Aww thanks. Yeah shader coding gives instant feedback, thats one of the things I love about it.

aww thanks man!

Yeah! Go go go! :)

I'm glad you like it! If you are new to shader coding then I suggest my beginners tutorials to get you up and running.

Oh Yea watched them ALL & followed.....& eagerly awaiting more......imo its one of the few areas that presently seems to lack abundant content as far as vids/youtube TUT's. If you don't mind, Where/how did you learn it primarily ? university? books/self-taught ?

I learning all of this by just looking at a lot of other peoples shaders and then playing with my own. There is still plenty of stuff I have to figure out. I wish there were some good video tutorials to learn from ;)

at least you are walking the walk !

Awesome. Thanks for watching!

Glad you liked it!

Thanks!

Thanks for watching!

Thanks!

Awesome!

Great!

Aww thanks! Its people like you that make the internet awesome!

Yes you are right. I do these off the top of my head and sometimes I get it the wrong way around.

​@@TheArtofCodeIsCool I see, it happens to the best of us. I noticed it when I drew the vectors to understand the function. Keep up the good work, your channel is a gold mine 👍

Thanks! And thank you for watching.

:)

Yes. You are right :)

nope, the aspect ratio change does not change the fact that the boxes go from -.5 to +.5 even if you didn't do any aspect ratio correction, the boxes would STILL go from -.5 to .5, they just wouldn't be square. If ever you get confused, its always a good idea to just output your uv coordinates as colors to get a better picture of whats going on. Check my video on KIFS fractals for a good dose of UV visualization.

You got it. :)

I explain a version of the distline function in 'the simplest 3d' though this version is different. In this case:The formula for the line is a+ab*t which in plain english says, start at point a, and then go t steps in the direction of b. This way, no matter what t you pick, you'll always have a point on that line. t = clamp(dot(pa, ba)/dot(ba,ba), 0., 1.) 1) dot(pa,ba) projects pa onto ba (and vice versa). This gives us a number that tells us the similarity between these vectors (1=they are the same -1=they are opposite, and anything inbetween or outside). Other people have made videos that explain the dot product. 2) /dot(ba,ba) normalizes the value obtained in 1 such that a point closest to a maps to 0, and a point closest to b maps to 1 3) clamp 0 1 makes sure that the value doesn't go out of the 0 to 1 range (the a to b range) because we want the distance to the line segment, not the distance to the infinite line. 4) length(pa - ba*t) -> pa is the position of point p with respect to point a, ba*t is the position of the closest point on the line segment with respect to point a. pa-ba*t is the vector from point p to the closest point on the line segment. We take the length of that to get the distance to the line segment. Perhaps I should have explained that in the video but it is already a long video and it would have added 10 minutes ;) Hope that helps!

That's perfect. Thanks for your help!

Nevermind, rendering to a framebuffer object with quarter of the screen resolution worked! I got a nice 60fps now :)

Nice!

There is an explanation of a different approach in 'The simplest 3d' but here is a quick explanation of this approach. t = clamp(dot(pa, ba)/dot(ba, ba), 0., 1.) dot(pa, ba) is a measure of where the closest point to p, that lies on the line segment is. This value will be 0 when point a is the closest point on the line to point p (when pa and ba are perpendicular) and it will be the square distance from a to b when point b is the closest point on the line to point p. We want to normalize that so that it is 0 when point a is closest, and 1 when point b is closest and anywhere in between 0 and 1 if some point in between a and b is closest. To do this, we just have to divide by the distance between a and b squared, which is dot(ab, ab). If you don't see this, just write out the dot product: ba.x*ba.x + ba.y*ba.y You can see that this is almost pythagoras, its just missing the square root, hence the distance squared. So now we have a number that is 0 when point p is closest to a and 1 when its closest to b, but what if its closest to a point that lies outside of the ab line segment? Well then it will either be smaller than 0, or larger than 1. We want to get the distance to the line segment, not the infinite line, so we have to restrict our value t to the 0 1 interval, hence the clamp.

@@TheArtofCodeIsCool OMG, THAT WAS SOOOO CLEAR AND EASY EXPLANATION!! THANK YOU SO MUCH!!

Not sure, as I have never used that particular editor but maybe doesn't support loops? You could 'unroll' the loop, which is basically just pasting the looping code multiple times.

@@TheArtofCodeIsCool I will try. :-) Still strange that this seems to be editor behaviour. My thought was, a programming language is a programming language is progr... Keep up the great work!

Because the brightness formula is 1/length(j) or 1/(the distance from the current pixel to the glowing dot). Because of the size of our boxes and dots, that distance is smaller than 1 so you get a brightness that is 1/(something smaller than 1) so you get a brightness value that is larger than 1 which is white. Another way of looking at this is that the glowing dots are too large for the boxes they are in such that we only see the white hot centers. What I'm doing here is effectively scaling them so you can see the whole dot.

In order to speed things up, and keep it parallel, we only consider the box the current pixel is in, and all of its direct neighbours. That way we only have to draw at most 9 lines per pixel.

