when is godot removing this IUD bullshit? sorry I mean UID

>>108972609
no idea

>>108972609
Moving to my own engine was the best gamedev decision I ever made to be honest. No more fighting structures not made by me nor for my use case, if I don't need it it doesn't exist, and if I do need it I'll know how it works.

>>108973243
i have decided to do the same but i cant switch mid development, this single feature gave me depression because i cant really update and older versions are totally unmaintained and ignored. the engine source code is the most vile c++ codebase I have ever seen with my eyeballs, so if any critical issue arises im cooked
im never again trusting a random engine with 20+ years of legacy code despite its recent rewrite attempts. it's bloated as fuck and full of spaghetti. the lead dev goes into full cope mode when someone rightfully complains about perf issues with this shit engine, and its users act like they're from a cult just like rust. also he's a bit of a grifter

>>108972587
whats the affordable (poor) way of getting some nice tween animation into my engine?
live2d free requires getting a licence to distribute
dragonbones gave me the wrong vibe (very chinese!)

>>108973243
whats wrong with godot? As long as you use the graphics server directly it should be fine no?

>>108973440
What engine? And yeah once you're mid development it's a bit too late, assuming you really are mid development and not 10% of the way and being way too optimistic.
>>108973838
Haven't tried Godot, only Unity and Unreal. After starting my own engine I just don't feel like the vast majority of games made by genuine 1 man teams (or 2-3 men, just not the "indie" with 40 devs and AAA publisher backing) necessitate the amount of features that make a premade engine worth it.

Like what do you need that you can't program yourself?

>>108972609
that's not how you pronounce godot

>>108973867
>Like what do you need that you can't program yourself?
I guess the answer is why program something that already exist in a engine?
If your goal is making a game wasting time doing a engine kinda sucks I guess. Personally I just pissed off at editor driven dev it makes it hard to understand.

>>108973867
premade engines are nearly always going to be faster than making one yourself

>>108973243
Post code

>>108973879
How do you pronounce it?

>>108973890
>I guess the answer is why program something that already exist in a engine?
Because you want it done to spec for your specific use case and every single line of code that doesn't do any work for you is more mental load and dead weight slowing you down.

You're also not making an engine like some people seem to think of it as. The game and the engine are not separable, they're one and the same.
>>108973901
Faster in terms of development? If you can remain disciplined and work through it, sure, but it didn't work for me, all I was doing was working on top of a gigantic black box that forced me into creating things in a way that I didn't like.

>>108973941
pedot

>>108973941
Godot, not Godot, and definitely not Godot.

>>108973945
>all I was doing was working on top of a gigantic black box that forced me into creating things in a way that I didn't like.
That's the control freak fallacy yes

>>108973968
Maybe, but I also thought all of that before I started in my own engine. Maybe I'm in the minority, but I doubt I am the only one in the world.

Can't know until you've tried, and I'm my own proof of that because I sure as hell spent many years thinking my own engine would be a massive waste. That may still be true depending on your goals, but most solo devs do not have such lofty goals and if they did the engine would not be the only impossibility anyway.

>>108973998
People have been able to make pretty much any game with Unity so what makes your particular use case so special that you need your own engine?

File: echo.png (49.1 KB)

Step 1: make an echo server [DONE]
Step 2: make an MMORPG

Only 1 more step to go, I'm almost there!

>>108974134
Just because they could make it doesn't mean they didn't fight the engine. I don't want to put effort fighting the engine or dealing with their choices.

Focusing on your own specific use case will also always give better performance than a generic engine by a large margin, unless you fuck up majorly.

>>108974262
Fighting the engine is still less work than making your own, that's the point

how do i make cool and flashy in-game UI.............

>>108973879
kek

File: works on my machine.gif (546.2 KB)

>>108972587
Hello anons, this might be a question more for /pcbg/ but in your experience what's the minimum and good enough amount of VRAM for running UE5? I'm a uni student who typically uses overbuilt lab computers but I'm looking to build my own setup. I'm a level designer primarily so taking advantage of post processing effects would be ideal.

Basically I wanna get a good bearing before I start narrowing down the list. Thanks

>>108974271
It's really not, though.

>>108973901
wrong because using other engines is so depressing I would never finish the game while working on my own is enjoyable

>>108974648
In nearly all cases it's not, unless you have a simple game and you're an experienced programmer, or you're a large studio who has outgrown the workflow
If you have simple presentation requirements like a 2D game with no modern graphical effects and you're a good programmer is the only time I'd consider it better to make your own engine

>>108974633
On 8GB of RAM and 6 of VRAM it was a miserable experience in UE4 already. No idea beyond that since I upgraded my PC.

>>108974633
no idea about UE5 but I use UE4 with 12gb vram (3060) and its pretty comfy
however, my job with UE4 is primarily using Niagra, so I'm only building things like VFX and also toolings to send over to the art/designers team. They all use 16-24gb vram GPU.

