Can C actually do Perfect Bézier Curves?

Рет қаралды 20,529

Күн бұрын

Streamed Live on Twitch: / tsoding
Enable Subtitles for Twitch Chat
Panim Playlist: • Panim
References:
- en.wikipedia.o...
Support:
- BTC: bc1qj820dmeazpeq5pjn89mlh9lhws7ghs9v34x9v9
- Pay for my VPS: zap-hosting.co...

Пікірлер: 79

@jgfjfgjfhjf 4 ай бұрын

honestly, you've successfully indoctrinated me into wanting to use C

@multicoloredwiz 4 ай бұрын

He's very u happy he's done so though. I think

@MrMultiDesigner 4 ай бұрын

he have a pinned tweet that says if u going to learn C because of him, you're wrong

@jgfjfgjfhjf 4 ай бұрын

@@MrMultiDesigner usually, people who have twitter don't know they're wrong. Turns out it can be the other way round too

@OneMilian 4 ай бұрын

I use C since 4 Years nonstop, i always love over analyzing different binary object dumps. It literally keeps me my whole free time occupied!!!

@saogregl 4 ай бұрын

He successfully convinced me C is the simplest language there is

@RickeyBowers 4 ай бұрын

There is are many good topics along this trajectory of development, well done.

@ramilefu 4 ай бұрын

Takes 20 minutes to make sure we all have the appropriate pre-req knowledge before diving in. What an absolute Chad.

@dkaaakd 4 ай бұрын

This implementation has a flaw: editor displays curve (x(t), y(t)), but animation engine uses it as (t, y(t)). Essentially, during "animation" we assume that dx/dt = 1, but it is not the case for bezier curve we're drawing (one can note that projections of dots onto horizontal axis have different density).

@TsodingDaily 4 ай бұрын

Yes, I know that. These are just minor details that can be figure out later. What I needed is an editor.

@starup4960 4 ай бұрын

Damn I wanted to comment that but didn't wanna look stupid in case I was wrong, so I spend the last couple hours programming a little animation to convince myself first 😭

@TsodingDaily 4 ай бұрын

Didn't you have fun programming tho? ;)

@Czeckie 4 ай бұрын

i dont see how to fix that without solving a cubic equation each frame

@starup4960 4 ай бұрын

@@Czeckie I did this: while (Math.abs(p_next.x - target_x) > threshold) { while ((p_next.x - target_x) > threshold) { t -= step p_next = c_bezier(t, p1, p2, p3, p4) } while ((p_next.x - target_x) < -threshold) { t += step p_next = c_bezier(t, p1, p2, p3, p4) } step *= 0.1 } Pretty sure that's neither "technically correct" nor fast, but it sure looks convincing enough! Assumes the x coordinates of the points are increasing though.

@Daniikk1012 4 ай бұрын

For Tsoding (If he still didn't find the answer himself) and others: About Inkscape being light: the setting is per document. Every time you open a new document, it creates a copy of the "default document", so to change the default background, edit the default document. About the const local variables being able to be used in static variables, apparently (Just learned this) it is because static variables that are LOCAL to the function, even though static, are initialized the first time the function is called (I have no idea how that's implemented, does it generate a flag and initializes when the value is not set or something?). You also should be able to do that at global level, but it didn't work because Tsoding trying to create an array of dynamic size, not dynamic initialization. Dynamically sized arrays are not allowed in statics, even local.

@hwstar9416 4 ай бұрын

I think it's a gcc extension. tried it with msvc and tcc and didn't work. I think static is initialized at startup not when function called for the first time.

@Daniikk1012 4 ай бұрын

@@hwstar9416my bad, the standard does specify that statics are initalized at startup, before main (I was confused because in C++, they are guaranteed to be initialized the first time the function is called), so it seems to be an extension. What's weird is that it works even with -ansi -Wall -Wextra -pedantic, which SHOULD turn all extensions off.

@charlieking7600 4 ай бұрын

Theoretically telling, C is Turing Complete, so it can do. Practically telling, it's made in Quake 3 source code. Most of curved surfaces in game are based on Bezier curves (they're called "Patches"). These are used in many other games, including Doom 3 and Xonotic.

@blackhaze3856 4 ай бұрын

Can your Adobe Illustrator do that?

@Paulo-c3n 4 ай бұрын

(he is on linux) xdddd

@OneMilian 4 ай бұрын

Honestly I want a drawing Pad because of him. I always draw my ideas and write stuff and ideas about it. Mypaint and gimp could be alot of fun for me. I am a nerd because i love writing my own manuals

@OneMilian 4 ай бұрын

Man honestly all you commenters are so cool who know tsoding!! Im pretty depressed because i have nobody in real life to talk. I cant life like this no more, i need to find an exit. I will never give up

@phantomforcesisdumbgame6964 4 ай бұрын

I just realized the subtitles are chat wtf thats so smart

@esepecesito 4 ай бұрын

My 2 cents: 1) ABSOLUTELY there is mathematical gate-keeping BS everywhere. The goods ideas are (a) not so many (b) very simple 2) I cannot recommend enough the video about Bezier Curves of Freya Holmer

@sadiewopen 4 ай бұрын

For moving the head you should try making your curve go below y=0 before going up. That way it would look like it has some momentum

@mattshu 4 ай бұрын

Thank youuuu

@yehoslavrudenco4549 4 ай бұрын

I don't just get the difference between и and ы, these are two completely different sounds I cannot really confuse, I'd rather confuse t and d, or p and b. And it was surprising for me when a Russian speaker said that my и is actually an ы.

@afterschool2594 4 ай бұрын

Man your explanation for bezier curve is the real how knowledge should be explained by a real intellectual not a gate keeper.

