cpu rasterization is blazing fast edition
>https://www.youtube.com/watch?v=U7_j5dSWkzg
>https://www.youtube.com/watch?v=2cOqAHzIfQE
who would've thought that keeping it simple is the easiest way to make game? i was surprised that it even ran on pentium 2 tho.

>>100043427
>blazing fast
compared to what?

>>100043674
to molasses

>>100043674
compared to people misconceptions, they tend to forget that simple instructions, cache, and memory prefetchers are used to the max due to the simplicity of arrays, as in pixel arrays, and that's before using things like SIMDs and better algorithms (dirty rectangle optimization).
people forget that emulators were the main users of cpu rasterization backends and it worked fine, you can emulate any console similar to NDS and older (and any game that was possible on those) in a stable 60 FPS with all the graphical effects and whatnot.
it's a fucking 2D game, people keep asking about game engines for 2D games, it's easy to get 10k entities on full screen with raster on the browser, you can even circumvent the issue of hd screens by using the backend hardware acceleration to scale the screen instead of scaling natively on the cpu. you can even have post processing shaders and it won't be any harder.
you can also piggyback existing engines tool chains and just export everything to yours and run it natively, you don't need to remake every tool from scratch, that's how you get to use that cracked unity version or pirated tool you didn't want to pay for without getting into trouble.

You know that there was a time before 3D accelerators where all games were done in software renderers right?

>>100043427
There's also llvmpipe, which is pretty fast and useful for debugging without potentially freezing your desktop if you fuck up.
GPU still has a lot more brute power though.

I think I'll rewrite my engine but this time I'll focus on what my game needs instead of trying to make a general purpose engine.

>>100044756
good call

>>100045404
after 3 years of writing my engine in rust I think I'm just gonna scrap it and use c# or java

>>100045415
why not stick with rust if you have 3 years of existing work?

>>100043427
>pixelshit

File: working-wheel-physics.webm (434 KB, 1024x600)

434 KB WEBM

3... 2... 1... liftoff!

>>100046340
those physics are indeed working

>>100046340
amazing. you should work for NASA

>>100043403
unironically now, has ANY of you EVER shipped game or even left a project in a playable state?
because to me this seems like /fucking around with rendering general/

>>100047643
I don't know about whether or not others here have finished a project, but I certainly have never finished a project in general ever and I feel attacked

>>100047643
Made pong for the last jam but got no time

>>100047643
unironically yes but it was a shitty phone game that I made like 10 years ago

nodevs rise up

>>100047643
Be the change you want to see

How does one brainstorm ideas for a game? I wonder if I should do the "Sakurai Way" where I should look at a game genre, deconstruct it, ask "what makes it fun" and "what doesn't", analyze the fun and "unfun" aspects, give one of the aspects a simple twist, and go from there.
I.e how he made Smash as a response to fighting games and based it around a "damage accumulation"/knockout mechanic; how he made Kirby Air Ride as a response to racing games and based it around making drifting into a "skill" with a "gauge", etc.

>>100047643
well yes i made a sokoban game with c and sdl, it could load arbitrary levels so it's as finished as it can be, but that was 10 years ago
now the closest thing i have was an FTL knock off where boarding other ships happens by sending a part of your own ship, oxygen mechanics still applies, so that's how you can transfer O2 to a ship without air.
but then i got filtered by game design and couldn't add anything of value to FTL.

What's the best programming pattern and what is it the event pattern? being able to fire an even that you can catch anywhere anytime is super handy, but makes everything a non linear clusterfuck

>>100048292
>a non linear clusterfuck
skill issue

>>100048536
>programming pattern
stopped reading right there

>>100047937
if you don't have an idea of what your game should be like and have to go through this process, your game will be soulless

>>100048292
the best programming pattern is having no patterns at all. your code should be flexible and not constricted to imaginary structures

>>100047643
No but ask me what I think about OOP vs Entity-Component System.

>>100047937
Sakurai is a hack. If you don't have a reason to make a game already don't bother.

>>100047643
I go here to shitpost but if I had a chance of actually making a community, I would stop visiting this site because the time I spend here is already too much.

I'm such a fucking mathlet holy shit
it's a miracle I've gotten this far

>>100049243
>>100048926
Then how do you come up with ideas for your games?

File: _b42d93ba-3d0c-4bbb-b881-(...).jpg (203 KB, 1024x1024)

203 KB JPG

>>100049243
>don't bother
>>100048926
>your game will be soulless

>>100049914
It is a mix of "I love this game, but I wish it had X" and, while playing other games ,thinking "this is a good idea I'd like to see in my game" plus the occasional epiphany.

>>100049946
That's kinda what's part of Sakurai's thought process when making his games though.

>>100049955
>>100049946
See https://www.youtube.com/watch?v=7VD0K1Sr5u0

>>100049589
All you need to know is dot product, cross product, norm, matmul and trigonometry.
It is that easy

>>100049914
I make what I want to play

File: _0fd4eeba-7d67-48e1-804e-(...).jpg (130 KB, 1024x1024)

130 KB JPG

>>100049589
>>100048951
.t

>>100049968
>>100049955
I don't really care for him anyway, because he has directed ... how many games, two? You want to pay attention to the man himself https://gamedevgaiden.neocities.org/quotes#Shigeru_Miyamoto

>>100050050
https://en.wikipedia.org/wiki/Category:Video_games_directed_by_Masahiro_Sakurai

>trusting the japaneses

>>100050070
>japaneses
it's spelled japanese my dude

File: _c66eac3e-d4d0-48f8-a7c8-(...).jpg (165 KB, 1024x1024)

165 KB JPG

Caught a live pic of the crabs whom reside in this thread.

>>100050098
jamp on dees nuts

>>100050061
So two games? Every kirby feels the same, and I wonder if the success behind every smash lies withing the inclusion of new characters or they do something novel. Anyway, is up to you. I don't think kirby is a bad game, just that Sakurai is a one hit wonder. If I were you I'd read every interview with Miyamoto, he always has great things to say about the art of making games.

>>100050029
I'm getting filtered by a problem that doesn't use any of this, it's that bad

>>100050098
I don't care how to say the name of your chinese cartoon makers.

