I was wondering if there was a way to do Ray tracing with polynomials, seems like there currently isn't (efficiently anyway). This video is a gem for understanding what the current limitations are before making your own.
One guy actually did it, google "Raytracing a Blackhole" Okay, its raytracing just one particular version of curved spacetime, but you get the idea, its cool(and quite simple to implement too).
If you're referring to the hypothesis that we live in a simulation - and setting aside any opinion of plausibility of the theory - we couldn't really say for sure. If there's nobody there to see something, is it visible? If a tree falls in the forest, does it make a sound? Do these phenomena exist even if there are no eyes to see, noses to smell, ears to hear, nerves to feel, tongues to taste? Is Schrodinger's cat dead or alive if we have no observable evidence of either state? This is a fun one to think about: We can't visually observe the cat without opening the box, but maybe we can observe the temperature. If we put an ice cube on the box, it will melt faster if the cat is alive than if the cat is dead. This alone will constitute observation. In fact any interaction between the cat and the outside of the box is "considered" observation, even if we don't pay attention to the ice cube. So are our senses simulated? Do our eyes see raytraced images, i.e. only the observed "exists"? Or are the stimuli - the photons - simulated? If we were building this simulation, we wouldn't be able to tell if any given photon is going to interact with an observing organism without simulating every photon to its eventual absorption. So it's more likely that for such a rich and complete simulation. photons would *need* to be simulated from their source rather than traced from our retinas.
Swoorup Joshi yeah right. it is so much better to see a lecture given by a native english speaking guy. Also, this Professor is really presenting it nicely.why even bother about rasterization? throw it out of the window, in 4 years will have real time raytracing.
+plavix221 if it takes 2 minutes to render the sample image of a bmw with cycles in blender, and we want 60 fps, we need gpus that are 7200x as powerful. If it takes four years for power to double, a conservative estimate since we are reaching the size limit of traditional transistors, we can find out how long it should take. with x= number of years, 7200=(1/4*x)^2,x=339. However, if we can keep up moore's law, 7200=(1/2*x)^2,x=169. Of course, we can always say that you can use $10,000 worth of equipment, which is when the consumer market usually starts getting it. I'd say about 20 seconds is more like what you would see. Now it's only 1200x less powerful than what is needed. now let's say they can put up with 20 fps for some reason. We will also say that moore's law still applies, doubling once every two years. 1200=(1/2*x)^2. In 69 years, you'll be able to spend $10,000 and watch the bmw benchmark at 20 fps. Finally, let's say right now you have some uber $1,000,000 array that can render it at blazing speeds of 0.2 seconds per frame. being 12x less powerful, you can either spend $12,000,000 and play at 20 fps ($36,000,000 if you want 60 fps), or keep waiting until prices go down to be able to do this. I haven't taken into account optimizations in render technologies, but the gist is you CAN do it in 4 years, if you have $2,500,000 or so laying around. This does not take into consideration any other parts of the PC you may need :^)
+plavix221 nope. It all depends on how you consider the surface. It can be one sided,two-sided (outer surface-inner surface) but,"digitally speaking",often a sphere has many sides since we have to use triangles or quads to describe it. There are other exotic digital representations(actually,the real term to use is *approximations*) but we get in the realm of dot-clouds,fractals,voronoi diagrams and "who knows what else"... :P
5:50 This is an incredibly bad practice that nobody does for what should be obvious reasons. He says they do that but as a professional in the industry I can tell you, no. Nobody in their right mind does this.
![Lp. Сердце Вселенной #60 РОЖДЕНИЕ ЛОЛОЛОШКИ [Финал] • Майнкрафт](http://i.ytimg.com/vi/YoR0pAV9FVQ/mqdefault.jpg)
Lecture starts at 7:33
My hero
Thanks
ma boi
You deserve more than likes
Thanks. Good.
I was watching cats and ended up here watching a lecture on ray tracing
hahahhaha!
Congratulations!
