cool shaders edition/gedg/ Wiki: https://igwiki.lyci.de/wiki//gedg/_-_Game_and_Engine_Dev_GeneralIRC: irc.rizon.net #/g/gedgProgress Day: https://rentry.org/gedg-jams/gedg/ Compendium: https://rentry.org/gedg/agdg/: >>>/vg/agdgRender bugs: https://renderdoc.org/Requesting Help-Problem Description: Clearly explain your issue, providing context and relevant background information.-Relevant Code or Content: If applicable, include relevant code, configuration, or content related to your question. Use code tags.
>>107207725I code in high level languages all day at my job. Let me be a caveman reinventing the wheel in my free time.
>>107207863you can even remove half of keys from your keyboard while coding, as an extra challenge. It makes the same amount of sense (unless you are writing the C++ compiler and need to stub it somehow before it self compiles)
>>107207899>It makes the same amount of senseNo it doesn't, you just have some weird hate for C.
>>107207899>you can even remove half of keys from your keyboard while coding, as an extra challenge.I do this. It's like driving manual, really lets you feel that extra control
>>107207899let him use void* for everything if he wants to why do you give a shit, you have no stakes in his engine, day jobs are where you can be obsessed about practicality
>>107207927>>107207920>>107207913none of the C fags given any explanation for while their kind of handicapping programming makes more sense.I guess it is a tie with team of programming with keys removed (also no reason to do it).
>>107207964I gave you straight away the most simple explanation, I just like it. It's fun. It's enjoyable.If I wanted the most easy optimum shit to write I would prompt some javascript engine with grok.
>>107207964I'm a C++ fag and I get it. C compilers are way faster, there is way less complexity, very easy to use. I'm never going to use it for a project when C++ exists but people who pick it are not retarded for preferring simplicity.
Different anon, but>>107207490>You make tasks x2 x10 x100 more expensive to do (in your time) and more buggyNot really.>for NO FUCKING REASON AT *ALL*.The reason is that I like C and hate C++.>>107207725>C++ - the same liberty, and then x10 more because you don't spend 90% of time doing malloc / free and trying to manipulate strings and arrays like a fucking RETARDED CAVEMAN so you can go on and write actually more complex and cool programs.This doesn't match my experience writing C. I use sds strings and kvec.h. It just works.
>>107208027>I'm a C++ fag>C++ compilers are hard to uselol sure you areCompilation speed is an "issue" if you structured your code in some silly way, or have PC old enough to join the army.
>>107208016>I gave you straight away the most simple explanation, I just like it. It's fun.glad to clarify there is zero reason to do it
>>107208067Having a good time is a valid reason by itself.Why would you even attempt to make your own engine if it wasn't for that one single reason that you enjoy writing your own code and reinventing what has already been done.
>>107208060I use c++23 and modules, it's improved a lot and it's faster but it's not as fast as C, usually it slows down because people do shit like use nlohmanjson in single header mode and then include it everywhere. I use conan / cmake / clion. I've been doing C++ for ~8 years now and work on a mixed C / C++ codebase for work. It's a different paradigm entirely but C is not "deprecated" and it's arrogant to claim so. You sound unhinged let people use whatever lang they want.
>>107208099>Why would you even attempt to make your own engine if it wasn't for that one single reason that you enjoy writing your own code and reinventing what has already been done.1) actually be better at something (that matters): e.g.: could be more open, and most portable to any system - for example.2) or if really aiming at reinventing - sure - but why reinvent it the wrong way, one that is stupid by not reducing productivity, for... for the lols?Ok then why not program it without procedures? copy-paste entire functions into one another, use longjmp or something. or write a state machine.Or avoid loops and copy-paste the code?>what? oh that would be crazy, unproductive and introduce bugs that grow exponent-...ohh...
>>107208162>I use c++23 and modules, it's improved a lot and it's faster but it's not as fast as C,I use c++14/17 and no modules, and the compile time is not an issue (except first compilation of big program - you wait the first time).More cores / better CPU does help.>usually it slows down because people do shit like use nlohmanjson in single header mode and then include it everywhere.right. do not do that stupid stuff, indeed.Perhaps in the end the "forced" isolation of program modules and not including all this shit has some long term benefit in the code?Maybe real friends are the #includes we commented out all the way.
>>107208190>Ok then why not program it without procedures? copy-paste entire functions into one another, use longjmp or something. or write a state machine.>Or avoid loops and copy-paste the code?Yes, why not. All those things can be fun. Programming is a game by itself, a puzzle, an enigma to solve.I don't know if you've studied music a bit but J.S. Bach for example used to give himself tons of very strict arbitrary constraints for his compositions. Yes, for the lulz.