>>100045463
Different anon but also a Rust veteran
Rust is great for the hottest parts of the flame graph and the most stable parts of a tech stack
It is not suited to parts of a codebase where the highest priority (even higher than performance at runtime) is being able to make changes quickly
Someday there will be an interpreted language with extensive hot reloading support and seamless Rust interop, and these two languages together will clean house and take over the game industry
That day has not yet arrived, so right now the best you can do is bespoke asset editors and maybe a shitty third-rate scripting system

>>100043403
/tpg/ won't ever respond so I'll ask here:
I'm looking to get a laptop and after like a decade of procrastinating also want to get into gamedev: Is there really any necessity to getting something "modern" for a gamedev laptop? I was looking at the 10th gen intel P15 workstations by Lenovo because they're in the $350-$450 range and still modern enough to for replacements parts to be pretty available.

I will write my own ECS today

>>100051263
post when you're actually finish a game with it, aka never

>>100050661
>extensive hot reloading
skill issue.
rust has incremental compilation. your program should compile instantly or you're doing something wrong.

>scripting
Why is this general so into this meme? Or is it just one poster?
Games are written in C++. Nobody actually does scripts for anything and every time they do it turns out a mistake.

>>100051386
>Games are written in C++
name 5 indie games made in C++ in recent years, and no, unreal doesn't count.

>>100051427
>name 5 things written in English, and no, books don't count

>>100051427
You're delusional

>>100051458
testaments, bills, recipies, menus, this thread

>>100051458
>>100051461
and you're retarded, unreal blueprints do not count as C++, even AAA studios rely on blueprint spaghetti to make games, how the fuck can you prove a small indie dev used C++ and not blueprints?

>>100050661
>It is not suited to parts of a codebase where the highest priority (even higher than performance at runtime) is being able to make changes quickly
so "Rust is a pain in the ass to write" is true. I'll pass on that shit, thank you.

>>100051327
Reloading compiled Rust code is an area of active R&D and I won't pretend to be more clever or experienced than the dedicated people who are already putting a lot of work into it
The most troublesome part is preserving state across reloads and there are gorillions of ways to do that, all with their own peculiar drawbacks that are essentially intractable for certain use cases
>>100051386
>Nobody actually does scripts for anything and every time they do it turns out a mistake.
Well memed!
In truth, even a "scripting" language suitable for games should have some form of JIT available, or even be capable of optimized AOT compilation into a dynamic library (I almost mentioned all this in my earlier post but decided not to bother)
The hot reloading part is especially difficult to design a language around and instantly bloats the scope of any project far beyond the amount of work it takes to make the actual game
So, sure, making your own scripting language is normally a dumb idea if the alternative is "just make the game" and especially if there are ready-made tools that can cut down your iteration cycle time by >80%, >80% of the time

>>100051808
It's for problem spaces where good code is a pain in the ass to write no matter what language you use and that's all I'll say in its defense

>>100051820
>Reloading compiled Rust code is an area of active R&D and I won't pretend to be more clever or experienced than the dedicated people who are already putting a lot of work into it
>The most troublesome part is preserving state across reloads and there are gorillions of ways to do that, all with their own peculiar drawbacks that are essentially intractable for certain use cases
You didn't understand the comment at all. Hot reloading is stupid and as you say in general intractable. All you need is save states and fast recompile.
>>100051820
>JIT
>AOT compile
Lmao. You don't need nor want scripting retard

time to make a new bingo sheet with hot reload, scripting, rust, and programming patterns in it

File: actually-working-wheels.webm (2.56 MB, 1024x600)

2.56 MB WEBM

The implementation is a mess and full of errors (like using wheel normal/tangent instead of surface for friction) but it works reasonably well.

>>100051981
truly the quadfecta of tinker trannyism and I guess vulkan too

>>100052028
That's really cool.
Great progress

File: lisp horse middle finger.png (131 KB, 900x600)

131 KB PNG

>>100051902
OK

>>100051427
Factorio is written in C++ and uses Lua for scripting.
They use SDL

>>100052140
>and uses Lua for scripting.
It uses Lua for mods. You really shouldn't worry about mods.
The game itself doesn't do very much scripting at all which should tell you something unless you're a tinker tranny that loves fantasizing about JIT AOT compiled scripting languages.

>>100051981
It's gedg bingo card. Still better than the agdg one.

>>100051386
its like 2 posters with a fetish for it
they have mental problems because they're projecting what they think makes a game a game onto other
pretty sad, really

>>100052480
oopsie, nevermind, i somehow thought he was talking about mods
carry on

File: 3841672_2-1290570974.jpg (230 KB, 1030x800)

230 KB JPG

>>100052028
Reminds me of Blockyard.

>>100051877
sad that rust trannies like to harm themselves.

holy fuck I did it
I'm a math genius

nevermind still a mathlet

>>100053248
>>100053548
bro...

File: Screenshot_20240417_234831.png (216 KB, 1920x1080)

216 KB PNG

Good morning frens!

I wanna make a roguelike. So I decided to just start building one.

Wish me luck anons

t. webdev

>>100053613
Have you ever played Rogue?

>>100053650
Not the original, no. I did play a fair bit of Powder, DCSS, and some Nethack.

>>100047643
I have but this thread doesn't like people who actually finish projects

>>100049914
>>100047937
If you don't naturally have ideas for games you want to make then that's concerning

>>100050661
>Someday there will be an interpreted language with extensive hot reloading support and seamless Rust interop, and these two languages together will clean house and take over the game industry
This is the ulimate "I think about making games rather than make games" take

Why is no one here competent

>>100053613
>https://www.youtube.com/watch?v=JxI3Eu5DPwE
here fren, it's from the author of the gameprogrammingpatterns.com website
don't forget to add juice, Cogmind lived by the juice, good luck!

>>100054112
why does no one here know how to report and hide posts that are extremely low quality

>>100054477
/g/ doesn't have mods

>>100051427
who the fuck writes an indie game in C++? Either you're using an engine and you're successful or you're making a hobby project to post on youtube. C++ is for on-the-spectrum script kiddies and AAA optimizer tech artist freaks.

