Somewhat related: it's interesting to see how physics engines for videos games has improved so greatly over the last few decades. One of the most recent physics engines I think is really cool to see in action is the one present in The Legend of Zelda: Tears of the Kingdom! This game allows you to build infinitely many devices to traverse a vast land, and simulates the physics of our own world quite accurately. This is a big step from Super Mario 64, one of the first 3D games with a decent physics engine, which also featured glitches that would defy physics in reality
AlsonC
@brandonlouie Really interesting comparison, and I definitely agree! Another hyper-realistic game I would compare it to is GTA V, which can have up to hundreds of players at a time while maintaining gravity, the physics of vehicles, and much more.
sritejavij
I remember reading about the difficulty in producing accurate water physics, and how the new GTA 6 game has a novel in-house water physics engine. I wonder how exactly a new way of rendering water graphics would look like and what that entails within the math itself.
RishSharma7
I'd like to add onto this convo by mentioning the physics engines involved in animated tv shows, like Avatar, are also really impressive. 15-20 years ago, the effort that was required to ensure that dozens of people who can perform gravity-defying feats in a single scene are still affected by gravity unless they do something to counter that (like jump, or bend, or something) was enormous and required extreme attention to detail. It's really cool to see how new technologies and physics engines are normalizing really accurately modeled simulations, and we can see that all over the entertainment industry.
keeratsingh2002
How do we decide on the right trade-off between realistic physics and compute resources when creating particle systems for real-time rendering, like in video games?
maxwelljin
@keeratsingh2002 I think it depends on our performance budget and the trade-off between visual fidelity and realism. Some games do not need high-fidelity particle effects if the art style is visually appealing. However, for games like Microsoft’s Flight Simulator, making the particle effects as close to reality as possible is preferred, since many pilots are trained using the simulation game.
508312
I think for me it is interesting to consider how it is much more complicated simulating physics of real world than rendering. In renders we came quite close to real world, like shooting rays and converging on a true color of a pixel, that encapsulates pretty much everything there is to it. But for particle systems we do not have an approach to create everything out of them(like in real world). We have select 10-15 or so general simulations which are quite close to real life, but if you think about it it is not that much. To animate and generate scenes people take a lot of liberties. Like clipping in animation is something that will happen, objects not deforming with collision, objects always being connected, indestructible. I guess the thought I wanted to express is how closely we encapsulated real word physics with ray tracing, one solution to rendering everything. But how animating scenes and particle systems are so far from having unified approach(singular solution), making us use set of disjoint simulations to create realistic effects.
GarciaEricS
This is a bit unrelated to this slide, but it doesn't fit well in any particular one slide. I wonder if it would be possible to have some particle system where the system is dictated by the state of particle groups and groups could be condensed or expanded (adding or removing particles) to add / decrease the amount of resolution in an area. If we need more numerical accuracy, we would add particles or remove them if we know the area needs low precision, sort of like what we did for the last part of homework 3.
Somewhat related: it's interesting to see how physics engines for videos games has improved so greatly over the last few decades. One of the most recent physics engines I think is really cool to see in action is the one present in The Legend of Zelda: Tears of the Kingdom! This game allows you to build infinitely many devices to traverse a vast land, and simulates the physics of our own world quite accurately. This is a big step from Super Mario 64, one of the first 3D games with a decent physics engine, which also featured glitches that would defy physics in reality
@brandonlouie Really interesting comparison, and I definitely agree! Another hyper-realistic game I would compare it to is GTA V, which can have up to hundreds of players at a time while maintaining gravity, the physics of vehicles, and much more.
I remember reading about the difficulty in producing accurate water physics, and how the new GTA 6 game has a novel in-house water physics engine. I wonder how exactly a new way of rendering water graphics would look like and what that entails within the math itself.
I'd like to add onto this convo by mentioning the physics engines involved in animated tv shows, like Avatar, are also really impressive. 15-20 years ago, the effort that was required to ensure that dozens of people who can perform gravity-defying feats in a single scene are still affected by gravity unless they do something to counter that (like jump, or bend, or something) was enormous and required extreme attention to detail. It's really cool to see how new technologies and physics engines are normalizing really accurately modeled simulations, and we can see that all over the entertainment industry.
How do we decide on the right trade-off between realistic physics and compute resources when creating particle systems for real-time rendering, like in video games?
@keeratsingh2002 I think it depends on our performance budget and the trade-off between visual fidelity and realism. Some games do not need high-fidelity particle effects if the art style is visually appealing. However, for games like Microsoft’s Flight Simulator, making the particle effects as close to reality as possible is preferred, since many pilots are trained using the simulation game.
I think for me it is interesting to consider how it is much more complicated simulating physics of real world than rendering. In renders we came quite close to real world, like shooting rays and converging on a true color of a pixel, that encapsulates pretty much everything there is to it. But for particle systems we do not have an approach to create everything out of them(like in real world). We have select 10-15 or so general simulations which are quite close to real life, but if you think about it it is not that much. To animate and generate scenes people take a lot of liberties. Like clipping in animation is something that will happen, objects not deforming with collision, objects always being connected, indestructible. I guess the thought I wanted to express is how closely we encapsulated real word physics with ray tracing, one solution to rendering everything. But how animating scenes and particle systems are so far from having unified approach(singular solution), making us use set of disjoint simulations to create realistic effects.
This is a bit unrelated to this slide, but it doesn't fit well in any particular one slide. I wonder if it would be possible to have some particle system where the system is dictated by the state of particle groups and groups could be condensed or expanded (adding or removing particles) to add / decrease the amount of resolution in an area. If we need more numerical accuracy, we would add particles or remove them if we know the area needs low precision, sort of like what we did for the last part of homework 3.