>>107208239>Programming is a game by itself, a puzzle, an enigma to solve.Only at the beginner -> intermediate level
>>107208266I have only three years of job experience which makes me qualify for that threshold.
>>107208239>Programming is a game by itself, a puzzle, an enigma to solve.if you speak for yourself.I rather enjoy creating interesting/good software.Not to mention, that is what all end users expect in the end.
>>107208282Even if I tried like you say I would probably end up being the single end user of my own engine.
/g/ still arguing about languages, it's all bullshit just use what you enjoy.Check out my pong clone made in Raylib.https://aussie114.itch.io/yet-another-pong-clone
started with some sphere physics
>>107209112Are you doing proper physics or is it just based on distance and radius?
>>107209170If it's a sphere then "distance and radius" is proper physics
>>107206715>Rationale: Splinter Cell Chaos Theory ran on a single-core 733MHz Pentium 3. Is your game more advanced than Chaos Theory?Yes my game is very advanced it has over 100 objects with physics in the scene. It already requires more than Pentium 3.I plan to have 500 objects with physics so I need multithreading.>>107206599I do not accept input lag.>>107205953I will not hire you because you are a noob. Noobs are useless. You need to create projects on your own first, to stop being a noob.
>>107209216This thread was better when you weren't here
>>107209170I'm doing "proper" physics in principle. I went through the xpbd paper and implemented that, so I've got the basic framework to apply forces to objects, find collisions and then solve constraints by adjusting the position and then the velocity.However, i can only find sphere collisions for now lol. For cubes and other models I need to implement something to find those collision points first (probably GJK).
I'll post a webm once I get a decent one, still figuring out how to convert from OBS to webm without losing fidelitybut anyway, figured out how to kind of hide the screen space look. I got completely filtered trying to project it to world space, so I tried something differentbasically, the shader already uses 2 gaussian blurs as part of the effect to create the edges flow texture (I think the real word is called line integral convolution)so what I did was create different threshold weights, and instead of just white brushes lines across the image, there can be up to 3 different colors dependent on threshold. It'll also change intensity of the line based on luminance value and distanceso for large areas of bright color, like the water, skybox, sometimes ground. it's pretty noticeable that it's screen space. But as long as there's changing color/luminance/distance *somewhere* in the scene, it kinda masks and hides it. each color also has its own intensity setting and can adjust based on luminance/distance so each color can have different weighting.still not perfect but I like how this looks
>>107203967You could obsess over multicore scalability, but realistically is any hobbyist from this thread going to need blazing fast performance?
>>107209404Multithreading is the biggest noob trap for engine developers because you really don't need it and doing it well is pretty complicated
>>107209421It's both necessary for good perf and easy. Why do noobs love to speculate on noob traps?
>>107209492What are you doing that needs multithreading
>>107209508Anything? 4 cores = 4x perf = 4x as much shit on screen.
>>107209517Graphics are handled by the GPU not the CPU
>>107209529Physics eg
>>107209548Physics doesn't parallelize well, worst case scenario is usually totally serialAnd a single thread can literally handle thousands of objects, so I ask again what the fuck do you need mulithreading for
>>107209570physics parallelizes fine. "worst case serial" so this hypothetical worst case is as good as if you didn't parallelize at all. AI is another example.
>>107209611>physics parallelizes fineNo it doesn'tThe worst case, where every physical object is touching each other, is actually really common in a typical video gameLike if you have a giant pile of ragdolls the parallelism isn't gonna save you
>>107209624Did you forget about physx anon? You can do it with better performance it's just hard.
>>107209673What about physx?Solving physics entanglements in a stable fashion is a serial task, very smart people have tried to change this and haven't managed it afaik
>>107209624>every physical object is touching each other>Like if you have a giant pile of ragdollsyour game only has a giant pile of ragdolls, and nothing else? If I have two giant ragdoll piles, suddenly parallelism helps again.>is actually really common in a typical video gamecommon in test demos maybe. What kind of dumb level design would that be?
How necessary is it to learn shader autism for gamedev? As opposed to just using premade shader resources from a marketplace or a free repo.I'm not a math guy so I struggle to get my head around a lot of concepts. I see watch tutorials of what's supposed to be the basics and they're making all kinds of complex spaghetti node networks with dot products, powers of, 1-, multiplying this, subtracting that, doing something with the texture coordinate of a normal. Fries my brain lol.
>>107209699Most video games center the action around the player and stuff the player isn't interacting with are put to sleepSimulation type games where this isn't true benefit more from parallelism
>>107209708>I see watch tutorialsMeant to say I watch*
