5 ms·

Everyone says that a powerful GPU is needed to drive this thing, but I don't think that's a must. Careful art direction based on current hardware limitations ca

by naoru 9y ago

Everyone says that a powerful GPU is needed to drive this thing, but I don't think that's a must. Careful art direction based on current hardware limitations can produce nice-looking virtual worlds — not photorealistic, but still immersive and fun. Nintendo pulls this thing consistently.

bni 9y ago
This is so true. Case in point: Brass Tactics uses cartoonish "game of thrones intro" graphics that is quite optimized. A current graphics card can even render the game world from a different overhead view on the monitor, with capacity to spare.
phkahler 9y ago
I would love to walk around in a VR world that looked like the old vector displays. Slightly glowing wire frames but with occlusion. This technically doesn't even need polygon rendering, though doing so with mostly black textures except the edges might be cool. At todays resolutions I think small cylinders along the edges of flat black surfaces might look better. And that's just one stylized world that doesn't require a mammoth GPU to handle.
kalkut 9y ago
I absolutely agree ! I do see the appeal of photorealism but lots of excellent 3D experiences are far from photorealism. A sensible art direction can go very far, and I'd even argue that a less than ideal graphical experience can be forgiven if the content is meaningful and engaging enough. I think we all have examples of this in mind
bhouston 9y ago
> Everyone says that a powerful GPU is needed to drive this thing, but I don't think that's a must. It is not a must, but it is a mostly. Most people want something approaching realism by default. Stylism is okay but it isn't the primary use case. The video game sales show this. Top selling games are mostly ultra realistic, Witcher, Call of Duty, Modern Warfare, Battlefront, etc. Yes there is Nintendo's low end strategy but most are ultra realistic.
mbell 9y ago
You also don't need to render at full resolution. A significant benefit of higher res VR displays is not being able to physically see the pixels / screen door effect of the screen. You still get this benefit by rendering at a lower resolution and upscaling to the display resolution. Certainly you will be able to see 'render pixels', but that's a much less significant problem and just reduces the realism of the scene.
- remcob 9y ago
  If you add eye/focus tracking you can selectively render the part in focus in a higher resolution and use lower resolution for peripheral vision. The end result is that you are rendering at exactly the bandwidth that the human eye can handle, which is optimal and probably not that high. Going forwards, this is IMHO where the real optimization gains are.
  - piracykills 9y ago
    That's a really interesting idea, but can you imagine how incredibly annoying it would be if there were any level of latency? Seems like even a few 10s of milliseconds of latency would make that rather horrible. Especially considering how fast your eyes can dart around a screen, could we even do that today?
    - knodi123 9y ago
      I dunno. There are a lot of flaws in your visual perception as it is (blind spot, nose hiding, differences in light/dark/color sensitivity based on location in FOV, etc). The brain corrects for things like that automatically. Might work just as well here.
nsxwolf 9y ago
I do not agree. I’m getting pretty tired of the shiny cartoony untextured polygon VR look. I’m ready for more R in my VR.
- andrewmcwatters 9y ago
  Personally, I'm ready for more _fun_. I don't care how realistic it is. The new experience alone is captivating. We have "realistic" in the games industry already, and its produced the worst games of all time.
tootie 9y ago
Atari sold millions of 2600s that had 160x192 resolution because joysticks are easier to use than hand waving.
simooooo 9y ago
Foviated rendering will half the requirements

by naoru 9y ago

bni 9y ago

This is so true. Case in point: Brass Tactics uses cartoonish "game of thrones intro" graphics that is quite optimized. A current graphics card can even render the game world from a different overhead view on the monitor, with capacity to spare.

phkahler 9y ago

I would love to walk around in a VR world that looked like the old vector displays. Slightly glowing wire frames but with occlusion. This technically doesn't even need polygon rendering, though doing so with mostly black textures except the edges might be cool. At todays resolutions I think small cylinders along the edges of flat black surfaces might look better. And that's just one stylized world that doesn't require a mammoth GPU to handle.

kalkut 9y ago

I absolutely agree ! I do see the appeal of photorealism but lots of excellent 3D experiences are far from photorealism. A sensible art direction can go very far, and I'd even argue that a less than ideal graphical experience can be forgiven if the content is meaningful and engaging enough. I think we all have examples of this in mind

bhouston 9y ago

> Everyone says that a powerful GPU is needed to drive this thing, but I don't think that's a must. It is not a must, but it is a mostly. Most people want something approaching realism by default. Stylism is okay but it isn't the primary use case. The video game sales show this. Top selling games are mostly ultra realistic, Witcher, Call of Duty, Modern Warfare, Battlefront, etc. Yes there is Nintendo's low end strategy but most are ultra realistic.

mbell 9y ago

You also don't need to render at full resolution. A significant benefit of higher res VR displays is not being able to physically see the pixels / screen door effect of the screen. You still get this benefit by rendering at a lower resolution and upscaling to the display resolution. Certainly you will be able to see 'render pixels', but that's a much less significant problem and just reduces the realism of the scene.

remcob 9y ago
If you add eye/focus tracking you can selectively render the part in focus in a higher resolution and use lower resolution for peripheral vision. The end result is that you are rendering at exactly the bandwidth that the human eye can handle, which is optimal and probably not that high. Going forwards, this is IMHO where the real optimization gains are.
- piracykills 9y ago
  That's a really interesting idea, but can you imagine how incredibly annoying it would be if there were any level of latency? Seems like even a few 10s of milliseconds of latency would make that rather horrible. Especially considering how fast your eyes can dart around a screen, could we even do that today?
  - knodi123 9y ago
    I dunno. There are a lot of flaws in your visual perception as it is (blind spot, nose hiding, differences in light/dark/color sensitivity based on location in FOV, etc). The brain corrects for things like that automatically. Might work just as well here.

nsxwolf 9y ago

I do not agree. I’m getting pretty tired of the shiny cartoony untextured polygon VR look. I’m ready for more R in my VR.

andrewmcwatters 9y ago
Personally, I'm ready for more _fun_. I don't care how realistic it is. The new experience alone is captivating. We have "realistic" in the games industry already, and its produced the worst games of all time.

tootie 9y ago

Atari sold millions of 2600s that had 160x192 resolution because joysticks are easier to use than hand waving.

simooooo 9y ago

Foviated rendering will half the requirements