Wow agdg is worse than ever lmao

>>108974738
Noted. I thought of using my 1060 and seeing if it's a passable experience but that's good to know
>>108975008
>however, my job with UE4 is primarily using Niagra, so I'm only building things like VFX and also toolings to send over to the art/designers team. They all use 16-24gb vram GPU.
Good to know thanks. FWIW I watched a video of an indie team talking about what they use and it seems like 12GB is pretty good but you can manage with 8 if you really need to. I'll keep all this stuff in mind thanks anons

>>108975086
I started making my game 2 weeks ago, figured I would check out the 4chan gamedev community, found that general, now I have super aids and cancer.
I remember going there like 4 years ago and it was pretty comfy.

File: coarse_lod_zoom_out_lakey_island_compressed.mp4 (3.2 MB)

Raymarched Voxels with Vulkan and C.

here is an aggressively compressed video with no LOD transitions to just give an idea of scale.

Yes, the voxels below are not high res, the land mass is large. So the view from above is a LOD.

But this video gives an idea of the scale from ground to above.

So the voxels look 'large' when close up in this video because I did not implement the LOD transitions. In gameplay, the voxels will be small.

>>108976107
https://www.youtube.com/watch?v=0inbi3nByJE

>>108973941
goD ohh

>>108976107
https://www.youtube.com/watch?v=T9pC2NX--nc

File: 1a.png (528.9 KB)

>>108977567
look at these fucking landscapes man, lost technology

File: 1780570569410.png (38.3 KB)

For a game that uses a C++ and OpenGL2 backend and Lua (jit where supported) for the game logic, which platforms do you think will be able to run this?
Do you think 3DS is capable of running a game like that?

>>108977619
Apparently the 3DS supported OpenGL ES 1.1. Note that ES is the "embedded" subset of OpenGL, not the same as the regular desktop version.
I personally consider OpenGL ES 2 to be absolute bare minimum for anything I'd write, but you do you.

>>108977592
nice, I always loved those Comanche voxels
https://github.com/s-macke/VoxelSpace

>>108977796
There are apparently some mesa drivers for 3DS that can port most of GL2 ES functionality

>>108977918
>>108977796
https://github.com/cpp3ds/gl3ds

>>108977844
how the fuck they have generated these color maps? and heightmaps?

>>108977592
>height maps
>the most common way of generating terrain to this day
>"lost technology"
>>108978544
they're literally just noise patterns with colors painted over them...

>>108978832
I was thinking about voxel landscapes being lost technology not heightmaps

I've been doing computer/software engineering for years, I am burned out and want to have fun with coding again like I used to have.

Is this a good hobby to pick up or will this just make me more miserable? I barely care if I make a game, if I make one I would probably make something isometric.

>>108979534
It's nice to follow your curiosity, always a good fun

>>108974341
you animate it

>>108979534
Game programming is honestly the most fun type of programming there is

>>108977592
I liked it best in Comanche and the M1 Tank game Novalogic made in its prime, called 'Spearhead'

File: companyofheroes-scrn4.jpg (31.3 KB)

I tried to avoid reality as long as I could, but it seems that I must develop my own engine if I ever want to make an online MP RTS. I hoped the recent UE5 RTS games meant someone had figured it out, but apparently UE5 just handles the graphics and audio, the rest still had to be custom made.

>>108979482
anon a heightmap is literally a data representation of a "voxel landscape" where each texel is interpreted as a voxel

>>108979935
there's the Spring engine

>>108979983
If the feature set isn't full of shit, this just might work. Thank you.

>>108979935
>>108979983
>>108980054
I know a lot of this isn't engine-bound, but every game made in the Spring engine has always looked more or less the same to me. I guess nobody bothered changing the shaders or fixing the floaty vehicle movement.

>>108980122
DORF is made in the Spring engine and looks completely different to the regular TA slop

What's the coolest feature you added to your gayme engine?

>>108980127
>DORF
That's OpenRA.

>>108979534
it's really fun. Computers and programming is fun, convoluted abstractions and other bullshit people people on top of it suck but with game engine stuff you can avoid most

>>108976107
Why do C trannies need to mention their project is in C? As long as it's a compiled language with no garbage collection, it makes no difference. Rust I can still understand because of the lack of support, but there is literally nothing special about writing your renderer in C, except for more coding.

t. C only coder

1 polygon jam

>>108980390
oh my bad
check out OpenRA then

>>108980361
Template events

Waow I'm a big boy foss-contributor now. Slouken merged my PR into SDL.

>>108980682
What do you get in return: sex, money, . . . job?

>>108980780
I already have all those things.

>>108980780
He gets my personal "Hooray". He can put it on his CV.

>>108980816
your joking

>>108980844
There are more job-and-wife-having guys on 4chan than you might think.

