Hope you all enjoyed this episode, and that the visualization was helpful! Join the episode discussion thread on my Discord in the #raytracing-series channel: discord.gg/thecherno And of course for all of your math needs, visit brilliant.org/TheCherno to get started learning STEM for free, and the first 200 people will get 20% off their annual premium subscription.
27:35 Hahaha didn't expect you'd actually talk about it and prove it! I'm glad I commented my misconception in the previous video, else I'd wouldn't realise it
You know you made it in life when you start using your own engine to teach people things that you know🤜🏽🤛🏽😂. Mad respect for Cherno, the guy is a legend🔥
That's the greatest 3d visualization I've ever seen in this area. Keep going on. Coud you please share what CPU you have ? Any plans to cover other geometrical primitives except spheres?
Really fun series and really cool now that we are getting a bit into the meat of it all. This showcases what raytraced Rendering actually is. Where basically you are creating an entire image from just rays being thrown around all over the place and then those rays are doing other things to create a perfectly accurate colored pixel. All that math and power being thrown around simply to create an accurate final color. Can't wait to see how all the bouncing works and then reflections and shadows and RTGI and such like that. Are your plans to go through each concept of RT in its own video? Like a video on just reflections and how it all works, just lighting and how RT Lighting works and how exactly RT shadows work or just throw it all into the same video? Also look forward to seeing BVHs and the lesson on why RT is so expensive in the end even with these machines that supposedly can perform trillions of FP calculations a second. Always cool to see this stuff come to life so to speak
This is kind of blowing my mind.. I was sure raytracing was using more complicated math than just the solution to the quadratic equation, but I can see that it's just building on basic algebraic building blocks. Really great visuals and insights, thanks for sharing.
Might have been worth mentioning that normalizing vectorA translates to "vectorA / sqrt(dot(vectorA, vectorA))". Just because it makes it very clear how normalizing is ofc more work than just taking a dot product. normalizing = dividing each vector component by the square root of the vectors dot product with itself. since the dot product of a vector with itself is inherently it's squared magnitude
Just with your Maths-(Deep-As-The-Ocean)-video I was able to come up with simple and very bad code using OpenGL (The Basics I learned from your OpenGL Series (Just the first few episodes)). Finally I already have implemented reflections with different reflectabilities and several spheres. The problem is, that my architecture is not scalable at all. Everything is just running in one huge pixel shader. But it works really well and i‘m so excited about, what I was able to do. I really have to thank you for the indepth explanation of everything.
Hey cherno, you haven't done a C++ series video in a while, and just wanted to say, if the series is still alive you could make a video on constexpr keyword as you didnt cover that yet
It's so simple.. The compiler will calculate the result before compiling the code.. Just type "constexpr" and then hover your mouse into the return variable And the compiler will show the result/output/value
@@footballCartoon91 there are more cases to it for example a constexpr constructor and destructor, i still dont fully understand the keyword i think it would be great to have a short video on it in the series
ok, I've worked out what the problem was. When we are shown the correct shading on the green billboards, we are viewing the scene through *Hazel's* camera and that is pointing in the opposite direction to the *ray tracing* camera.
If I’m not mistaken at the end you’ll have to convert all your math to triangle intersection since that’s what GPU hardware is meant to do, correct me if I’m wrong
GPUs are actually good at doing just about any array math. The real reason to use triangles is because you need some way to define a surface and while things like spheres, cubes, toruses, etc can have their surfaces defined using a simple equation, you want to be able to render arbitrary shapes like characters or buildings and you can approximate arbitrary surfaces by breaking them down into a bunch of triangles.
think about light rays bending around corners, the particle photon (vector model) is not entirely accurate, think 1 and 2 slot photon wave dispersion experiments
@@samochreno I don't think I'm negative, are objective. I followed the other series, they were very interesting but they stopped mid course. Most people idolize you, I think there must be a place for a post like mine, don't you ?
![Oscar Maydon x Fuerza Regida - Tu Boda [Official Video]](http://i.ytimg.com/vi/_ymicn0_GYc/mqdefault.jpg)
Hope you all enjoyed this episode, and that the visualization was helpful! Join the episode discussion thread on my Discord in the #raytracing-series channel: discord.gg/thecherno
And of course for all of your math needs, visit brilliant.org/TheCherno to get started learning STEM for free, and the first 200 people will get 20% off their annual premium subscription.
;)
27:35 Hahaha didn't expect you'd actually talk about it and prove it! I'm glad I commented my misconception in the previous video, else I'd wouldn't realise it
Also amazing, I've never seen Ray tracing being demonstrated backwards like this!
I physically feel my brain cell number increasing with this series
You can't even recall your real name!
Mine can’t yet comprehend. Yours must be very advanced.
@@QuintinMassey Quite plainly. You can't even recall your real name. Seek help,
hahahaha you put my feelings into words
This raycasting series is the best visualization I've seen online. Thank you for the videos, they've been very useful.
You know you made it in life when you start using your own engine to teach people things that you know🤜🏽🤛🏽😂. Mad respect for Cherno, the guy is a legend🔥
I love how you used Hazel to visualize it!
That's the greatest 3d visualization I've ever seen in this area. Keep going on. Coud you please share what CPU you have ? Any plans to cover other geometrical primitives except spheres?
24:20
The math was so deep, the camera had a hard time staying focused :)
lol
Cherno back at it again with the fresh fade 😎
Really fun series and really cool now that we are getting a bit into the meat of it all.
This showcases what raytraced Rendering actually is. Where basically you are creating an entire image from just rays being thrown around all over the place and then those rays are doing other things to create a perfectly accurate colored pixel.