>>100054499
It does but they only take action against people they lose arguments to

>>100054562
>who the fuck writes an indie game in C++?
Competent developers

>>100054578
competent developers who want to waste their time hyper-optimizing code and spending hours fixing niche library bugs instead of making anything.

>>100054644
C++ libraries aren't unstable, it's the gold standard for gamedev

>>100054644
You're some kinda python freak or what?

>>100054562
Actually, I forgot about LoliSim, he released the game with the code and it was C++, not that i expect anybody here to remember it.
But I'm 100% sure people are using C instead of C++, C++ just became a gigantic bloated monstrosity that either :
A - it's too complex for new people to get into and understand
B - the people who do use it are so tired of it they switch to something simpler (aka C) or change languages all together

>>100054562
https://www.youtube.com/watch?v=XOJ3Gg8lx1Y

this nigga right here.

agdg crabs bored again? Tired of worshipping ecelebs?

>>100054892
He actually released it? I can't believe I missed it.

File: 538.jpg (41 KB, 421x834)

41 KB JPG

>>100054892
>I forgot about LoliSim
about WHAT
wtf kind of game is that

Is /vg/ down?

>>100055068
A 4X game with an emphasis on the exploit and expand parts

>>100055048
on 8ch before the purge, i wish i didn't delete one of the version i had
>>100055068
2d gta game with fully simulated town and npc schedule where you get to kidnap and impregnate any loli in town, with a customisable dungeon and everything, truly a game among games.

>>100054923
why is he so cringey?

>>100047643
If only there were links to progress days in the op where you could check out and play games made by this general

>>100055439
>games made by this general
haha

>>100054892
I can vaguely remember terrible pixel art and vaginal circumference stats or something, that must have been almost a decade ago. i had no idea there was ever a playable version

>>100055621
>vaginal circumference stats
what

>>100054892
>A
not my problem
>B
not my problem
still using c++

>>100055621
you forgot about the accurate genetics algorithm and data based on race so you can breed the perfect sex organs stats.
goddamn the dev was a gigantic autist but i do miss his genius

>>100054562
i just realized that this guy said C++ is for skiddies, this website is amazing

>>100050661
It's called C + lua jit

What is the modern way of rendering unicode text? Is dynamic atlas generation with caching the way to go? I know I can generate the atlas on game/engine startup but that's only good enough for a limited character set, I need that chinese audience for my nonexistent game.

>Matrix types have a matrix stride (GL_MATRIX_STRIDE), which represents the number of bytes from the start of one column/row vector to the next column/row (depending on whether the matrix is laid out as column-major or row-major). The column/row-major ordering can be queried through GL_IS_ROW_MAJOR, which will be one for row-major ordering.
Are you fucking kidding me? How many variants of this shit am I supposed to implement? My elegant implementation is now ruined by lots of metadata and adding strides to data transferring.

>>100056019
Depends how correct you want to get
You can generate all the common chinese characters in an atlas

>>100056085
There are standard memory layouts you can use for uniform blocks which gaurantee a particular format
https://www.oreilly.com/library/view/opengl-programming-guide/9780132748445/app09lev1sec2.html

>>100056152
My float[16] being spread to four times the size is extremely confusing to me.

>>100056243
I'm pretty sure float[16] should be packed to 64 bytes under std140, same as mat4.
Also consider using std430 if available, which has better packing when using e.g. (vec3, float)

File: Untitled.png (261 KB, 1171x402)

261 KB PNG

Informative videos but I hate these clickbait titles and thumbnails so fucking much.
Like,
>I need a quick refresher for this rendering technique. There's a video about this. Let me look for it in my playlists
Then all I can find is just these nonsense.
https://www.youtube.com/watch?v=i6VVegoRuy0

>>100056855
Well I'm not clicking your shitty video. Summarize it.

>>100056855
I don't watch these kind of videos anymore, they are all time wasters. nobody ever watched these and became more productive as a result. in fact you'd be more productive putting that time into devving

>3rd party proprietary software
Not even once.

https://youtu.be/lwlRDhPZxLc?si=cGg-lvGbXE1flV3U

Any good c++ books?

>>100057575
yes

>>100056855
you think that's bad? I have a semi popular game and retards send me these kinds of videos 3 times a day because they think it would quadruple the performance

>>100051386
most unreleased indie games people do youtube diaries and post here are made in c++. the ones that get finished are java or c#.

>>100058109
Nobody makes games in Java
People make games in C++ and C#

>>100058250
Minecraft
kys

>>100058286
>games
>s
Plural words mean more than one ESL kun

>>100058286
exceptions don't contradict generalizations brainlet

>>100058295
>>100058297
Slay the Spire
KYS

>>100058306
exceptions don't contradict generalizations brainlet

>>100058322
my "exception" is the biggest indie game in history
now shut the fuck up, both of you got destroyed

>>100058333
Game/engine dev thread
Post your game/engine written in Java sir

>>100058333
Do you know how to count retard?
Popularity has nothing to do with it

>>100056105
idk what you mean by correct but yeah I think a 2K (or maybe even 1K) atlas is good enough, the hard part is the caching. I got a quick and dirty dynamic generation going but no caching could fuck me perf-wise.
My current strategy is to just FT_Load_Char the unique glyphs in a frame and generate an atlas of that, no caching just raw-dogging and the perf is pretty decent (0.6ms) for a frame with 6 different sized sentences. I guess I'll have to experiment with this

>>100058620
>idk what you mean by correct
Displaying all Unicode correctly is an enormously complicated task, but just rendering fonts for a game for the basic languages you want to support is much easier. You can fit all the characters in the Basic Multilingual Plane into one texture so long as your font isn't too big
https://en.wikipedia.org/wiki/Plane_(Unicode)#Basic_Multilingual_Plane

>>100047643
I have, but I don't consider myself part of this thread (despite it being an engine-deved game)

>>100058695
If you're referring to text shaping then I already got a basic shaping implementation with bidirectional text going which seems to work fine (at least for Arabic), but I guess there's always edge cases in some weird obscure languages, and if/when I get to that bridge I'll cross it.