Me too lol
youtube algo is little miracle 😂
Lmaooo
Lecture starts at 7:37. Before that he speaks with the students about projects and miscellaneous spatial rendering problems
thanks man
I read you at 7:15 =(
@@gaxkiller lol
Very good lecture! It's really interesting that Nvidia's CEO basically explained pretty much the same thing in RTX's launch event.
I just got back from Pixar's Elemental after having watched this video last week - and my mind is blown. All of the things!
All of them!
This is so cool when you finally understand the math behind it! xD
RTX ON
That must be why the frame rate of this video is so low
I was wondering if there was a way to do Ray tracing with polynomials, seems like there currently isn't (efficiently anyway). This video is a gem for understanding what the current limitations are before making your own.
If you close your eyes, Kermit the Frog will teach you the basics of ray tracing.
lol
Thank you for the great explanation ....really helpful introductory.
"Don't worry about the holes"... my new mantra
Now make a raytracer for curved spacetime
The Misanthropic Nihilist It's much harder to represent a curved line than a straight one.
You could ray march as a series of small line segments, obviously much more expensive that way
One guy actually did it, google "Raytracing a Blackhole"
Okay, its raytracing just one particular version of curved spacetime, but you get the idea, its cool(and quite simple to implement too).
Checkout Scott manleys video on wormholes. He wrote a raytracer for non Euclidean space to show what it would look like going into a wormhole
It's basically the same thing, you just change the metric
33:00 "Run the code... dooo...."
That should become a meme...
I have NO clue what's going on. BUT, could this describe the holographic universe?
If you're referring to the hypothesis that we live in a simulation - and setting aside any opinion of plausibility of the theory - we couldn't really say for sure. If there's nobody there to see something, is it visible? If a tree falls in the forest, does it make a sound? Do these phenomena exist even if there are no eyes to see, noses to smell, ears to hear, nerves to feel, tongues to taste? Is Schrodinger's cat dead or alive if we have no observable evidence of either state? This is a fun one to think about: We can't visually observe the cat without opening the box, but maybe we can observe the temperature. If we put an ice cube on the box, it will melt faster if the cat is alive than if the cat is dead. This alone will constitute observation. In fact any interaction between the cat and the outside of the box is "considered" observation, even if we don't pay attention to the ice cube. So are our senses simulated? Do our eyes see raytraced images, i.e. only the observed "exists"? Or are the stimuli - the photons - simulated? If we were building this simulation, we wouldn't be able to tell if any given photon is going to interact with an observing organism without simulating every photon to its eventual absorption. So it's more likely that for such a rich and complete simulation. photons would *need* to be simulated from their source rather than traced from our retinas.
TL;DR dooo concretely proves the Universe is an illusion
Well by now it's a trending topic to the gaming community. Can't wait to see what are they going to do with 'Ray Tracing' on upcoming games.
From what year is this lecture? I am loving it!
2009
So this is Ray tracing before it was cool? nice.
well I was just wondering what the name of the video is which would be played on Friday , seems to be very interesting.
It's pr0n
Why I just found this ray tracing comments are best comms on UA-cam ever
This lecture is GREAT! Thanks
Damn. I'm the chosen one
What class is this?
Never mind, google answered my question. It's "CSCI 420 Computer Graphics" for those curious.
NVIDIA brings me here
anyone knows where I can find the videos he is talking about?
very interesting and well explained, ty!
wow very informative
Swoorup Joshi yeah right. it is so much better to see a lecture given by a native english speaking guy. Also, this Professor is really presenting it nicely.why even bother about rasterization? throw it out of the window, in 4 years will have real time raytracing.
