I don’t know all the nuances of the problem, but from my understanding, it’s due to the how the waveguide works. A waveguide points light in a certain direction, usually through rectangular, reflective tubes. In order to support internal reflections inside that waveguide, Microsoft has to limit the exit angle to 35 degrees, which impacts the FOV.

Splitting Waveguides

The Redmond giant has been struggling to find a solution to this problem, with executives previously saying its unlikely to get much better. Now, thanks to internal research, Microsoft has managed to double the current field of view and publish a patent describing the system.

Microsoft is splitting image to the waveguide into two parts for each input pupil, resulting in two separate exits. Though both have a limit of 35-degrees, together they put the diagonal field of view up to 70. With the proper design, its inventors believe a diagonal FOV of up to 90-degrees is possible. Crucially, the tech has been tested on hardware already, meaning it could make its way into the device’s next iteration. The question now is whether 70 or even 90 degrees, is still enough. The HTC Vive, in comparison, has a diagonal FOV of about 145 diagonal degrees, while the Rift sits at 120. The difference even between the two is noticeable, so the HoloLens may still be quite jarring in comparison. Despite this, it’s a big step forward in HoloLens development and very welcome news. After hearing the Windows 10 Creators Update won’t reach the device, it’s great to hear some positive progress.

Microsoft Has Finally Found a Solution to the HoloLens  FOV Problem  but Is It Enough  - 10Microsoft Has Finally Found a Solution to the HoloLens  FOV Problem  but Is It Enough  - 36Microsoft Has Finally Found a Solution to the HoloLens  FOV Problem  but Is It Enough  - 15Microsoft Has Finally Found a Solution to the HoloLens  FOV Problem  but Is It Enough  - 32Microsoft Has Finally Found a Solution to the HoloLens  FOV Problem  but Is It Enough  - 66Microsoft Has Finally Found a Solution to the HoloLens  FOV Problem  but Is It Enough  - 53