>>100058796
Either implement something simple yourself or use an official solution, don't try and handle every retarded language rule on your own and make your own custom Unicode implementation
character width and kerning pairs are realistically all you need for languages people care about

>>100058286
the developer of this game WAS nobody, it sprang from the void carried by hatsune miku, the rule stands unchallenged

>>100058823
Yeah I'm not retarded enough to literally write everything from scratch. I'm using freetype with harfbuzz and sheenbidi to handle all that for me, so I guess all that remains is for me cobble some caching up and then move on with my life

File: 1467789713106.jpg (83 KB, 378x378)

83 KB JPG

>sly lisp code sometimes compiles but usuall doesn't

haha wtf
No clue what is going on.
(slynk-mrepl:send-prompt) in the repl works but it won't compile.

Btw I can render a window with some c bindings. I just wish the interactivity wasn't fucked. Maybe I should switch from sly to slime.

>>100054298
I saw that one! I also saw Josh Ge's "How to Make a Roguelike", Brian Walker's Procuderal level design in Brogue, Thomas Biskup's video about Ultimate ADOM, and a vid by Herbert Wolferson.

There's quite a few talks out there which touch different topics. I've been doing some research on those lately.

A roguelike sounds like the most fun type of game to make
No worrying about graphics bullshit, pure rabbitholing on game design and procedural generation

>>100059423
Eventually I do want to add some graphics and some spell effects. But for now I can keep that part very simple yeah.

Unironically Powder was the game that got me into the genre. It seems to have mixed reviews within the scene, but I found it very accessible for newcomers.

>>100051386
>Nobody actually does scripts for anything and every time they do it turns out a mistake.
[Citation needed]

File: 1c6.png (146 KB, 680x679)

146 KB PNG

>>100053248
>>100053548

>>100060858
People do use scripting languages but I agree with his sentiment that this general likes to embed scripting languages in their engines for no good reason

>>100044233
Yeah even in python or js software graphics is way faster than most people think.

>>100060930
I'd like to see you do a software renderer in Python

>>100060936
I did a few years ago. I drew this 3d animated butterfly mesh thing you could fly over some perlin noise terrain on a canvas in tkinter. The whole aesthetic was absolutely righteous and it ran fine on my slow 1.6 ghz netbook, maybe I can find the code and show you....

>>100060947
That means you have to blit over a million pixels
In Python

File: canvas.gif (1.3 MB, 1200x800)

1.3 MB GIF

>>100059531
>Eventually I do want to add some graphics and some spell effects
You can get a surprising amount of eye-candy with just plain text and colors. Checkout the old versions of cogmind and cave of qud before they started adding complex sprites or the samples of the ratatui lib(see pic)

>>100060964
doing visual effects with text like cogmind looks like vomit

>>100060955
Oh obviously it wasn't full screen, it was a little window, maybe a few hundred pixels on a side.
Also tkinter was doing all the polygon filling and that's all C (either in TCL/TK or in X I can't remember exactly how they do it.) Python just does a few matrix vector multiplications (just a few hundred or so per frame.)

>>100053613
GMI vibes.
>>100053736
We do like people who post their projects instead of bragging about them.
>>100058295
https://github.com/Anuken/Mindustry
>>100060897
I don't think, except for one poster, that this general do games, let alone embed scripting languages.
>>100044233
>you can emulate any console similar to NDS and older
Not only that, but the most optimized DS emulator (drastic) uses cpu rendering.

>>100060964
Huh, neat. Console release is one of my goals though. I doubt it'll gain much traction without at least some basic spritework.

>>100060981
Yeah text would be hard to sell on small screens. Unless it's very stylized like Door in the Woods.
https://www.youtube.com/watch?v=zZT-wdbZZbM

>>100058250
https://store.steampowered.com/app/1162750/Songs_of_Syx/
recent game made in java that I just bought. I haven't been spending any money on c++slop though.

>>100061008
That's pretty cool. I'm just throwing some random OGA art against it for now. Will probably commission a pixel artist somewhere in the future to make it my own.

>>100061489
do you check the programming language a game is made in before you buy it?

I already made a list of my own, but does anyone else have ideas or suggestions how to make a traditional roguelike more accessible to the mainstream gaming masses? Vs it being a niche genre with ASCII graphics.

First class controller support (especially if targeting consoles) and intuitive UI/UX is a given, while keeping the depth that makes the genre great.

Anything else?

>>100061949
good graphics and presentation
a captivating theme

File: crypt-of-the-necrodancer-(...).jpg (108 KB, 1280x720)

108 KB JPG

>>100061949
>I already made a list of my own, but does anyone else have ideas or suggestions how to make a traditional roguelike more accessible to the mainstream gaming masses? Vs it being a niche genre with ASCII graphics.
Do you know what "roguelites" are?

>>100061988
Yes. I deliberately don't want to make a roguelike. That market is saturated already.

I want a traditional roguelike with no cross-character power progression, outside of achievements, hero unlocks, bonus levels etc.

>>100062019
No roguelite* damn autocorrect

>>100061988
necrodancer is a roguelike
binding of isaac is a roguelite

>>100062019
And you think the market isn't saturated with traditional roguelikes?

>>100062061
Not the kind I want to make.

File: StW.png (878 KB, 1040x2276)

878 KB PNG

>>100062098
Anyway, check out shiren the wanderer, which plays like a consolized version of a traditional roguelike.

>>100062141
Thanks, I'll check it out!

>game used to have localisations to any random language even on retro consoles
HOW? FUCKING HOW? I refuse to believe that text rendering and localisation is still a hard problem decades later, how the fuck can you handle left to right, right to left, verticale text, cursive text, arabic, million of chinese moon runes, having correct kerning PER FUCKING FONT???
I'm about to use cairo to draw full rendered text boxes offline and just load that instead and draw it, fuck everything about text encoding and text rendering, humanity failed to improve on language.

>>100063297
It used to be as shrimple as loading a monospaced bitmap font and using a subset of the language you wanted to support. For example french and spanish support was free if you gave up on all accented characters, sure it wasn't correct but it still did the job.
You can't do that anymore these days.

