In this case, is Monte Carlo integration used to achieve realism or to make computations more efficient?
Shruteek
The true nature of light shadowing compared to random sampling and center sampling makes me wonder how computer graphics would deal with the quantum nature of light - if it is even practically possible. For example, if a model shines a point source of light on a circular object, Fresnel diffraction should yield a bright Arago spot on the opposite side instead of pure darkness. Similarly, the way light reflects and refracts through a glass of water is unaccounted for (flipping the reflected image and certain angles completely obscuring the objects around it). Does the scale of macroscopic lighting in computer graphics allow all these properties to be neglected?
SeanW0823
For each pixel on the ground, can we average multiple random samples to get an approximate result to the true answer?
akshitdewan
I think I understand monte carlo integration conceptually, but I'm having a hard time making the connection to this example. What's the relevant integral that we're sampling here?
chetan-khanna
This sharp shadows generated if we sample the centre of the light generated by the current ray-tracing model we've seen actually made me look back into ray-tracing a little bit more.
I was somewhat interested in how we can use ray tracing to model some phenomena of light, particularly diffraction (this felt—although I wasn’t sure—slightly more difficult to model, given we are ray tracing after all, and assuming (in our simplified model) that rays are shooting in a straight line).
I found an academic paper (http://web.cs.wpi.edu/~emmanuel/publications/PDFs/C72.pdf) that tries out various methods of modelling diffraction and then using ray-tracing; I also found another paper (https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6814971) that actually suggests a new technique for diffraction (and for simulating the wave properties of light more generally) in ray-tracing.
In this case, is Monte Carlo integration used to achieve realism or to make computations more efficient?
The true nature of light shadowing compared to random sampling and center sampling makes me wonder how computer graphics would deal with the quantum nature of light - if it is even practically possible. For example, if a model shines a point source of light on a circular object, Fresnel diffraction should yield a bright Arago spot on the opposite side instead of pure darkness. Similarly, the way light reflects and refracts through a glass of water is unaccounted for (flipping the reflected image and certain angles completely obscuring the objects around it). Does the scale of macroscopic lighting in computer graphics allow all these properties to be neglected?
For each pixel on the ground, can we average multiple random samples to get an approximate result to the true answer?
I think I understand monte carlo integration conceptually, but I'm having a hard time making the connection to this example. What's the relevant integral that we're sampling here?
This sharp shadows generated if we sample the centre of the light generated by the current ray-tracing model we've seen actually made me look back into ray-tracing a little bit more.
I was somewhat interested in how we can use ray tracing to model some phenomena of light, particularly diffraction (this felt—although I wasn’t sure—slightly more difficult to model, given we are ray tracing after all, and assuming (in our simplified model) that rays are shooting in a straight line).
I found an academic paper (http://web.cs.wpi.edu/~emmanuel/publications/PDFs/C72.pdf) that tries out various methods of modelling diffraction and then using ray-tracing; I also found another paper (https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6814971) that actually suggests a new technique for diffraction (and for simulating the wave properties of light more generally) in ray-tracing.