4 ms·

Every time i see this quality of computer graphics i ask myself: how do i still know its not real, what little detail is still missing to photo realism...

krm01 6y ago
It's almost always the lighting.
- qzw 6y ago
  Yes, and particularly the way light reflects off of different materials is often still in the uncanny valley. Perhaps ray tracing will finally get us there in the next few years.
  - WrtCdEvrydy 6y ago
    Lighting and water are kinda hard to nail perfectly... there's just something about the real thing that's hard to duplicate.
    - dylan604 6y ago
      And fire. Fire always looks fake in GGI whether that's in games or in movie/tv
      - holoduke 6y ago
        More generically: fluid simulations. Fire, smoke, water etc
  - krm01 6y ago
    That's the thing, light doesn't just reflect off materials. A bunch of lights gets absorbed by it, changing the color and the amount of light that is reflected back, from a deeper surface area. There are just so many things happening with light, most of it we don't fully understand yet, but we seem to instinctively see when it's off.
    - JoshTriplett 6y ago
      Path tracing can replicate that effect, including things like an object near a window lighting the room with reflected light in its colors, and refraction effects through some materials. See https://raw.githubusercontent.com/Dalamar42/rayt/master/samples/cornell_box.png https://raw.githubusercontent.com/Dalamar42/rayt/master/samp... for one example; notice that the white light reflecting off the green wall makes the side of a nearby object look green, for instance.
      - zokier 6y ago
        Path tracing with colors still presents some challenges. Most mainstream renderers use RGB model, but of course real world has the full EM spectrum. For example for the Cornell box scene, imagine that the wall was pure green material and the box pure yellow material. Then there shouldn't be that visible secondary bounce, but with naive RGB path tracing there would be, because it wouldn't be able to distinguish between pure yellow and mixed yellow. In practice I would expect the effect to be pretty subtle in most scenes, but e.g. scenes lit by sodium streetlights could have stark difference. For example something like https://commons.m.wikimedia.org/wiki/File:Red_and_black_cars_under_low_pressure_sodium_lamps.jpg#mw-jump-to-license https://commons.m.wikimedia.org/wiki/File:Red_and_black_cars...
    - qzw 6y ago
      Yes, there's definitely more going on than just reflectance. That's why most virtual materials appears too glossy to me. From what I understand, ray/path tracing will enable better materials modeling as well, given enough processing power. But I'm not a graphics person, so would be happy to be corrected.
    - nuccy 6y ago
      Actually we do understand all the things which happen during normal conditions with light (i.e. when we speak about: bunches of photons and not a single one; objects big enough so that quantum effects can be neglected; relatively low intensities, so that self collimation and other effects related to it don't play a role), at least from physics stand point. Maxwell equations do a good job for us to explain light and its behaviour. Though, obviously, there are so many effects coming from those equations: refraction, reflection, diffraction, interference, dispertion, scaterring, etc, which just are numerically complicated all to be taken into account for a rendering engine.
carrolldunham 6y ago
There are some lighting differences but in this case the main 'disparity' seems to be that everything is in focus at once, since they are not attempting to simulate a camera photograph.
demarq 6y ago
One area I feel modern graphics struggle with is GI. https://news.ycombinator.com/item?id=22722734 https://news.ycombinator.com/item?id=22722734

krm01 6y ago

It's almost always the lighting.

qzw 6y ago
Yes, and particularly the way light reflects off of different materials is often still in the uncanny valley. Perhaps ray tracing will finally get us there in the next few years.
- WrtCdEvrydy 6y ago
  Lighting and water are kinda hard to nail perfectly... there's just something about the real thing that's hard to duplicate.
  - dylan604 6y ago
    And fire. Fire always looks fake in GGI whether that's in games or in movie/tv
    - holoduke 6y ago
      More generically: fluid simulations. Fire, smoke, water etc
- krm01 6y ago
  That's the thing, light doesn't just reflect off materials. A bunch of lights gets absorbed by it, changing the color and the amount of light that is reflected back, from a deeper surface area. There are just so many things happening with light, most of it we don't fully understand yet, but we seem to instinctively see when it's off.
  - JoshTriplett 6y ago
    Path tracing can replicate that effect, including things like an object near a window lighting the room with reflected light in its colors, and refraction effects through some materials. See https://raw.githubusercontent.com/Dalamar42/rayt/master/samples/cornell_box.png https://raw.githubusercontent.com/Dalamar42/rayt/master/samp... for one example; notice that the white light reflecting off the green wall makes the side of a nearby object look green, for instance.
    - zokier 6y ago
      Path tracing with colors still presents some challenges. Most mainstream renderers use RGB model, but of course real world has the full EM spectrum. For example for the Cornell box scene, imagine that the wall was pure green material and the box pure yellow material. Then there shouldn't be that visible secondary bounce, but with naive RGB path tracing there would be, because it wouldn't be able to distinguish between pure yellow and mixed yellow. In practice I would expect the effect to be pretty subtle in most scenes, but e.g. scenes lit by sodium streetlights could have stark difference. For example something like https://commons.m.wikimedia.org/wiki/File:Red_and_black_cars_under_low_pressure_sodium_lamps.jpg#mw-jump-to-license https://commons.m.wikimedia.org/wiki/File:Red_and_black_cars...
  - qzw 6y ago
    Yes, there's definitely more going on than just reflectance. That's why most virtual materials appears too glossy to me. From what I understand, ray/path tracing will enable better materials modeling as well, given enough processing power. But I'm not a graphics person, so would be happy to be corrected.
  - nuccy 6y ago
    Actually we do understand all the things which happen during normal conditions with light (i.e. when we speak about: bunches of photons and not a single one; objects big enough so that quantum effects can be neglected; relatively low intensities, so that self collimation and other effects related to it don't play a role), at least from physics stand point. Maxwell equations do a good job for us to explain light and its behaviour. Though, obviously, there are so many effects coming from those equations: refraction, reflection, diffraction, interference, dispertion, scaterring, etc, which just are numerically complicated all to be taken into account for a rendering engine.

carrolldunham 6y ago

There are some lighting differences but in this case the main 'disparity' seems to be that everything is in focus at once, since they are not attempting to simulate a camera photograph.

demarq 6y ago

One area I feel modern graphics struggle with is GI. https://news.ycombinator.com/item?id=22722734 https://news.ycombinator.com/item?id=22722734