>>100063588
>You can't do that anymore these days.
There's no reason you can't

File: spleen-font-accents.png (775 B, 417x73)

775 B PNG

>can't get pixel-perfect rendering right
>find out size and position are swapped
Odd that it seemed to sortof work before then. Probably bug introduced while fixing precision issues with UV before I switched from u16 unorm to plain floats.

>>100063588
Bitmap fonts do support accents, letters like é or à are separate codepoints anyway, so it shouldn't be hard to add them.

>>100063791
If you do that you're gonna get bad reviews like
翻译不佳 感情受伤 东方陷落 by third worlders.

>>100064172
Yeah sure on a modern system with unicode that's not a problem, I was more thinking about retro consoles and their weird random encoding.

Suzanne works but there's glitching on Sponza

>>100064238
just learn england you thurdie

>>100065200
Software rendering needs to be so perfect to never have artifacts lmao.
Does every other piece of code have 50 tiny bugs also but we just don't notice them?

File: fi8LE7VJ427te4RvVtBWFV-1200-80.jpg (72 KB, 1200x675)

72 KB JPG

If your game doesn't run on embedded linux, you should seriously optimize your game.

File: 1712276968247081.png (357 KB, 578x748)

357 KB PNG

>>100043403
Is C# a decent language to use for game servers if I want to add multiplayer to my game? I'd ask the web dev general, but web devs have different priorities compared to game servers. I was wondering if anyone here ever used C# to add multiplayer to their game, and if so, what they liked/disliked? A quick google search seems to suggest it's a decent choice, and I guess there's no perfect choice. Outside of some basic NodeJS + SocketIO stuff, I haven't done backend programming for games.

>>100067941
It's not the language that matters, as long as you can serialize structs and send them down the wire, preferably using UDP, it's just down to design of your networking architecture.

File: networking.jpg (221 KB, 1440x1396)

221 KB JPG

>>100068064
>It's not the language that matters
That's mostly what I figured. But if I'm learning a language specifically to do server-side shit, I'd like it to be at least good at that.

>>100068083
You're using different language for client and server? This sounds horribly inconvenient as you can't reuse code. What sort of game are you making? Networking demands of an FPS game and some turn-based thing are wildly different.

>>100068168
Godot actually allows for C#. Not making an FPS, but I would like real time multiplayer (eventually).
>What sort of game are you making
First one or two will be turn based (so a glorified board game). My plan was after adding quality of life features like a server browser and maybe a lobby chat, I'd move onto figuring out real time multiplayer. Basically translate this turn based 2D game into a real time 2D game.

>>100068267
The thing about networking especially real-time games is that the server and clients need to have the same representation and understanding of the game state, at least if you do it the proper exploit-resistant server-authoritative way. You obviously need to be able to share most of your code in this sort of situation. I've never used existing engines myself so I don't know what issues you'll face, but best of luck.

>>100067692
what version of OpenGL does it run? Only OpenGLES?

>>100067941
You can make a server in python if you want to, C# is more than adequate, you'll actually find its concurrency way simpler if you want to handle a lot of player connections. Just chose whatever language has the best tooling.

is accessing an array element by index faster than accessing it with a pointer?

>>100069902
an index is a pointer
a pointer is an index

>>100069902
this is almost at the same level as
>array of structs
or
>struct or arrays

>>100069966
is it really?

>>100070155
pointer to deez nuts

>>100070148
which one is better tho? I thought struct of arrays is better memorywise

>>100070187
makes sense, thanks

>>100069902
x86 memory operands can do addressing with 2 registers, one which can be "multiplied" by 1, 2, 4 or 8 with no penalty as it has dedicated AGU units for that.
ARM has something similar
RISC-V in theory too, with instruction fusing.
There's compiler optimizations too.

It doesn't matter.

>>100070148
The difference between AoS and SoA is significant when considering memory bandwidth and/or amount of cache line fetches.

>>100070155
yes
a pointer is an index in memory

>>100069902
you're asking a language specific question
what qualifies as an "array"?
if you're talking about C, then it's the same thing, a pointer is just a memory index

>>100070361
its only significant if you dont use data from the whole struct

>>100067941
It's fine but isn't C# managed? If you need something real-time that needs fast updates the GC may be a problem.

>>100056855
>Informative videos but I hate these clickbait titles and thumbnails so fucking much.
Yes, but I could give less of a care because these videos are short and SO worthwhile. Who cares about clickbait if that's the case?
i.e. the 12000 FPS one covers:
- How to optimize rendering by taking advantage of assumptions, in this case, voxels having discrete coordinates and symmetric geometry
- Chunking data into smaller encodings by taking advantage of assumptions and "chunking" that can be used to group data for like objects in proximity
- Batching faces into one plane instead of NxN planes
- Instancing one plane and using the symmetry of a voxel to draw six other faces
And so much more. It's probably the best video I've ever seen on rendering optimization and it's not even close.

>>100070689
If you cant figure this shit out yourself you're a crappy programmer

>>100070701
>If you cant figure this shit out yourself you're a crappy programmer
I believe that's a lie.
Every idea you've ever had is something you've received. Being obstinate that everyone must lift themselves up by their own bootstraps in order to prove their worth is therefore hypocritical and evil.
You should celebrate and glory the things you've been blessed with.

>>100056855
>trusting the guy that thinks that you can't fit a 32 wide chunk x index in a 5bits number.
https://youtu.be/40JzyaOYJeY?si=BxkmhPOOdDQVsUMF

>>100070898
>Every idea you've ever had is something you've received.
You've never had an original thought or arrived at something through deduction?
Most of the voxel optimizations in that video are stuff that you could arrive upon just by thinking about the problem for 5 minutes

>>100070916
Then explain how.

>>100071004
NTA and didn't watch video, but 2**5 = 32

>>100071108
Dude. 32 wide chunk has 33 possible values.

File: are-you-serious-or-trolling.png (5 KB, 686x37)

5 KB PNG

>>100071143
How many elements does this array contain?

>>100071193
30

>>100071193
32, which is enough to encode vertices for 31 voxels.

I'm no longer the stupidest of /gedg/. Today is a good day!