@bukitoo8302 4 ай бұрын

the thing is that int N=2; char array|[N] defines a memory allocation, and not allowed. but int N=2; and const char N2 = {N}; is used as an initializer... and this is valid

@hwstar9416 4 ай бұрын

its actually not standard C. it's a gcc extension

@ovi1326 4 ай бұрын

you can cast a function pointer to void* just fine in iso c (c99 onwards at least) but you have to do the cast explicitly like `(void*)func` as for static arrays requiering comptime initialization I have no idea why that doesn't work for array sizes (maybe the compiler is interpreting it as a variable size array based on passing an ident associated with a variable even tho it's comptime known?) also c23 introduces constexpr which should just work as expected

@ovi1326 4 ай бұрын

Also it's defined behaviour to take a typed pointer convert it into a void* and cast it back to the same type, vtables in wasm work the same way they do in x86. Now taking a function pointer casting it to void* and then casting it to char* to read the assembly is undefined behaviour.

@thebatchicle3429 4 ай бұрын

@@ovi1326I don’t think it’s portable to cast a function pointer to void* because they may have different sizes

@ovi1326 4 ай бұрын

@@thebatchicle3429That could be the case but usually the function pointers are either the same size or smaller than data pointers (like in the early days of wasm64, I did a quick search and wasn't able to find any other platform that did thay). And it's defined behaviour to cast UINT32_MAX to size_t and then back to uint32_t, there is no way overflow would happen.

@anon_y_mousse 3 ай бұрын

You know, just reading the chat comments can be hilarious sometimes. Like the people arguing over compiler implementations and pushing to answer a question that's answered in the standard document by looking at implementations which may not be relevant to their own usage anyway. Compiler Explorer and speed up certain testing, but no compiler should be used as a replacement for reading what it says in the actual standard. I still don't get why so many people have problems with accidental assignments in conditionals because the majority of the time you're not testing for flat equality, but then I also don't make such mistakes when writing code even when using flat equality tests. Although, towards that end in my own language, I did make it so that simple assignment in a conditional isn't possible with just = and you'd need to use := to force an assignment. I definitely agree that mathematicians using complex formulae with "unreadable" symbols is gatekeeping. I haven't used much complex math in my career, but what I have used was much easier to understand when written as code than as a squiggly formula. One final thought I had regarding hot reloading and v-tables is that the method that most implementations aim for is an optimization technique and given when C++ was first developed it makes sense because even at millions of function calls on the older hardware that it would've been run on, it's possibly too much to ask of the hardware. However, in the current day, they probably should use the indexing method because even at billions of calls it's still relatively speedy and allows for hot reloading.

@jonathanriessner4389 4 ай бұрын

Wikipedia is actually gatekeeping the hell outa their information. for me its one of the most important sources, but thats because I have professors that explaining the shit whats on there. without the privilege of my education I wouldn't have a chance to understand shit on this "platform "

@zerocase1092 4 ай бұрын

It's really not, it's the mathematicians. Someone can probably add this explanation since Wikipedia is community driven.

@hussein-alemam 4 ай бұрын

Wait is the subtitle actually the twitch comments?

@phantomforcesisdumbgame6964 4 ай бұрын

bro i just realized that lmao its so smart

@gutoguto0873 4 ай бұрын

what switch comments

@isaklovas 4 ай бұрын

@@gutoguto0873 he meant to say twitch

@hussein-alemam 4 ай бұрын

@@isaklovas yes, I edited it

@gutoguto0873 4 ай бұрын

oh ok

@bbq1423 4 ай бұрын

42:47 Can't do it at compile time since N is not static. Compiler can't know if someone somewhere casts away the const and modifies it.

@SaidMetiche-qy9hb 4 ай бұрын

I think it would be cool to data drive panim, and have an editor for the "setting" file for it which can allow to configure the interpolators and tasks

@TimTeatro Ай бұрын

10:00 I respect you very much, but I have to disagree: the complexity of mathematical language is not due to gate keeping. That's like saying APL or C++ are gate-keeping languages. I think, if you had to reboot mathematics as a green-field project, starting from set-theory (or some other fundamental basis) and work your way to algebraic and differential geometry, combinatorics and statistics, you'd end up with something just as scary looking. I think every programmer has done that thing where we look at a solution, say 'that's way too complicated, I can do it better', then we dig into it, see all of the edge and corner cases and 6 months later we have almost identical code to the original and find ourselves saying "oh, now I see why it was done like that."

@UnrealCatDev 4 ай бұрын

what have we learned today? Being polish is cheating therefore I am a cheater.

@chexi 4 ай бұрын

полный кайф, классическое программирование. жаль понимаю 0.1% от всего)))

@SadShiry 3 ай бұрын

SUPRAAAAAAA

@diegorocha2186 4 ай бұрын

Blululululululululu indeed!!! 38:00

@puncherinokripperino2500 4 ай бұрын

do NURBS

@DrSpooglemon 3 ай бұрын

How much Adderall does this guy get through?

@magnum21th.c 4 ай бұрын

😂gatekeeping❤

@rudy-ishikawa 2 ай бұрын

I'm Brazilian and I always have trouble when I hear that word, for me it sounds like " gaykeeping " 😅

@mokopa 3 ай бұрын

What you call "gatekeeping" is actually just "rigor"

@TsodingDaily 3 ай бұрын

Does not change anything. The worst communicators always sacrifice clarity in the name of "rigor". Because for them it's just a gatekeeping tool. We are humans at the end of day. Leave your rigor to computers and automatic proof assistants. We already automated this shit long time ago.