>>108980857
why am i not that guy

>>108980870
What guy are you

>>108980881
not job and not wife having guy

>>108979534
there is nothing special about gamedev. it won't make you any less miserable if you're already burnt out. what works for me is making small projects. domain doesn't really matter as long as you can get invested, and make some quick progress.

>>108973243
careful, youre going to upset all the incompetent retards

>>108980682
nice. I got some PRs merged into GLM a while back. nothing groundbreaking, but still.

>>108980447
>its okay when rust niggers do it

File: nadine costumes.jpg (1.5 MB)

>>108972587
>find a new 2D game engine
>It's just a basic front end to SDL2
What's the point of all these "engines" whe you can just use SDL directly?

>>108980127
>DORF is made in the Spring engine
Wait, it is? The same engine that has ZeroK and BAR? How tf did they do that? Those games look so primitive compared to half of the features in DORF

>>108981217
you might prefer to use a scripting language to make coding easier, and to avoid overhead of compilation and of setting up a new project in C/C++.

>>108980437
>>108979669
>>108979588
>>108981135
Thanks guys, I'll start on some primitive stuff and try to work my way up, see how far I get.

>>108979534
It can be fun to have something visual: All progress is visible and tangible and not just puts(status ? "It worked" : "It failed");

File: 1739917546991366.png (982.3 KB)

>>108981152
Very cool. A tiny piece of your code is in some of my software. And now a tiny piece of mine will be in a lot of other people's software.

>>108980526
>a single fullscreen triangle and entire game rendered through a shader
problem, officer?

>>108980682
That's fucking awesome. Congrats and thank you.

File: noclip.webm (3.8 MB)

also uhhh you can do this now

>>108972587
there is something genuinely deeply beautiful about installshield installer dialogs.

especially the full-screen blue gradient ones.

it was like embarking into an adventure, a new world was unfurling infront of you. even today i can feel the magic when i close my eyes and put myself infront of that crt screen again.

my game has just been printing money every day since it released

>>108982001
>>108982139
I really like the visuals in your game. What's your tech stack? How long have you been working on this?

File: 1763250165494859.jpg (49.3 KB)

Gaben said that my game constitutes a hate crime

>>108983016
Started staring at disassembly and making the track editor a couple years ago, actually learned IDA Pro and did the decompilation last summer. There's not much of a tech stack just it's mostly just SDL3 and a handful of other libraries for small specific things.

>>108982503
printf("money\n"); doesnt work in the real world, unfortunately :(

>>108983324
At so it's an actual decompile and rebuild of fatal racing? Cool

>>108983215
there's always itch

File: whalejs_nsis.png (45.3 KB)

>>108982224
it's NSIS for me, but yeah, there's something nostalgic about those old school installers

>>108980447
i use lisp btw
t. lisper

how long has your engine been in dev

>>108985331
Spent 3 weeks on the initial DX11 version, then 2 on the DX12 version

>>108985334
wow that's quick what features does it have

>>108983600
Yep, I've started to deviate from the original code fairly significantly now but the physics/input and software rendering is still basically as near as frame and pixel perfect as you can be to the original. That may change in the future though as I'll have to make even more changes to improve the netcode for higher than LAN latency and I think most people would probably rather have things like unlimited framerate than frame perfect control input in a racing game. We ain't speedrunning mario here after all.

>>108985331
Since August but I don't separate my engine from the game so it's not really a good indicator.

>>108973537
blender

where is the jai leaker anon, we need the newest version pretty please

>>108973537
Aren't tweening functions very simple math formulas? If you want full-blown live editing you really don't need to implement much on top of interpolation. Just a basic UI tool for editing keyframes + playback pretty much.
What exactly are you trying to animate?

>>108985658
they are not very simple no but you can just copy them, its unecessary to make them yourself.

>>108985380
Not him but I'm >>108985613 and my first Vulkan commit was 10 months ago and I would say that
>Forward+/Clustered Forward
>GPU mesh and light culling (and drawcalls)
>debug visualizations
>glb and ktx2 import
>some amount of work going into the non-renderer stuff
As well as a bunch of autism about scaleable/proper systems (e.g. proper mesh and material instances, CPU to GPU buffer uploads, everything working with buffer device addresses, GPU scene vs CPU world decouping etc), so it wasn't just "get this thing on screen as fast as possible".

...was in about 6 months ago. Throw an extra month on it maybe, but that's about it. I had never written a single line of Vulkan, OpenGL, DX11 or DX12 before this. I also worked full time when doing this (granted it's European full time so 40h a week is actually 40h a week).

My point being these things, especially Vulkan or DX12, really don't take anywhere near as long as I've seen people claim. You'd think diving straight into Vulkan would be a herculean task but if my lazy ass managed people really should stop being afraid of it (or at least stop trying to scare others off like crabs in a bucket).

