Scratchapixel 4.0, Learn Computer Graphics Programming | Hacker News Reader

Scratchapixel 4.0, Learn Computer Graphics Programming

scratchapixel.com

270 pointsmagnio54 comments

An interesting excerpt from the section “3D Computer Graphics Primer”:

> The understanding of light and how we perceive it has evolved significantly over time. Ancient Greek philosophers posited that vision occurred through beams of light emitted from the eyes, interacting with the environment. Contrary to this, the Arab scholar Ibn al-Haytham (c. 965-1039) introduced a groundbreaking theory, explaining that vision results from light rays originating from luminous bodies like the sun, reflecting off objects and into our eyes, thereby forming visual images. This model marked a pivotal shift in the comprehension of light and vision, laying the groundwork for the modern scientific approach to studying light behavior.

Indeed! If you are interested in reading more about this, I recommend the article “Light Through the Ages: Ancient Greece to Maxwell”[1]. It is a very fascinating article written by J J O’Connor and E F Robertson in 2002 that takes us through a journey of how our understanding of light has evolved over the last few millennia.

[1] https://mathshistory.st-andrews.ac.uk/HistTopics/Light_1/

> "...vision occurred through beams of light emitted from the eyes, interacting with the environment"

How did they explain lanterns and torches in this theory?

Until you can observe that light doesn't travel instantaneously, it would be consistent to believe that a beam leaves your eyes, bounces off a surface, and informs you (as though it were a limb, perhaps) when it hits a light source.

The more I think about it the weirder it gets.

It’s day time, the sun is up, it’s a cloudless day. Ambient light is bouncing all over, including into a south facing window (i.e. no direct sunlight.) How does the Eye Light Emitter do its job in this case?

Now, I don’t mean “this concept is impossible” because that’s clear. I’m genuinely curious how this would have been explained by someone at the time.

Vision rays emanate from your retina, get refracted by the lens, exit the pupil, go outside the window, get scattered in the air, and - some of them - end up hitting the sun.

You should read scratchapixel really)

https://www.scratchapixel.com/lessons/3d-basic-rendering/int...

In the real world light travels from the light source to the eye. But this lesson xplains that simulating this with a computer is not efficient. A more efficient technique is to follow the path of light but in reverse, starting from the eye and waking back to the lights making up the scene. Again read the lesson(s) it's all explained and that's why the lessons are there in the first place.

Except that would never explain shadows. Why does my eye beam not let me see what is in shadow, if there is no such thing as light beams coming from a light source?

Because the beam coming from your eyes and hitting a shadow would never reach a light source (you could say it gets absorbed before it reaches a light source).

What do you mean lanterns and torches?

On the macroscopic scale, light rays can be replaced with vision rays, and everything works out the same. That's how ray tracing works. People thought eyes were doing ray tracing.

That's how we still do (simulation of) ray tracing (Whitted).

What do you think he means? It's a good question.

I remember a guy who had a novel theory (mostly as a joke, I think) that light bulbs don't emit light, they simply suck away the darkness. He literally called lightbulbs "darksuckers."

Roger Wilcox, from the 1990s?

http://www.rogermwilcox.com/darksucker.html

It was someone I knew IRL but I guess that's where they got it from

A far as I know, in raytracing, you bounce your ray from the surface's normal and if it hits a "light source" then it's brighter (by the inverse square distance to the source). Maybe they explained the same way, except they used a different name?

I think this is plausible. I guess they thought unlimited ray bounces were a thing.

That's why the lessons are online. You can read through them and everything is explained there. In ray-tracing light most often travels from the eyes to the surface and from the surface to the lights (direct illumination). Various laws (such as as the inverse square and lambert-cosine laws) are applied to simulate reality.

I can't recommend that maths history site enough, I've read through basically all of it multiple times and sent in many corrections over the years :)

As far back as Aristotle, anyone who put a moment's though into it realized that eye beams were nonsense. They are like modern superstrings - any theory that includes them works works just as well if you remove them and keep the rest of the theory.

http://web.stanford.edu/class/history13/earlysciencelab/body...

From the getting started section:

> Terminals are part of the MacOS since it is Linux-based.

This is not true. MacOS is based on Unix.

