Welcome to hypography, Josua!
Please feel free to start a topic in the introductions forum
to tell us something about yourself.
A general comment on your post: I think you should avoid paying too much attention to the NerveGear
and other devices described in Sword Art Online
and other cyberpunk fiction, because ... well ... they’re fictional, and not intended to suggest in detail devices that are actually possible. When Reki Kawahara writes that the NerveGear read and writes to the brain using “high-density microwave transceivers” or that “30% of the NerveGear's weight is from its internal battery”, it’s just for literary effect, to make the story feel more realistic, not a technical suggestion to somebody trying to make a actual video game interface.
A specific comment:
Rendering would be a monster to deal with, aside from the CPU. Rendering the level of realism offered by the NerveGear is practically near-impossible with current generation GPUs. Because of this, the GPU needs not only to be compact & power-saving, it must also be powerful enough to render that much pixels.
I think you overestimate the difficulty of rendering pictures realistically enough that they are indistinguishable from reality.
A single high end GPU like an Nvidia GTX-1080
can rendering 33,177,600 12-bit pixels (8x UHTV resolution) at 60 frames/s. Test show that this resolution is nearly beyond the limit of human perception to distinguish from higher resolution, except for the detection of aliasing effects (like the “screen door” effect) due to the arrangement of the pixels on the screen.
The human retina
has about 120,000,000 rod cells, which cannot distinguish the color, and 7,000,000 cone cells, which can. The refractory period is about 0.004 s (analogous to a frame rate of 250 f/s).
So, ignoring that retinal cells are effectively 1-bit detectors – that is, they fire or don’t fire when stimulate to their threshold –a single present day GPUs are within a factor of 16 of rendering more data than the human eye can possibly perceive. It’s common place to use many GPUs in parallel (for example, ORNL’s Titan
uses 261,632). So I think we can reasonable say that were it possible to interface with every neuron in the human retina, we would not lack for computer processing power to stimulate them as quickly as they are capable of being stimulated.
It's said that it was due to the microwave transceivers that redirected all brain signals from the brain to the game with the device. However, as we all know, microwaves don't do something like that in our reality.
They are a form of electricity that could be used to stimulate the brain. It would be horrible and something such as Infrared is much better, but they can do what was described.
I think Josua’s statement is accurate. The “something” to which I think he’s referring, taken from Kawahara’s light novels, is
“The NerveGear's high-density microwave transceivers are capable of accessing the user's brain, allowing it to send fake signals to the five senses of the user. The transceivers are not only capable of inducing fake sensory signals but can also block every movement command from the brain to the body, preventing the player from moving their body while in FullDive to avoid injuries”
Unless your brain cells have had been ontogenetically
engineered to be receptive to it (in the visible 4.3 x 1014
to 7.5 x 1014
Hz, range), individual neurons can’t be measured or stimulated using EM radiation
of any kind.
Microwaves are not a form of electricity – a flow of electrons in a conductor – but EM radiation in the 3 x 108
to and 311
Hz range. Infrared EM radiation is in the 3 x 1011
to 4.3 x 1014
The genes available to ontogenetic to make nerve cells sensitive to and emit EM radiation are limited to those that sense and emit visible light, and lower-frequency EM radiation has proportionally lower resolution, so I doubt EM radiation lower than the visible range can be used to sense or stimulate individual neurons.
So, for the NerveGear to work anything like described in its fiction, its “microwave transceivers” would need to be replaced with “visible light transceivers”, and the user’s scalp, skull, and selective brain sections made transparent to visible light. Presently, the only known way to accomplish the first two of these is to remove the troublesome section of scalp and skull and replace them with something transparent, like a glass window.
If you give up on the requirement not to stick stuff through your skin and skull, and no messing with your somatic DNA, you can do optogenetics much more comfortably by implanted fiber-optic fibers in your brain. This is far from the neat, non-intrusive consumer product depicted in SAO, but at least it’s physically possible.