>>108985380
In the DX11 version there's a common struct for game objects, with parameters like shader, mesh and texture(s) to use, as well as scale, RGB-factor and tiling factor. Frustum culling is done on the CPU. Assets are streamed in based on visibility and uploaded in an asynchronous manner (so streaming without stalls).
The DX12 version is less feature complete (no streaming or audio) but has a more efficient design for rendering: Entirely GPU-driven, so the CPU does almost nothing and the GPU all by itself does frustum culling and draw call generation using bindless resources.
Now I have a full-time job and don't work on it anymore.

>>108985677
wow nice one

>>108985331
over 11 years. it barely does anything.

make games not programs

>>108985745
how do they compare in performance?

>>108986997
The DX11 approach has O(n+m) time complexity on the CPU for issuing draw calls(n=number of objects to cull, m=number of draw calls to issue), which also has to be happen on a single thread.
The DX12 GPU-driven approach has O(1) time complexity on the CPU for issuing a render operation, and a low constant factor at that. Telling the GPU to render 10 million objects is, for the CPU, just as fast as telling it to render just 10 objects.

The end result of is that you're much more likely to be able to achieve 100% GPU utilization.
If you had that to begin with (for example on a laptop with a weak embedded GPU but decent CPU) you'll see little to no benefit.
On a machine with a high-end GPU you'll be able to achieve full utilization, assuming there's no other bottlenecks, while also freeing up a CPU core to do other work.

You probably wanted a simpler number, but the matter of the fact is that a vertex shader or pixel shader is no faster simply because it's running under DX12 versus DX11.

>>108987207
I mean how much faster is it to do an entirely GPU driven pipeline compared to a CPU one

>>108987220
Not him but I found it to be much more logical. Same with using buffer device addresses everywhere I could. It has seemed to me like when it comes to the renderer, more advanced often means just more straightforward. My own sanity was the reason for me using bindless, BDAs, etc rather than performance.
>lights rendered on the GPU? cull on the GPU
>drawcalls consumed by the GPU? create them on the GPU
>need data in the shaders? just pass an address to the start of the buffer and index into it
Which is why I HATED the 2-3 days I tried OpenGL before dropping it for Vulkan. So much bullshit that makes no sense, it's all just data why do I have to follow all these stupid rules and weird systems?

Sadly textures/images remain something you can't just dump into a buffer, however that makes some sense since the way they are stored on the GPU is considerably more complex than just "here's a linear set of bytes".

>>108987263
Right forgot the answer itself:
If we're talking developer time, at least in my case it wasn't faster or slower. You have a bunch of data you need to do some operations on, you have to write that code regardless, your choice if to do it in your language CPU side or HLSL/GLSL GPU side. The difference in dev time, assuming you plan for it from the start and you can use stuff like BDAs, I think is negligible.

>>108987282
I'm asking how much faster the game runs

>>108987326
Oh in that case it purely depends on your game, but I doubt it will ever be slower. A GPU can handle many many more operations in parallel, and it doesn't inherently have to wait on the CPU to pass shit over to it, it can in theory just run at full blast regardless of how the CPU is operating.

Basically it's effectively always faster to some degree (on PC platforms anyway), if it will be so much faster that you notice it depends on your use case. Also you're of course moving operations from the CPU to the GPU, so you are shifting the load, and the difference depends on the hardware difference (but surely nobody is running a Ryzen 9 with a GTX1660 or something right?).

>>108987359
there's always data that needs to be sent unless literally the entire game runs on the GPU
I still want to see how figures on how much faster it is

>>108987427
>there's always data that needs to be sent unless literally the entire game runs on the GPU
Obviously, but if your mesh culling happens on the CPU those are al calculations not happening on the GPU.
>I still want to see how figures on how much faster it is
That would require someone take a game that's doing this stuff on the CPU and create a version where it's all on the GPU while intentionally changing as little as possible outside of that. I can't even seem to find any benchmarks for this. The only way you'll get a real answer is if you go ahead and make one yourself I think. Basic renderer with meshes and lights, spawn a few tens of thousands or more and flip between the CPU branch and GPU branch, see what performance differences you can find.

>>108987491
>That would require someone take a game that's doing this stuff on the CPU and create a version where it's all on the GPU while intentionally changing as little as possible outside of that.
Didn't you say you made a DX11 and DX12 renderer for your engine? That's why I was asking

>>108987550
Not him anon, said so in >>108987263.

Regardless, going from DX11 to 12 would almost certainly change too many things outside of the bindless stuff and poison the results.

>>108987565
>going from DX11 to 12 would almost certainly change too many things outside of the bindless stuff and poison the results.
No it wouldn't

File: oneBitUpdated.jpg (589.3 KB)

messing around with my one bit outlines shader. just wondering, does anyone have a GPU suggestion? my 1050 ti died on me, and now i'm using my super old 750 ti.
is 5050 good? or should I save some extra $$ for a 5060 with 16gb or something? im the anon sometimes posting the painting shaders btw, im not an imposter into the thread.

