วีดีโอ

I actually find working with light themes at the moment easier on my eyes. However you are not the only one asking for me using a dark theme again. Therefore I have switched to a dark theme for future videos. However I still have got some more in the pipeline, which have already been recorded in advance and will therefore be with my current bright theme. I hope you will endure those until darker times arise ;)

Light mode for life, let's keep things actually visible.

Even though, as I pointed out I currently deem light themes more readable for my eyes as well, I understand, that many people are watching those videos in the evening in dimmed down light conditions. Therefore I can understand the request to have an overall less bright appearance. I hope to have found a darker, but yet clearly visible theme to be used in the future, that allows both sides to be happy.

@@MrJakob Light mode is great during the day. I love using both with my system automatic switching twice a day

Thank you it’s good to be back. I hope you will enjoy the new videos. Let’s learn together! That’s half the fun 🤓

That makes a lot of sense. Thanks 👍

Thanks that looks interesting. I will take a closer look once I got some spare time 👍

Too bad you don’t want to support me. But I am glad you liked the content.

Thanks for the feedback. I got quite some videos still in the queue, which are prepared and ready to be published. For newer ones however I will see, what I can do however. 👍

You are very much welcome. Making those videos is fun for me as well 😃

Thanks. My color scheme is a slightly adapted Solarized light. You can find my shell/environment configuration on github: github.com/jakobwesthoff/dotfiles Especially the alacritty and tmux configs are mostly interesting for you if you want to checkout the font and color scheme.

Thanks. Yes I discovered glam by now as well, but same as you I wanted to do things from scratch to learn 😁

The implementation is purely cpu bound, as none of the computation is done on the gpu. It would be possible of course to use the gpu to enormously speed up the calculation. The main speed bump for the implementation was to utilize all processor cores using the rayon library. There is a session about when I discovered that as well.

Thanks for your kind words. You might be right. I will consider this, when I start working on this again in the future.

What you need to understand raytracing is a basic knowledge about geometry, which includes vectors and a little trigonometry. If you have basic understanding of those concepts you are perfectly fine, as the book is very detailed in its explanations. You might even be able to follow the book, without that knowledge, but I guess it wouldn’t be much fun, as you will be able to „copy“ from the book, but you won’t understand what you are doing exactly. I hope that helps a little.

@@MrJakob Thank you for the reply!

for some reason inticatif breaks CodeLLDB for me if i run it through the vs code debug button(or F5). however it builds fine and displays if i do a "cargo run" not sure if you have the same issues or not(maybe you know how to fix it?). well now that i typed this all out i remembered you are on IOS so maybe not :) also does the Body enum implementation cause any issues later on? i have never seen a ray hit return the object itself before and i would normally just have a HitRecord struct that gets returned with the hit information(t, ray, normal, etc) by the intersect function. normally the final calculations don't need to care about the object that the ray hit. only the ray and the t value(to calculate the hit position) and normal which we haven't gotten to yet. another thing im thinking about is with reflections and bounce lighting. wouldn't each bounce do a copy of the whole object for each bounce? i initially did it using a normal struct but converted it to use the enum dispatching because i wanted to learn a different way it's interesting but im not sure yet when i would ever need to reference the original object. though i suppose you could use that as a way of attaching materials and properties directly to the object. and referencing it later. again normally i would store color info in the HitRecord and modify it iteratively with each bounce) where it seems like with the enum dispatch approach you would need to collect and calculate everything at the end using all the Body's that get returned. anyways lol sorry for the wall of text.

I am glad you like the videos. As you can see it was a big learning for me as well. I tried things and later realized they are not what I wanted. Therefore they had to go again. But essentially that is why I started this series in the first place. I hope you will enjoy the rest of the videos as well. I will hopefully find the time to do further work on this project and someday „complete“ it. My spare time is quite limited at the moment. But I am confident to be able to release some new videos in the future. Comments like yours actually motivate me even more to do so.