File: what.jpg (46 KB, 817x599)

46 KB JPG

>>100071255

>>100071348
Chunk with a width of 1 voxel has 2 vertices, left and right one.
That's 2 possible values, 1 bit is enough, so far so good.
Chunk width of 2 has possible 3 values, which needs 2 bits.
Chunk width of 32, 33 values.

>>100071434
A voxel is a value in a 3D grid, where the fuck are you getting 2 vertices from?

>>100069902
>doesn't know how memory works
>>100071434
>doesn't know what a voxel is

https://www.youtube.com/watch?v=4xs66m1Of4A
this is the end boss of voxel optimizations
imagine meshing every voxel face like a pleb

>>100071450
>>100071474
Holy fuck, people here are autistic.
The context is https://youtu.be/40JzyaOYJeY?t=409

>>100071510
I'm not watching the video
If he's conflating voxels with vertices he's a fucking moron

File: 1691502327596779.png (5 KB, 190x266)

5 KB PNG

>>100071434
>1 voxel has 2 vertices

>>100070931
>You've never had an original thought or arrived at something through deduction?
I don't believe in original thought, but as for deduction, your ability to uncover truths largely comes from information you've received and playing, which is one of the reasons I'm opposed to the ethos of 'figure everything out yourself'.
Some examples from that video:
- I started to think about symmetry back when Casey posted that video about drawing a circle and showed that you can reflect across a circle in two different ways, negation and swapping, so when I saw Vercidium use the symmetry of a VOXEL to instance one plan six different ways, I was able to understand more deeply and that's helped a lot in reasoning about transformations
- I knew that you can only see three sides of a cube in perspective projection from reading Kaiji, which helped me understand back-face culling voxels and only showing visible faces of the six sides of the chunk mesh(also the idea of inverting the voxel so you're inside the die and that same three plane principle holds)
>Most of the voxel optimizations in that video are stuff that you could arrive upon just by thinking about the problem for 5 minutes
I know you're being hyperbolic, but that seems unrealistic for someone that doesn't have a strong background in geometry or Graphics programming. Even then, he might reveal principles you haven't thought about, so there's still value in it.

>>100071925
>I know you're being hyperbolic
Actually I'm not, things like merging faces and sorting them by direction are the first optimizations I thought about when creating a voxel renderer
And I guess the rest of the video is basically compressing the data format which is also basic stuff
>I don't believe in original thought
What kind of retarded belief system is this

>>100071991
>What kind of retarded belief system is this
>"For who makes you different from anyone else? What do you have that you did not receive? And if you did receive it, why do you boast as though you did not?"
>1 Corinthians 4:7
God Bless you.

>>100072016
A evangelicuck I see. Waiting for the rapture while doing absolutely nothing?
Does God have original thought?

>>100072016
Very bizzare interpretation of that passage

>>100072098
Yeah.
I accidentally posted the "New International version" which is less accurate. The King James translation says it better:
>For who maketh thee to differ from another? and what hast thou that thou didst not receive?
The first half is a straightforward rejection of pride and holding yourself above others because of what you have. Material things, like money or luxuries, are the obvious source of pride, but I've seen more people become prideful because of appearance, status, or knowledge, which can all be taken away, so those are things you "have" not an intrinsic part of you.
>now if thou didst receive it, why dost thou glory, as if thou hadst not received it?
This is the part which contains the second meaning. At first, it's that you shouldn't boast about the things you have as if they weren't given to you, which you might think means "never be happy about your gifts because you didn't earn them", but I think it actually means "You should treat your blessing as blessings and be happy for them".

So I find the idea of 'original thought' to be backwards since it's not truthful to the reality where you observe knowledge and receive information. It gives you an ego which obscures the truth from you. It's still important to isolate yourself from The World since hyper social norming will also obscure the truth. What most people would call 'original' I would instead call 'authentic'.

>>100072572
>you observe knowledge and receive information
You also process information and that's what "original thought" is

wtf is going on

>>100072815
Someone in the thread who has outsourced all of his thinking.

>>100072882
and that's how you get 231 posts with 0 progress

discuss progress
I finished seamless travel between portals today and now I'm working on my first person arm

>>100073077
Working on GUI, specifically figuring out how to best handle clipping.
At first viewport seemed like a good option to handle it for me, but the number of viewports is too limited. Clipping rectangles myself isn't hard but I'm trying to figure out a nice abstraction which can be used together with font rendering, since I need to cut characters in pieces if on an edge and hence adjust UV.

>>100069902
a compiler will optimize it to the same thing in most cases. they are basically indistinct.

in other cases, it's just cpu architecture dependent.

>>100073077
Did some refactoring of the wall placement logic.

Do unity developers really use the unity physics engine to code things?
For example if I wanted to engineer cars in a game. Do they really use rigid bodies and stuff?

I was looking at this and it seems like a poor way to control all aspects of physics

Recommend some programming/compsci tutorials in the same format as Ray Tracing in One Weekend. Doesn't have to be computer graphics related, just anything programming or compsci.

>>100074177
>Recommend some tutorials
petition to ban this from the thread

I'm getting filtered by maths again

>>100073880
isn't unity's in-house physics engine still really new?
i assume most people are still using physx just like literally fucking everyone and everything
there's no real reason not to any more (though it would be nice to have an AMD GPU port) after it went open source, havok is out of the question, and bullet actually has a litany of problems, design issues, build system issues, compatibility issues

>>100075197
What's wrong with Bullet?