>>108988039
>is 5050 good?
widely regarded as "should not have been released" by gaymers

>>108985331
Since December 2025

Here is my post;>>108976107

>>108988352
ouch. i dont game a ton so its still prob a good upgrade from the broken 1050 ti, and especially the 750ti
i guess ill deal with the 750ti for a while and save some extra $$ towards a 5060 or maybe 5070

>>108986868
based

File: postProcessFakeLighting.jpg (516.6 KB)

improve the speed of my fake lighting post process. its that (uses APV+Spherical Harmonics data and faked realtime lighting), then experimenting with a laplacian of gaussian on the lines. its kind of slow... for now.
rip to you, 1050 ti, some day i will improve to the point that my shaders will run even on a broken gpu.

>>108985331
First, define what an engine is. What I have is more a collection of stolen ideas.

>>108990074
a a p i

File: clouds2.webm (3.3 MB)

Did you guys know the clouds in Whiplash are physical objects that exist in space with real coordinates and you can fly to them? Well they are.

more people should try rpgmaker
especially kids
its easy as fuck and doesnt require any prior knowledge

>>108990433
Is there a way to make it perform well?

All Rpgmaker games I've played were resource hogs with high input latency that are just frustrating to play.

That's why I refuse to play porn "games made in rpgmaker, even when I'm super horny and they feature my fetishes.

>>108991024
>That's why I refuse to play porn "games made in rpgmaker, even when I'm super horny and they feature my fetishes.
Fuck man, I know that feeling. They use it like the worst visual novel engine ever, or even worse, actually try and use the awful FF7 or whatever clone mechanics into their game. Like come on, that shit is not congruent to a porn game.
>were resource hogs with high input latency
Literally electron-tier shit now: nw.js specifically.

It would be nice if there was an engine that was just a basic, possibly grid-based overworld, not dissimilar to the older pokemon games, a dialogue system, and a scripting system that isn't pure ass. They coule make the same kinds of games, just less awfully.

>>108991024
i think porn game makers are just retarded
they make the absolute basic visual novel crap, not understanding anything about the engine, probably filling the map with parallel events that check their own state 60 times per second or some crap like that.
although the engine by itself is not for action games, it is like a single processor web browser, chromium web instance.
But it is easy af to use tho.

>>108992089
you have to always think about optimization when using it, but that is good, its common discipline.

File: file.png (82.7 KB)

thank you ai, very cool

>>108992089
for example im yet to see anyone really trying to use branching dialogues, like really branching, not yes no sarcasm. And you could make it, very easily in that engine.

>>108992089
>they make the absolute basic visual novel crap
there are some high-effort porn games out there. but yeah, most are just weeb shovelware.

>>108973243
yes, you can add the features you need, it's also less bloat and you can make it compatible with vintage hardware if you really wanted to

File: dlss5.jpg (58.2 KB)

>>108989587
Is this some sort of DLSS 5 on meme?

File: 1780175039545588.png (1.5 MB)

ive been on a break for 2 months, i said I was gonna start again in June but I've just been playing runescape instead. This is on-topic because this is a general and generals are glorified chatrooms

File: l3.mp4 (3.6 MB)

lighting

>>108992484
trippy

>>108992089
porn games being slow and clunky gives them soul

>>108985629
Has anything been added that's useful? I'm still on the leaked version from early 2025.

Huh I thought that the visual C++ runtime was only for C++, but it turns out it also applies to C, I thought C would just used the UCRT (as that's how mingw works), but it still need vcruntime for functions like memset.
That info probably would have helped explain why anon thought the UCRT is not available on all windows PC's, I probably said that C doesn't require the hybrid UCRT trick that C++ uses (/MT + dynamically link to ucrt), since he is correct that C does not work on a freshly installed PC probably (but it would if they used the hybrid UCRT or they used mingw).
Because if you only used /MT, it would add like 5/10mb of debug info and that's mostly the ucrt (10mb without /OPT:REF, ~3-5mb with I think).
You shouldn't statically link the vcruntime if you shared C++ objects between dll's, but for my usecase, I don't link to C++ libraries (some are C++ libraries but they link as a C library/noexcept or they are header only). I like DLL's because it's faster to link and I like how it advertises the libraries that are used, and it makes the stacktraces show the module, it would be very annoying if I didn't use vcpkg however, so now all I need to do is add the linker flag to my custom triplet for hybrid ucrt.
but every DLL has it's own pdb debug symbol... that gets messy quick. it's fine without debug info because the export function (the first function into the DLL) is visible in a stacktrace. Or I could just rely on a 3rd party error reporting system or throw the pdb files into a folder and load it either with /PDBALTPATH or or adding it to a search path (to copy pdb's I use TARGET_RUNTIME_DLLS and regex replace dll with pdb, it seems to use the path in vcpkg/installed/, this will not work for non vcpkg libraries which is expected)

>>108993776
What is it about build systems that make people write paragraphs and paragraphs of useless information?