@@MrJakob awesome! it's really fun following along with you as you are learning it. unlike a classic university course where everything is set and the teacher is a master it gives me room to pause and go searching myself for different ways to implement a new idea. then come back and unpause you. a lot of times we end up writing the same/similar code but sometimes you do something different and ill implement both and figure out which way I prefer. i don't know how long it will take to make it through all of the rest but i don't have any plans on stopping. i have heard of rust for a few years now but never tried it(old c/c++ programmer that has been using python and C# for the last 10 years). so far it's a pretty nice language and i have only missed python a little :) im not sure what your main languages are but any plans on using rust outside of this project?

That sounds like fun 👍. I have actually used rust for some other small projects I needed to be done privately. They are on my GitHub as well. Furthermore I actually wrote a certain service at work in rust quite recently, because it was a nice fit. I like the language and the community around it and don’t think I will be stopping writing it soon. I have rough plans to maybe write a minimal kernel in rust, just because I have never written a basic or layer before. But that will most likely be another TH-cam series then 😅

@@MrJakob Nice i have some ideas about trying some neural networking stuff in rust once im done with the ray tracing. it's always a good sign when you learn a new language to use it beyond the learning project.(or in my case thinking about it) the kernel sounds super interesting i would definitely watch that. i tried implementing my own kernel years ago in C but stopped after basic io.

Thank you very much. I am sorry for not having released a new video in such a long time. There will be more in the future, that I can assure you. In the last two month I nearly didn’t have any spare time to work on this project. But I am seeing the light at the end of the tunnel. Therefore I hope there will be a new session out soon, covering refraction and there stuff like glass 😁

@@MrJakob Didn't wanted to pressure you! I know how hard it is to find time these days. I discovered the series last week and have watched most of it and really liked the approach/commentary and of course the project. I just wanted to show interest to motivate you! Take your time

On macOS you can open this “Goto Folder Dialog” in any file picker dialog by pressing Command+Shift+G

That’s great to hear. Should you decide to publish your work on GitHub or the like, I would love to take a peak at it 😁

I haven’t yet started implementing cylinder intersection yet. Therefore I did not think about this properly yet. Once I am at that point, I hope to come up with an understandable explanation ;). Should I fail at that, please feel free to ask the corresponding question again, at that point :).

Yeah I guess so. However having png as an output format is at least a nice option. Now anybody can choose 😁

I guess that could be done as an optimization. It most likely would even be one, which makes a difference. However at the moment I think the raytracer is fast enough. Once I deem optimizations necessary, I would profile the code and start optimizing the parts that have the greatest impact first. Thanks for the hint however, I will keep it in mind.

Copilot is kind of an amazing machine learning system. However I usually know quite clearly how I want the code to be laid out, therefore something like autopilot kind of irritates me, more than it helps. But that is personal preference I guess.

@@MrJakob i agree with the code style comment, but the cool part about copilot is that after a bit of time on a project, it starts mimicking your code style.

@@ramongonzalezfernandez8904 that is interesting, maybe I need to check it out again some time. However I am still not that sure, if using copilot will hold up legally, as you are technically using someone else’s code, which might be under a license incompatible to yours. I know that there are different legal opinions about that out there. I guess we will only be sure once there is a big legal battle on that basis, to see who wins ;). Nevertheless it is an impressive technology.

I actually refactored the matrix implementation later on to use const generics as well.

Thank you very much. I hope you will enjoy the rest of the videos as well. Don’t hesitate to post more comments should you have any input or questions.

Thanks for the kind words. Please keep me posted, while you progress through the series, if you have any thoughts or questions.

As I haven’t tackled chapter 11 yet in this series, I can’t help with this, yet. However once I am implementing chapter 11, I will of course post the videos about it. Maybe those will help you to get an idea of what is going on then.

Yes, TH-cam will remove posted links, as it assumes it is spam.

Yeah things like that happen 😅