Michael Pollan

To be honest, I haven't given it much more thought than 'does it look good'. These days I actually prefer a method that does not use sine because the sine version gives bad results on some older hardware. I'm not familiar with the LCG algorithm you mention but a quick glance at wikipedia makes me think that this is a sequential algorithm where the new value is computed from the previous value. In shaders you can not do that because all pixels are calculated in parallel, at the same time.

@@TheArtofCodeIsCool Aaah you're right. You can immediately tell that I'm not a shader person :) Yes, the sequential nature of it is actually preferable, because you can 'seed' it in normal circumstances, but obviously that doesn't work with parallel execution, and especially when fragments don't cross-talk and have no common memory. Though I wonder if fragments could, in theory, share a memory segment similar to how they have a common access to textures? Is this why people use a noise texture sometimes, to guarantee a better distribution? (And is it faster when compared to sine? Probably not.)

hmmm. I'm not sure but it could be that you are trying to compile to an older version of opengl that doesn't allow dynamic array indices? Really not sure though.

@@TheArtofCodeIsCool Yeah WebGL version 100 do not accept dynamic index, i had to update the code to version 330. But you can get the same results in version 100 by accessing the index by hand. Thanks for your time ! :)

Hmm, works for me. Maybe try a different browser?

Great question. It is because the noise function always returns 0 when you enter 0 as a seed. This makes it that the cell with id = 0,0 will have 0 speed for both x and y.

@@TheArtofCodeIsCool Ooo, I noticed that and worked it out too! I'm getting better! However, I still got a bit lost. With this method don't you draw some of, if not all the lines twice? Why no just draw lines from p[4] to the right (or whatever) ... actually, stop. I'm going to try it out and see if I can do it. I'll be back.

Aww thanks man. I really appreciate the gesture but for now I'm happy if you can like / share my videos!

Well I've never had one so I guess I don't know. :) I'll do the second part of my fractal explorer soon which might be closer to the mark.

The Art of Code all the smartest people on earth have taken LSD! You should. It will change your life, you will not regret it. 👍😎

imo intelligence doesn't have much to do with it statistically. They reveal the perspective that there does exist other perspectives(all equally valid). I used to advocate its use to others in my youth too, thinking that one has never really been fully awake/aware otherwise. There may be some truth in that but don't EVER do it because your being pressured into it. & if you do decide to try it know that environment/atmosphere/company can be crucial. I honestly wouldn't trade my psychedelic experiences for ANYTHING(well maybe a random half of them for a few million $ LOL). However it is NOT for everyone.

realcygnus you are wrong, the smartest people do take LSD regularly. kzfaq.info/get/bejne/g7ZxecSZlc-4dIE.html People like Steve jobs, bill gates et al in Silicon Valley take LSD, it's no coincidence, so I'd say statistically it's the realm of smart people. Of course psychedelics isn't for everyone, but it's the domain for people who are Intelligent enough to want answers. Enthogens have always been the domain of truth seekers and smarter set.

Hanniffy Don't get wrong I am a fan(generally speaking). Not to put words in your mouth but drugs won't "make" you smart. It is true that psychedelics can(& usually do) show you that there is a Larger Reality, but from there its still up to you to evolve the quality of your consciousness which is more of a journey/path as opposed to an endpoint/goal. I think perhaps it could be a majority in the art world(like music). Its also important not to confuse subjective Enlightenment with objective intelligence. imo Gates is a bad example in that he would use a much larger percentage of his assets to benefit humanity if he was truly "enlightened" from his experiences. & btw it was Wozniak who was the actual "Brains" behind Apple(though I'm sure he tripped too since it was culturally common at that time/place). But imo these are both bad examples in that, the ability to make $ is a demonstration of neither spiritual enlightenment or intelligence(in a general sense). imo Enlightened ppl would be like Jesus, Buddha, Gandhi, or more currently the Dalai Lama & Tom Campbell. & the People that I think are thE most intelligent are typically physicists/mathematicians/information theorists etc. Like Newton, Einstein, Feynman or more currently Ken Tompson /Dennis Ritchie et al.(c & unix) or Linus Torvalds(Linux) & the numerable current theoretical physacists like Tom Campbell(on both sides of the spectrum !), Kip Thorne, Leonard Susskind, Jaun Maldencena & Nima Arkani-Hamed etc. & Many of the psychedelic philosophers/pioneers/advocates like Allen Watts, John C. Lilly, Terrence McKenna & Timothy Leary etc. were certainly geniuses in their own right. & you're right Psychedelics has ALWAYS been a(but not thE only) Path of truth seekers & BIG thinkers. There are just too many ppl to list in every category ! Anyway I Just think you MIGHT be in a belief trap, in thinking that its more statistically significant than it really is(specifically regarding intelligence). Academia is full of very bright people that never even considered it & ironically even some of the smartest ppl can have some of the dumbest ideas about life in general & Vice versa. I do respect your opinions however.