>>108993798
Because that's what build systems are. If you need a build system, something went wrong with the language, which is why C++ has build systems, meta build systems, and meta meta build systems: because it's the language that has most gone wrong of all.

>>108992484
nice

>>108993837
*C and C++

>>108977619
550G for 11 mushrooms is outrageous

File: post process lights and shadows.jpg (127.6 KB)

>>108992280
no its fake lighting on post process. pic rel is the super early attempt on it.

>>108992308
I've been thinking I want to make a game that's just basically a copy of high-level old school runescape PvM, but I'll probably never get around to it.

>>108993837
Assembly has build systems.

File: 85081f6f19253bad.png (18.8 KB)

bruh

>>108993837
>If you need a build system, something went wrong
Holy Dunning Kruger.

>>108993837
Every language in existence that involves any sort of compilation has some sort of build system. You probably just don't have a realistic definition of build system and probably don't have that much experience with other languages. And whatever previous experience you had, you were probably accustomed to being handheld the entire process for writing even a hello world. Hiding away complexity from the user does not make it go away. It's still there. It still exists. You will eventually notice it unless you only ever wrote hello world programs.
Even every web application has a build system nowadays no matter what lang you use because assets should be preprocessed to save bandwidth, lower the amount of HTTP requests, etc.
Go read about stuff and stop being a brainlet. Quit whining.

File: 1751930865415334.png (1.5 MB)

>>108995198
>>108995418
Since you want to be pedantic, let me be pedantic too:
>if your language needs a build system worth naming and is considered a whole system in its own right, it's a shit language
Better? Here, enjoy your 450+ page textbook on your build system kek
>web
Yeah webshitters doing something you're doing is a red flag telling you you need to reconsider your current workflow.

it's pronounced "smake"

another nodev filtered by cmake
he was ngmi

File: agdg-wiz-gob.jpg (383.7 KB)

guys, im making minecraft in godot, can i post it here ?

>>108996039
That's a great way for people to steal your idea.

>>108995947
kuh-mak-e

>>108995482
You don’t need a book to learn cmake you can just have AI create your cmake files for you which is what I do.

>>108996039
Yes please

>>108996075
Does it matter? Oh no. Yet another minecraft. The more the merrier.

its been a fat minute since ive toyed with games. is it unreasonable to use sqlite for gamesaves? what about occasional autosaves? state management? obviously you want to keep game state in memmory for fast access, but wouldnt sqlite:memory be sufficient?

>>108997098
It's unreasonable to use a database for anything in a game that isn't an MMO or live service game

>>108997098
Luanti uses sqlite to store chunk data so it depends on what you want to store.

>>108997098
sqlite isn't really tied to memory, it's more like it's tied to the disk, aka ACID-like which is basically for transactions of data, so that if you crashed in the middle of writing to it, it would still be in a valid state / rolledback, not useful for a game because a game is like a very big singular transaction, not a small transaction that sqlite is good for. This would be useful for server stuff like a player login system and asset DB.
In theory if you want fast save states you could use flatbuffers and store everything in a struct (fast but fragile to forward compatibility), tables are required if you want strings/dynamic arrays, but it's very difficult, ugly (every variable is a setter/getter, you could use the slower API where you copy the data into a plain class but no reflection), and it depends on code gen. It supports a little bit of json but it's very annoying to get working. You can do a bit of reflection on it, so if you wanted a save editor / view data in UI, that could work. And if you made updates, there is a method to making your saves forward compatible (if you had a automated test suite, you could add tests to make sure your old saves work on every update).
But it's a lot more simple to just serialize the normal way everyone does, which is with some save/load function in your objects, since if you actually used flatbuffers for everything, you won't have functions in your data (OOP), and you can't modify the array while reading, but if everything is a struct you could in theory use functional programming and use pools / freelists, and with that saving the game is just writing raw data (but with some of the slots being empty, since you are using freelists).
I think there was an open source 2d space game that just stored everything in a single chunk of memory, and multiplayer was done by just computing the delta and uploading it. It's not the best solution for scaling players depending on the game (lag compensation is a harder issue).

The feature system in my Minecraft clone is coming along nicely. Right now just dealing with simple anchors but I’m starting to visualize doing advanced stuff like cave splines. The main issue in my current system is chunks checking for overlapping features require their bounding boxes upfront before a feature is fully generated. Right now I have the bounding box calculation simply hardcoded, that is the maximum possible size of a tree. The alternative to generating a feature’s blocks upfront to get the proper bounding box which I’m trying to avoid lest chunks try to generate trees that they don’t touch.