+plavix221 if it takes 2 minutes to render the sample image of a bmw with cycles in blender, and we want 60 fps, we need gpus that are 7200x as powerful. If it takes four years for power to double, a conservative estimate since we are reaching the size limit of traditional transistors, we can find out how long it should take. with x= number of years, 7200=(1/4*x)^2,x=339. However, if we can keep up moore's law, 7200=(1/2*x)^2,x=169. Of course, we can always say that you can use $10,000 worth of equipment, which is when the consumer market usually starts getting it. I'd say about 20 seconds is more like what you would see. Now it's only 1200x less powerful than what is needed. now let's say they can put up with 20 fps for some reason. We will also say that moore's law still applies, doubling once every two years. 1200=(1/2*x)^2. In 69 years, you'll be able to spend $10,000 and watch the bmw benchmark at 20 fps. Finally, let's say right now you have some uber $1,000,000 array that can render it at blazing speeds of 0.2 seconds per frame. being 12x less powerful, you can either spend $12,000,000 and play at 20 fps ($36,000,000 if you want 60 fps), or keep waiting until prices go down to be able to do this. I haven't taken into account optimizations in render technologies, but the gist is you CAN do it in 4 years, if you have $2,500,000 or so laying around. This does not take into consideration any other parts of the PC you may need :^)
plavix221 DAMM your prediction was on point, around 4 years later we have it
I'm jizzing my pants right now.
Now we have real time ray tracing yeah
because we have very powerful hardware
Why does a ray hit a sphere in 2 places?????
I cant imagine that one.
plavix221 oh ok i see. there are acutally two sides a sphere has, not? XD
+plavix221 nope.
It all depends on how you consider the surface.
It can be one sided,two-sided (outer surface-inner surface) but,"digitally speaking",often a sphere has many sides since we have to use triangles or quads to describe it.
There are other exotic digital representations(actually,the real term to use is *approximations*) but we get in the realm of dot-clouds,fractals,voronoi diagrams and "who knows what else"... :P
+plavix221 he meant 2 surfaces of the sphere: front and back
Jack Le But which one? There is an infinite amount of them. XD
plavix221 whichever one intersects the ray ;) the sphere is hollow
The solution to the shadow problem sounds really obvious
A good lecturer makes everything they're explaining sound obvious.
It is when you are ray casting with no time limit but when you are attempting real-time cg (especially back then), that isn’t an option.
Even when you are ray casting with "no time limit", it might take too damn long or the machine crashes mid-trace (especially back then)
22:42 - One of the jobs I'd like to have also :)
you cant, niga
5:50 This is an incredibly bad practice that nobody does for what should be obvious reasons. He says they do that but as a professional in the industry I can tell you, no. Nobody in their right mind does this.
What do you do?
@@kamogelokekae6877 I work with realtime graphics for game dev
What if the rays were parabolas?
Test it
What kind of course do I go to learn about this stuff?
computer graphics and light transport
lecture is OK
Why did you put C (speed of light) on the desk. And why do you always tell "as you can C"? :D :D :D
4 D divide by W?
Look up homogenous coordinates and perspective projection
Didnt you get little nervous, when he starts talking about assignments ? Not cool, dude (xD)
*Not today UA-cam*
10:37 heeeeeere
is this a graduate course?
elementary school course
FU Nvidia... I'm still not going to pay $1000+ for a RTX 2080 Ti.
Just for gaming it's too much. The RTX could be really good for animators and 3D artists though.
coz this is inovation, u pay 4 that
There's not enough games that really requires RT, it's just a novelty...
Goob explications boys :)
13:28 iihaaaaaghhh
7:37
When universities really taught something, no like the Gender/Black/Mexicans/Etc. Studies courses they offer right now.
it depends on what you take. they still teach that stuff. what a dumb comment.
Gender studies are way more important than computer graphics BIGOT
@@IvanRandomDude Go back to get your soy.
@@victorsegoviapalacios4710 Agreed.
@@veloce5491 If you take gender studies you're wasting your time and not helping humanity
I'm not visualizing the illustrations. It would be better if it is 3D animation, or use some objects to illustrate.
100 lines of code to see if it’s inside a triangle is bollocks. I did it in 20
@@uku4171 I care.
how long are your lines?
graphics
if Louis CK had a computer scientist brother..
Traduzcan al.español
Nou