>>100075183
I know that feel bro

>>100075222
it's been a really long time since i looked at it and my memory isn't good at the best of times but, iirc
>it only works in like one very specific static build configuration (this isn't documented anywhere, and if they haven't fixed it even their readme for using it with vcpkg has it misconfigured) due to a fucked build system
>they might've fixed this but there was fairly extreme code duplication, i think to some degree it was because pieces of bullet2 were still interwoven into it but it seemed like every module had bits of duplicated code
>really shit docs
also they really ought to replace their opencl module with something less shit (AMD's opencl performance is completely fucked, no idea how, especially since HIP is much faster and shares the same backend and compiler) that can also access modern hardware features
vulkan compute shaders or intel's sycl would probably be a better idea
it would be nice to have an actual competitor with physx

and in general it doesn't seem like it's being worked on at all, the most recent commits to the library are spaced out by months and years

>>100075560
So really you have one complaint, that you have to statically build it, which isn't really a problem at all

can some kind anon explain why is it that for my model to stick to the camera view (like a gun in an FPS) I have to multiply it's model matrix by the camera's inverse view matrix

gunModelMatrix = glm::inverse(camera.GetViewMatrix()) * gunModelMatrix;

it works but I don't get it

>>100076300
you're attaching it to the camera

>>100044233
yeah, i wrote a 2.5D voxel raycaster (arbitrary columns of voxels in each grid cell) and it runs at over 100fps at 1080p, software rendered, with per-pixel lighting and fog, and 5DOF support (everything except pitch up and down supported)

just arrays of simple structures kept as lean as possible to maximize cache hit rate.
the per-pixel lighting and fog is even more linear, just a loop over all the pixels, perfectly prefetchable and no cache misses, it's almost free w/ SIMD.

Yes, on gpu it would likely be hundreds of fps*, but it's cool seeing what a cpu can do.

>>100075876
no, i'm not complaining about static linkage, i'm complaining that i had to spend hours figuring out what was going on with an obscure part of CMAKE back when i didn't even know it at the time before giving up and only finding out a potential solution after inadvertently stumbling across the information in a vcpkg github issue
and it wasn't just static linkage, there was some weird incompatibility with some tooling or something (this was years ago (then again, so was the last real commit))
this plus the horrible documentation is a pretty clear sign that it's either they don't know their own project or they're GNOME-esque cunts

>>100077169
you having trouble figuring out how to use it does not mean it has a 'litany of problems'
It's not as good as commerical solutions, but it's decent

>>100073077
Reading monogame source code rn, figuring out what's available to use. The available resources (official docs, API docs, video tuts by pajeets etc) are mostly superficial and low quality.

Luckily the code is straightforward to understand. I'm giving some basic structure to my game with eg a ScreenManager registered as a GameComponent (that nobody seems to use), a ContentLoader registered in the already-available service container (that nobody seems to use).

Also I'm ditching MonoGame.Extended and copy or reimplements parts I need from it, like the OrthographicCamera. That package is a buggy and unmaintained mess.

>t. roguelikedev

>>100077186
if your buildsystem is fucked, and you know this, and you tell no one about it, you are to blame
especially if it's cmake
especially if you have the audacity to publish an incorrectly configured package to a package manager
especially if your project's only documentation is a shit pdf that doesn't even contain an API reference
like even the bare minimum of doxygen docs would be better

it's a sign of overall project quality

also just going over the source i can't believe i missed this it's C++03
that's even worse than i thought
who knows what unspeakable horrors, horrid hacks, and terrible workarounds lurk in there
no wonder there was so much code duplication
i think even boost moved on from supporting pre-modern c++

>>100077297
>it's a sign of overall project quality
No it isn't, you're literally judging a book by its cover
None of the stuff you're talking about is related to the actual quality of the tool you idiot

File: boxes.png (16 KB, 806x606)

16 KB PNG

Figured out how Love2Ds model of pushing coordinate transformations on a stack to do transformations locally works. What screwed me up is rectifying Box2D's center of body coordinate with Love2D's top-left semantic. So the order is: Push, translate origin to the body, rotate the coordinate system according to the body, and translate the system up-left so the graphics draws correctly.
I'm happy about it.

>>100077313
pre-modern c++ sure as hell fucking is, hell even c++11 is pushing it with a library like bullet
there is literally nothing redeemable about that era of c++

no move semantics, no proper support for things like atomics and multithreading (and no, platform intrinsics aren't enough), the code's guarenteed to be clunky and difficult to work with, especially due to limitations with templates and constexpr, tons of situations, the compiler could optimize better, attempts at optimization which do the opposite thanks to language level guarentees for things like copy elision, no noexcept

>>100077410
Bullet's easy enough to work with you fucking roleplayer

>>100077432
sure, why not, that's not important anyways at this point, the library is going to perform worse, be at risk for more bugs, and might even do dangerous and retarded auto_ptr level shit to get around the lack of move semantics, or maybe even just to get around all the language level bugs that existed at the time
how the fuck did you manage to learn c++ without picking up a habit of avoiding pre-C++11 codebases on instinct
i'm pretty sure the core guidelines even says never dip below c++11

>>100077533
>how the fuck did you manage to learn c++ without picking up a habit of avoiding pre-C++11 codebases on instinct
It's called having a job
Many gamedev libraries are pre C++11, some of them are even in C!
Next you'll be telling us to use Rust

>>100077558
C would unironically be considerably safer than pre-modern C++ because it's simpler

>>100077573
pre-modern C++ is also simpler than modern C++
>safer
yeah we should use Rust right

>>100077595
yes but the language is a dangerous mess
C never got fucked up to that degree

>>100077603
pre-modern C++ is a dangerous mess
modern C++ is a dangerous mess

>retards talking about which memory "safe" languages people should use
go to back r*ddit to discus your mental illness

>>100077072
>it runs at over 100fps at 1080p
Cool, people also forget that cpu rendering doesn't suffer from gpu rendering limitation, aka having to render from the main thread exclusively unless maybe you use some modern API.
Even mobile has at least 4 threads, you can use a thread for rendering and another for game logic, and even use double buffer memory or some lockless technique to make it even faster.

>>100078087
And you'd still be 1% as fast as GPU rendering, wow

>>100078109
with 0,001% the power consumption and hassle
fps only matters as long as it doesn't dip bellow your screen refresh rate, even then, you can target 60fps, 30fps, or hell even 15fps because a consistent 15fps will look better than fluctuating 60fps. unless you're making a high octane action game, who the fuck would care if your 2D pixel game is running at 100fps and not a 10k?
what matters is that you get to support the widest range of machines by default when you ignore the shitshow that is gpu and all its driver issues and differences through all platforms.

>>100078579
You're consuming far more power doing CPU rendering than GPU rendering

>>100078579
>a consistent 15fps will look better than fluctuating 60fps
is this a troll?