All that math and power being thrown around simply to create an accurate final color.
Can't wait to see how all the bouncing works and then reflections and shadows and RTGI and such like that.
Are your plans to go through each concept of RT in its own video? Like a video on just reflections and how it all works, just lighting and how RT Lighting works and how exactly RT shadows work or just throw it all into the same video?
Also look forward to seeing BVHs and the lesson on why RT is so expensive in the end even with these machines that supposedly can perform trillions of FP calculations a second.
Always cool to see this stuff come to life so to speak
I loved this format and pace!
This is kind of blowing my mind.. I was sure raytracing was using more complicated math than just the solution to the quadratic equation, but I can see that it's just building on basic algebraic building blocks. Really great visuals and insights, thanks for sharing.
You have the best explanations, thank you for these videos, oh and btw, nice hair :)
Love using hazel as a visualization tool. More like this, please 🙏!
Amazing visualization! Love your work Cherno
I post this comment to increase the engagement of this video for RUclips. This series have been amazing so far.
Great job! Just a friendly reminder, your camera was focusing on the board at certain points of the video so I hope you can fix that next time.
Might have been worth mentioning that normalizing vectorA translates to "vectorA / sqrt(dot(vectorA, vectorA))". Just because it makes it very clear how normalizing is ofc more work than just taking a dot product. normalizing = dividing each vector component by the square root of the vectors dot product with itself. since the dot product of a vector with itself is inherently it's squared magnitude
why Hazel uses C# instead of C++ ?
Just watching this videos, will not be enough to digest the information :)
Starting to get quite spicy!
Just with your Maths-(Deep-As-The-Ocean)-video I was able to come up with simple and very bad code using OpenGL (The Basics I learned from your OpenGL Series (Just the first few episodes)). Finally I already have implemented reflections with different reflectabilities and several spheres. The problem is, that my architecture is not scalable at all. Everything is just running in one huge pixel shader. But it works really well and i‘m so excited about, what I was able to do. I really have to thank you for the indepth explanation of everything.
I cannot even program a Tetris game..
Congratulations
You guys are coding games!? I'm stuck with queues and trees only
So well explained with the code and live visuals side by side
You are very good teacher
Wow hazel seems amazing! I can only hope to become as good as you!
unbelievable tutorial I've ever seen
this is interesting, i’m curious how you determine the hit position of a more complex shape that’s not mathematically defined?
Loved it, thanks!
Hey cherno, you haven't done a C++ series video in a while, and just wanted to say, if the series is still alive you could make a video on constexpr keyword as you didnt cover that yet
It's so simple..
The compiler will calculate the result
before compiling the code..
Just type "constexpr" and then hover your mouse into the return variable
And the compiler will show the result/output/value
@@footballCartoon91 there are more cases to it for example a constexpr constructor and destructor, i still dont fully understand the keyword i think it would be great to have a short video on it in the series
What if someone does not want to write in c# is it also possible to do same with a c++ class here?
Nice job!
I coded this up in JavaScript, and to get the shading on the magenta sphere correct I had to do 1-light… can’t work out why it is different
ok, I've worked out what the problem was. When we are shown the correct shading on the green billboards, we are viewing the scene through *Hazel's* camera and that is pointing in the opposite direction to the *ray tracing* camera.
If I’m not mistaken at the end you’ll have to convert all your math to triangle intersection since that’s what GPU hardware is meant to do, correct me if I’m wrong
GPUs are actually good at doing just about any array math. The real reason to use triangles is because you need some way to define a surface and while things like spheres, cubes, toruses, etc can have their surfaces defined using a simple equation, you want to be able to render arbitrary shapes like characters or buildings and you can approximate arbitrary surfaces by breaking them down into a bunch of triangles.
Why does changing the width/height variable affect the density of rays? I would've expected it to just change the width/height of the final image.
Width and height signifies the number of rays being cast horizontally and vertically into the scene. The scene size isn’t being changed.
@@bishboria Yes but that doesn't explain why
@@DevJG2 because even if those are the names he's using, he's just using them to move in a grid some known units
very good 👌👌
Did you try jai?
As ever, thank you 👍👍
ray intersections could be a whole series on it's own...lmao
My gpu drivers doesn't support vulkan, is there any way around please ?
Hazel should have it’s own library of math functions. You should call it HazMat
Man c# is so much more readable. How do I get my eyes used to that c++ syntax ::>>
what is the most enjoyable sphere game, a flipper, or a just a sphere rolling game, world full of spheres and planes
binocular vr double eyed view
particle gas simulation
think about light rays bending around corners, the particle photon (vector model) is not entirely accurate, think 1 and 2 slot photon wave dispersion experiments
its more like a distribution of rays, not just one ray on exact course, statistics vs exact rays
if you have a bright light (hdr) when coming out of a tunnel, the light intensity will be even stronger on the tunnel rim because of the light bending
"To normaliZe or not to normaliSe"
Cherno, 27min40
I like how you start series after series and then abandon them all.
Why are you so negative
@@samochreno I don't think I'm negative, are objective. I followed the other series, they were very interesting but they stopped mid course. Most people idolize you, I think there must be a place for a post like mine, don't you ?
@@iseedumbpeople6369 im not cherno bro xD i just have a similiar name
My bad, having fever at the time didn't help. Sorry.
just wow.
Am I the only one to hear a big boom, very low pitched, every so often?
It’s almost painful seeing how slow these episodes come out, I hope your recent post shows you are still working on this
I'm trying to put the brain goo back into my ears
Woah
Handmade raytracing
;)
"C"??
Go away.