Re:Magic Leap is a Tragic Heap
by saltcured on Monday August 27, 2018 @01:19AM
(#17849584)
What do you think is "truly immersive VR"? The technical challenges for VR depend on what you think makes it "truly" work or makes you dismiss previous solutions. I think I experienced immersive VR over 20 years ago in the CAVE system, which had shutter glasses, head tracking with stereo projection on either multiple walls or a thing they called an ImmersaDesk, and a user-input device called "the wand." It used a multi-rack SGI Origin, which I am sure is outclassed by a modern gamer's PC. There were some artistic demos and many science/visualization application demos.
I saw these in multiple research labs, but I'm not sure they ever really served a real application base. They were in constant use by grad students and postdocs writing new experimental apps and demos. Unfortunately, I think real technical users often find that immersion is a gimmick rather than a real tool. You'd rather spend your money or compute resources on higher resolution and high dynamic range images than stereo rendering. And, when you have challenging visualization problems you are often struggling to sustain any interactive rendering and exploration at all, much less the high framerates needed for immersion.
You could have immersion at any time, if you are willing to adjust your problem size and expectations. Stereo rendering always means rendering twice as much as for a regular screen, so you are also going to require a larger hardware investment for any given application and quality metric. Natural, immersive navigation always means limiting your model such that the working set remains well cached and you don't have massive drops in rendering rate as the user moves. People get aggravated when ripped out of immersion by application stalls.
Also, it is hard to make an immersive display system low friction/accessible to casual use. But, if you have to design your application to easily degrade to normal screens and keyboard+mouse, then you cannot really focus on and exploit the compelling parts of immersive rendering and immersive user inputs. Can you ever find enough users to agree on this set of compromises to make a product successful?
I saw these in multiple research labs, but I'm not sure they ever really served a real application base. They were in constant use by grad students and postdocs writing new experimental apps and demos. Unfortunately, I think real technical users often find that immersion is a gimmick rather than a real tool. You'd rather spend your money or compute resources on higher resolution and high dynamic range images than stereo rendering. And, when you have challenging visualization problems you are often struggling to sustain any interactive rendering and exploration at all, much less the high framerates needed for immersion.
You could have immersion at any time, if you are willing to adjust your problem size and expectations. Stereo rendering always means rendering twice as much as for a regular screen, so you are also going to require a larger hardware investment for any given application and quality metric. Natural, immersive navigation always means limiting your model such that the working set remains well cached and you don't have massive drops in rendering rate as the user moves. People get aggravated when ripped out of immersion by application stalls.
Also, it is hard to make an immersive display system low friction/accessible to casual use. But, if you have to design your application to easily degrade to normal screens and keyboard+mouse, then you cannot really focus on and exploit the compelling parts of immersive rendering and immersive user inputs. Can you ever find enough users to agree on this set of compromises to make a product successful?