The process goes as follows
>Chunk checks for overlapping features
>Higher level feature provider that covers a 64x64 block area generates up to 6 trees at random x/z locations. The tree feature’s bounding box is compared against the chunk’s own bounding box. This is where the issue is.
>Overlapped features (trees) are returned to the chunk
>For each overlapping Feature generate it (during feature generation any blocks that are outside the chunk’s bounds are discarded)
Tldr: A feature’s bounding box is required before it’s actually created.

The solution as I see it is to pregenerate all the provider’s features blocks upfront when the anchors are calculated and cache the block changes along with the bounding boxes. This might be what Minecraft does I’ll have to check.

The problem is that I don’t actually cache the 64x64 provider chunks so I can’t cache their data either. At the moment for each 16x16x16 block chunk I’m actually creating a 64x64 provider chunk upfront. Which you’d think would be expensive but the performance is negligible since all it’s doing is finding up to 6 random anchor points on the x/z axis. Now that I think about it I’m pretty sure Minecraft does this for every feature that’s not a structure. Structure rooms might be the only thing that are cached through whatever system.

>>108993837
Lol. Post your clojure fizz buzz work in progress bro.

>>108997744
im not reading all that
post a picture

>>108997947
There’s no interesting picture to be shown. Once I get more features in I’ll post a pic just for u.

File: 1750315432810490.jpg (62.5 KB)

>yeah so my game idea is *game from 15-30 years ago* but worse in almost every regard
I sure hope you guys don't do this.

>>108998576
that was literally me, but I finally came to my senses and pulled the plug. still way too late.

>>108996039
>im making minecraft
why?

>>108998576
for real

I'm making my game in C++, and I'm currently implementing customizable controls. I created a gigantic enum for basically every single possible action the player could take (walk left, walk right, jump, shoot, etc.), and I created a map between those actions and keyboard keys. Now, if I need to check if the player is pressing a button, I check the action itself rather than the keyboard key. To assign a key to an action, I just modify the value in the map. This works, but is it "correct"?

>>108999391
it's good enough, you technically can use an array when your enum is 0-indexed and sequential but that makes it a pain in the ass to make quick changes during design.

>>108999568
Not him, but how is it a pain if you just have

enum Actions {
   JUMP,
   WALK,
   SHOOT,
   ACTION_COUNT
};

and access the array via eg. keybinds[JUMP]?
sorry I'm a nocoder

is it smart to have a return type other than void for some things?

>>108999668
What do you mean, some functions the sole purpose is to get some value. Recursion needs a return value. Returning success/failure is also often important. Granted, you could return things via reference/pointer in the args but usually you only need one thing.

>>108999668
>>>/g/sqt/

>>108999668
Of course.

>>108999888
>>108999999
>>109000000

>>108998576
that would hit harder if anyone here was actually making games

File: photo_2026-01-14_03-54-58.jpg (41.1 KB)

>>109000039
>>108998576
I'm re-writing a game from 30 years ago to make it better in every regard.

Why don't we all team up and work as an indie studio?
I'll be the CEO

>>109000039
I am making a grognard game, so it doesn't really apply to me because even grognard games today miss such basic quality of life features that it doesn't take much to do better.

audio-visual stimulus generator

>>108999659
I guess I was assuming you wanted to initialize it statically with a braced initializer, so your initializer elements are positional rather than associative. If you don't mind using an init method I guess you could access elements individually one at a time if you don't think that's annoying or ugly then good for you.

>>109000523
But you can just copy the enum definition and use multiple cursors to add the brackets and equal signs. Changing individual keybinds is as easy as using to each enum value at runtime.

enum Actions {
  JUMP,
  SHOOT,
  CROUCH,
  ACTION_COUNT
};

int keybinds[ACTION_COUNT] = {
  [JUMP] = 1,
  [SHOOT] = 2,
  [CROUCH] = 3,
};

(excuse the globals)

>>109000143
I'll design the logo and be in charge of casual racism around the office.

>>109000668
Array designated initializers are not standard C++ and I get that if you use C you don't even have a map class to worry about so you feel attached to not needing an associative container or trying to see how it could be more ergonomic to use one when it's not an absolute necessity.

>>109000855
yeah sorry. I'm a Cnile. But isn't using pure C (mostly) faster even with the C++ compiler?

>>109000950
>But isn't using pure C (mostly) faster
no

>>109000996
Because of vtable overhead or what?

>>109001015
there's no meaningful performance difference between C and C++, C++ can do almost everything C can after all

>>109001054
There's a difference in compilation time. But surely there are things that are faster but more cumbersome to implement in C, and vice versa in C++?

>>109001081
There's a ton of stuff C++ can't do that C can of course
There aren't really any things C can do that C++ can't

> cant make progress in game dev since lazy
> device to learn music instead
> open lmms
> spend the whole fucking day getting surge xt to work
> it sounds like shit
why is making music so hard. I might just fire up my ps4 and just use dreams to make music at least there it was straightforward.