I noticed that too, then it mentioned that compilation should work on macOS and Linux, even though they use Windows. But then they compile to a .exe in the test example to make sure everything is working. This kind of makes it a non-starter for someone who doesn’t already have some idea what they’re doing.

>MacOS is based on Unix.

It't not "based on". MacOS is a Unix.

Correct. This could be modified)

Looks great! I really have to supress my "AI thumbnails means low quality AI slop" sensors though because I've heard of this site before and know it's high quality.

I would personally try to have some non-AI thumbnails. But that's just me.

)) The AI-thumbnails will be removed in a few weeks. Totally agree with you.

What are you replacing it with?

ML thumbnails.

Just remove the images now, until you have real ones.

I've seen several blogs/sites of decent quality and technical depth start throwing random semi-related generated images into their articles, and it just makes me immediately wary of BS. No images is better than AI-generated ones if you want your site to look like it was made with care and has good information on it.

The thumbnails are fine imo.

Scratchapixel and Rougier's Python OpenGL (https://github.com/rougier/python-opengl) are two books that i want to see completed one day...

I just noticed that all the thumbnails are AI generated, I think most of them look quite good, only a few need to be replaced in my opinion: "What Do I Need to Get Started?", 8, 25, 26 look pretty bad; I would have chosen the same style as "1." and "2." and used it for all images as they do not have the typical AI look.

Quality writing and sloppy ML-genrated thumbnails really don't fit together, but I saw you already announced that you will replace them with proper images, so this will be a nice resource overall!

I know it probably doesn't relate to the content; but my bs/ai detector sounded the alarm after seeing the abundance of probably midjourney-generated chapter images.

Scratchapixel is the real deal but yeah, I know what you're saying. The "house styles" of services like Midjourney are so widely seen in scams and spam that they have become an automatic red flag. Doubly so when the material is supposed to be about non-AI means of creating images - why aren't you showing the actual product of following the course?

I was also confused by that, it's a weird choice considering the articles themselves are full of original diagrams and illustrations that could easily have been used instead. The AI generated images are just visual noise.

I'd say your bullshit detector needs some readjustment. You can make nice images with AI, especially as chapter images.

AI-thumbnails will be removed soon.

I would say generating chapter thumbnail images for a non artistic publication is probably the best use case for AI.

It isn't, because people associate these kinds of pictures with low-effort auto-generated clickbait and scams.

"people" as in "some people on HN"? Because I think as broader audience wouldn't necessarily mind or associate those images with scams.

At face value yes, maybe. But low-quality content is increasingly associated with them, so the effective impact can easily get negative. I'd like to compare this to typical startup landing pages or doxygen documentations. Both can be spotted from a mile a way and are strongly correlated with padded and low quality content.

Scratchpixel existed way before widespread generative AI. The content is great.

Any alternative resources to learn Computer Graphics Programming from scratch and indepth?

Specifically regarding OpenGL, my starting point was https://learnopengl.com

Very similar to drawabox.com in the "feel".

have been waiting for the texturing chapter to be done for what feels like years now... i actually think it has been years.

Will work on that next then). It will be cool -- and worth the wait. Sorry if the project got on stall for some years. Life can be difficult some times and it was for me for all these years. Yet I put my heart into writing what's about 500.000 words now which is equivalent of roughly 8 books of 200 pages. All this for free since 2009). I understand your frustration, but hopefully with time and the community support (don't expect any of the corporations out there to help us in that process) we will get these topics covered. But as I am no millionaire, I can only work at the speed at which I can feed myself.

sounds good!, i do very much appreciate all that you have written so far. its just that texturing is the one thing i really do find difficult and its frustrating its not there :) -- thanks again for all your hard work.

amazing

Honestly - mixed feelings. This looks more like a way to build up academic/collegiate level knowledge on computer graphics than a way of familiarizing yourself with shaders/rendering techniques.

It is also woefully incomplete.

What would you recommend?

Hum to familiarize yourself with shaders/rendering techniques, do you see any other way but to build academic/collegiate level knowledge on computer graphics? Would love to understand how you intent to achieve results without knowledge.

Yuck, AI thumbnails.

Reply on news.ycombinator.com