting that pixel.
With such a tiny screen, you would need a magnifying glass to see the hologram.
I would imagine it would be easier to use UV photo-active or electro-active film/glass written with a UV or electron beam and illuminated with a red laser. A 5cm cube would require 71,429 x 71,429 pixels or 5G pixels.
Re: Occlusion
Occlusion in a hologram is going to be a huge issue compared to regular (flat) 3D, since you can't cull an object's hidden geometry based on the viewing angle (of the observer as opposed to the camera).
So it's just another projected camera trick to give depth. I want to see genuine "walk-around" holograms. This stuff is boring.
Occlusion
There seems to be quite a bit of confusion here about what the word "occlusion" means - esp. relating to CGI.
Every object model has depth - even standard 2D gaming objects. This is because objects are manuipulated as 3 dimensional objects in "world space" inside the computer and then transposed into a set of 2 dimensional co-ordinates for display on screen et al.
Occlusion is always required in CGI (and holograms) because in the same way I can't see the back of your head when I'm looking at the front in the real world, neither would I want to in CGI.
Therefore depth calulations are always needed to firgure out the viewing angle to the faces of the object that can be seen, and thus "occlude" the faces that can't.
Well said Chris
We've got the standard problems with a new technology here. Hologram technology is theoretically better than stereoscopic technology, but because it's at such an early stage of development, the advantages are not obvious, and it appears worse than the old technology. The viewing angle might be restricted, but that is not a fundamental feature of hologram technology, whereas it is with stereoscopic viewing.
This is star-wars technology. I was amazed that it was even possible to create any kind of interference pattern from an LCD display. Certainly, the technolgy to create a display that could render a the full interference pattern for a full TV picture must be a long way off, but it is great that someone is researching this stuff.