>>109001149
everything worth doing is hard fag. don't expect to make any music that sounds good at all until you do it consistently for years

>>109001102
tf are you even saying dude

File: whl_phone.webm (2.8 MB)

I finally made it work on Android.

>>109001202
sorry I mixed that up
There's a ton of stuff C++ can do that C can't

>>109001217
learn to screen record saar

>>109001232
That would look just the same as recording of the desktop version. Kinda boring.

File: 1755500779840730.jpg (28.8 KB)

having a read

>>108998576
I do do it, and it's going to be a better game and you will seethe. Suck it.

>>109000143
I'll be the COO with actual talent who the board of directors kicks out by successfully falsely accusing me of sexual harassment

File: small.png (2.2 KB)

vibecoded 4kb shitpost (9kb -> 4kb with crinkler)
build + source included
have fun

https://www.mediafire.com/folder/m1ncq4a9e9m3r/small+test_tinydx11+++crinkler

File: f41316f9d8a82a61.png (341.3 KB)

>new crazy taxi uses AI
oh god

>>109000668
What's the pount of enums if you resort to manually convert them into integers, stick to data structures that use them as actual enums.

>>109002098
I literally have no idea what that statement even means

>>109002098
Nobody actually cares except deranged furfags who don't even buy video games.

is the "vulkan is manageable after you set everything up" a psyop? when you decide to add additional shit beyond the triangle is it just as verbose or actually manageable?

>>109002098
As someone who works in the industry, here's a scoop: EVERYONE uses AI nowadays, every single fucking one.
>>109002630
That's shit you'd have to set up anyway so it's not Vulkan specific. Any serious project will have you build the same or similar structures on top of OpenGL.

>>109002630
>additional shit beyond the triangle
you mean a loop to draw the triangle more than once?

>>109002630
Any library is easy to use once you create an abstraction around it

>>109002660
>EVERYONE uses AI nowadays, every single fucking one.
i dont

>>109002794
I meant at companies in the industry. Don't know a single one that doesn't. Unless you mean you also work in one, and yours doesn't, which is absolutely possible, but all the ones I've had a look into do.

t. working somewhere in Activision-Blizzard

>>109002660
>>109002805
What kind of AI tools do they use?

>>109002830
Depends, most programmers I work with use Claude or DeepSeek, not like anyone vibecodes though.

Art department uses AI for various things, though nothing that comes out of the AI ends up in the final product without quite a bit of modification. Think they're experimenting with 3D mesh generation but as far as I know it's not at the point where the meshes even work as base meshes except for super basic shit like a barrel.

>>109002630
it depends.
it's like saying C or C++ is easy after you set everything up as well. It does not mean that setting it up isn't a huge waste of time, you will still waste a sizable percent of your time trying to figure out how to get libraries and stuff working.
but why are you even asking the engine general? everyone who wants to make a game engine is a masochist that does everything in the hardest way possible. Of course we are going to psyop you into thinking this is easy, when a game engine would let you make playable games in a week, maybe a single day if you used AI.
the people here have spent years on 4chan.

>>108985331
like 2023...I gave up a few months ago. Made something cool, but enginedev is a waste of time desu. And AI takes the fun out of handmade stuff. The real brutalpill is that distributing your game, i.e. the most important thing, is a nightmare if you roll your own stuff and want it working everywhere. Just use an off the shelf engine.

>>109002630
You’re conflating Vulkan with software architecture. Vulkan on its own is easy given you’re using modern extensions. Building an RHI and further rendering systems is more challenging. However IMO AI makes it pretty easy. You can put together stuff like a resource manager and frame graph system, forward/deferred rendering quickly. You can literally have gpt put together a step by step guide or a multi week plan. Also AI is very very good at solving validation errors.

>>109004096
it's easier than ever to use your own stuff with AI no?
"Determine what needs to change and create a plan to get this to build on windows"
"Do it"

>>109004379
I don't think AI can do that

>>109004440
Try it with claude code

Haven't posted here in ages. Decided to try game dev again since AI can generate art for me.
Tried Codex for sprites and it keeps hitting me with bullshit like
>We’re so sorry, but the image we created may violate our content policies. If you think we got it wrong, please retry or edit your prompt.

Is there a way around this? I know there are probably better LLMs out there for this but I have Pro subscription trough my work.

>AAA has co-opted the indie ps1 aesthetic
it's so over

>>109004862
You have to explicitly instruct it that all characters are over the age of 18 which also causes it to give all female characters massive tits so nobody gets the wrong idea, for you know safety or whatever

>>109004862
local image generation works great. it's not like with LLMs where you need data center-class hardware to run it. when I first started, I still had a GTX 1060.

>>109005135
what game?