>>100078604
see >>100043427
steam deck was able to last 7 hours instead of the typical 2 hours.

>>100078744
>Elden Ring vs pixelshit game
Retard

>>100067941
>Is C# a decent language to use for game servers if I want to add multiplayer to my game?
I'd say yes https://github.com/runuo/runuo/tree/master/Server

>>100069143
Apparently no opengl, just sdl 1.2

>>100069902
https://stackoverflow.com/a/2305783

File: bodyweight-exercises-chart.jpg (533 KB, 1320x1867)

533 KB JPG

You are also building your own body right?

>>100078579
GPUs are.much more efficient per watt at graphics rendering since it has fixed hardware specifically for these tasks. They also have order(s) of magnitude more compute power than CPUs.
CPUs are good at executing serial code, but get beaten by a mile by GPUs on (massively) parallel tasks.

>>100079296
I pretty much stopped working out 6 months ago after 7 years of lifting

>>100076314
that did not help but I already figured it out in my sleep

>>100079296
No, I would prefer a dying body so I do not have to stay in hell for long.

File: Screenshot from 2024-04-1(...).png (15 KB, 1521x800)

15 KB PNG

Do you guys care about non-game GUIs in this thread? I just started writing one as an experiment.

>>100079296
Yeah, been bulkmaxxing for the last 10 years or so, 5'10" 354lb right now.

File: IvyMen_yZPQhHRJqv.webm (3.12 MB, 1008x724)

3.12 MB WEBM

Added billboard sprites

>>100081209
you can post whatever you want, but as a protip: nobody cares about UI features/updates. I see people record a promo video for their game on steam and so many of them like to show off their inventory management system for some reason

how do I improve learnopengl animation code? sending bone transforms as uniforms every frame is slow as shit. am I supposed to send them as vertex attributes instead?

all engine deving is cope for not having good art and assets
if you had those you'd just use an off the shelf engine

>>100081709
(You)

>>100081709
Retard. Enginedev is a cope for having zero creativity or interesting ideas to make a game with.

>>100081709
>he's not blendermaxxing and enginedevving at the same time
you stoopid

>>100081709
fact is if you aren't good at everything nobody's going to play your game anyway. using an off-the-shelf engine isn't going to make your game's mechanics not feel simultaneously too stiff, too floaty, and too generic because you don't understand the relationship between your animation's timing and the acceleration of your character's x axis.

>>100078087
graphics API's can benefit from multi threading, you can write memory to memory buffers from separate threads, and you could also make the gpu thread execute commands assembled in other threads (frontend-backend architecture), but you don't need to do this because it's not very hard to hit 200fps+ on a 10 year old intel IGPU using a low level graphics API's (and depending on what you are drawing, probably also 200fps in a naive / inefficient way).
Also how would you multi-thread CPU rendering? On graphic's API's it's trivial (and part of the reason why vulkan/dx12/metal is more verbose than opengl is because the API is designed around doing some work in different threads, but note that opengl 4+ is basically vulkan in multithreadding features).
Technically the API for CPU rendering is single threaded as well, the only difference is that the canvas could be accessed by multiple threads, or maybe you could buffer the frames so that you are rendering multiple frames in a single thread?.
I don't care if you use the CPU or GPU, just use any engine if you actually value your time making video games (it's crazy to think there are people out there who would rather spend 10 years making one game, when you could have completed 10 games in theory if you used an engine, wait, I'm an engine dev...).

>>100081670
>sending bone transforms as uniforms every frame is slow as shit

No it's not, look elsewhere for optimization

>>100081857
>when you could have completed 10 games
I can shit out 10 2D platformers/cookie clickers/TD/... games in less than a year, but that doesn't mean they're good, let alone interesting to me.
While engines provide a nice boost initially I find I always end up fighting them when implementing more complex features, so while enginedev might be slower initially I believe it is worth it in the long run, depending on what you want/need.

>>100081890
there is a noticeable fps drop when I include another animated model, like a 100 fps at least
is this really the most efficient way to do this?

>>100082023
100 FPS out of how many? If you're already at 10000 FPS then 100 matters little.
My rendering implementation is a bit esoteric since I try to draw everything in one go with vkCmdDrawIndexedIndirectCount, but I have a large buffer of transforms which gets indexed by a per instance offset + per vertex offset.

>>100082094
but what if I want 10 animated models? a 100 fps drop is huge. maybe it's the standard for animated models but idk

>>100082218
For stuff like this you don't need to look at FPS but at frametime, since FPS = 1/frametime.

Hypothetically, let's say each frame takes 1ms always, and each model adds 1ms more.
- 0 models -> 1ms, 1000 FPS
- 1 model -> 2ms, 500 FPS, 500 FPS diff
- 2 models ->3ms, 333 FPS, 166 FPS diff
- 3 models -> 4ms, 250 FPS, 83 FPS diff
See how the FPS difference gets smaller? But frametime difference stays the same.
So determine how much time each model takes on average, then extrapolate from that. Or just try adding an extreme amount of models and see what happens.

>>100082262
ok this does make sense, it fits what I see

>>100077558
>Many gamedev libraries are pre C++11
like what

>>100082262
Huh, NTA but that's kind of insightful, never thought about it that way.

holy shit meson is amazing

who's meson

>>100082262
I wish more people understood this basic math, like the retarded video of the guy making his game go from 10000 FPS to 12000 FPS

>>100086135
a build system for c++ & co that everyone gushes over
pretty much everything still uses cmake though

I'm getting filtered by the stencil buffer

>>100087644
i didn't even bother

>>100087644
just greedy mesh to get 12000 fps

>>100084684
witcher 3 source got leaked! get in frens

>>100088873
who cares?

>>100088873
>761.78 GiB
yeah nah
if someone is willing to upload only the source files I'll bite, otherwise it's a hard pass

File: bump-limit-beyond-300.png (874 B, 61x18)

874 B PNG

Did bump limit get raised or am I misremembering things?

Turns out it's 310.

>>100089965
is it? let's see

>>100089965
>>100090166
what's with the weird cutoff anyway? why not 300

>>100089904
>874 B 61x18
based bandwidth preserver