More details emerge about MIT Labs glasses free 3D display

Yesterday we reported that MIT Labs have created a glasses free 3D display prototype that allows multiple viewers to see a 3D image without glasses. There are no viewing zones and it is akin to holographic television in that the viewer can ‘look around’ objects as he or she moves their head.

We spoke to Douglas Lanman, Postdoctoral Associate MIT Media Lab, to find out more details.

3D Focus: What is the viewing angle for your prototype?

Douglas Lanman: The viewing angle for our three-layer prototype is 20 degrees, both horizontally and vertically (we support full motion parallax, not just horizontal-only modes). The viewing angle for our two-layer prototype, which includes a lenslet array between two LCD panels, is 48 degrees, both horizontally and vertically. One of the benefits of our latest work is to attain a large field of view using refractive lenslet arrays, but restore the resolution loss typically seen with lenticular displays.

3D Focus: What is the benefit for increasing the layers and what is the most amount of layers you have tried?

Douglas Lanman: Our latest work supports a sort of "Franken-Dispay" consisting of an arbitrary number of stacked high-speed panels, illuminated by any low-resolution multiview display. For example, our two-layer display can be thought of as a classic lenslet display (i.e., a 2D lenslet array glued to an LCD panel), covered with a second semi-transparent LCD. In this case, adding the second LCD panel allows the resolution to be enhanced, compared to conventional lenticular displays. If the panel is fast enough (say 360 Hz), then the resulting images will be nearly free of artefacts and appear at high resolution. However, if only 120 Hz panels are available, similar image quality cannot be obtained with just one additional layer. Yet, we show that (say) two 120 Hz panels can achieve high fidelity.

In summary, we show a single optimization framework that can trade-off the complexity of the display (i.e., the number of layers and their frame rates), for image quality, brightness, and other stereoscopic quality metrics (e.g. depth of field or field of view).

3D Focus: How is the video encoded? Have MIT created a processing algorithm?

Douglas Lanman: Yes, we have not only introduced the general display hardware, but also an "encoding algorithm" that determines the best patterns to display on all the layers to recreate the illusion of the 3D scene perceived by the viewer. The title of our publication is "Tensor Displays"; this is because the encoding algorithm utilizes multi-linear algebra, in which the images seen from different perspectives are encoding within a tensor (which is just a multidimensional array of numbers). While the mathematics is beyond the lay audience, the take-home message is that our work is very much focused on the mathematical encoding algorithm; specifically, we show how to choose the patterns to display across the layers to maximize brightness, resolution, and perceived image quality.

The patterns are not "hand-tuned", but are optimized using this formal tensor-based mathematical compression algorithm. This is one of the key differences with current commercial systems, like the 3DS, which use heuristics that lead to limited brightness and resolution when scaled to wide field-of-view television systems.

3D Focus: Have you tied 360 hertz panels yet or are they not available?

Douglas Lanman: Unfortunately, 360 Hz panels are not yet available. In our video we set the camera exposure to be slightly longer than the human eye, so that our 120 Hz panels can simulate 360 Hz display. Yet, our method can work within the constraints of current 120 Hz LCD panels. In fact, at SIGGRAPH, we'll be showing a working prototype using a stack of three 120 Hz panels. The images appear high-resolution and span of field of view of 20 degrees horizontally and vertically. While the images appear with some slight artefacts (similar to compression error you see in jpegs or DVDs), 120 Hz panels can work now, especially if you allow a third layer as we do with this demo.

FREE WEEKLY 3D NEWS BULLETIN –