Thread 105668822

Build a Roguelike-edition

/gedg/ Wiki: https://igwiki.lyci.de/wiki//gedg/_-_Game_and_Engine_Dev_General
IRC: irc.rizon.net #/g/gedg
Progress Day: https://rentry.org/gedg-jams
/gedg/ Compendium: https://rentry.org/gedg
/agdg/: >>>/vg/agdg
Graphics Debugger: https://renderdoc.org/

Requesting Help
-Problem Description: Clearly explain your issue, providing context and relevant background information.
-Relevant Code or Content: If applicable, include relevant code, configuration, or content related to your question. Use code tags.

Previous: >>105629295

still refactoring

>>105668849
You do have tests in place to support your refactoring, right anon?

>>105668866
tests?

>>105668866
>tests
found the webdev

>>105668951
Guilty lmao. Webdev by day, gamedev by night.

Idk anons it feels comfy to know that the shit I built won't regress later on with new additions or refactoring.

Only doing it for engine code, however.

>>105668972
You can know if things regressed by playing the game.

>>105669181
Just saving myself debugging time in advance

>>105669207
You're just wasting time instead of making a game.

Is there some kind of foolproof formula for knowing how many threads you should create
Should I do one for each logical core or what

>>105669220
Tell that to the other people arguing about vulkan vs opengl, c vs cpp etc.

Speaking of which, where's your game, anon? Or are you still stuck on the engine? Because I for sure am not.

>>105669230
I'm busy wasting time on engine tech and not making games.

>>105669246
Redpill me on your engine. What does it do?

>>105669255
It wastes my time.

>>105669272
Seems like we're all wasting time in this thread. Welp, back to coding my map generator.

>>105668951
whats wrong with tests? i use them to make sure my math and constraint solvers dont get messed up when i completely overhaul the algorithms. its faster than having to track down whatever is going on during debugging than to have tests compare the new implementations to good known states

>>105669454
How can you mess up something that doesn't exist?

>>105669454
Other anons haven't taken the testpill yet. I guess they're not used to building production-grade software.

>>105669498
Thank you for posting this sloppa franchise's worst girl, I really needed an idea for
>Unlimited test works
which are even more worthless than UBW.

>>105669454
testing for games is usually just starting the game and seeing if it works

>>105669498
Look at this >>105668972
>48 tests
Anon, you have 48 tests for your whole engine. 48 tests wouldn't be enough to cover even the inventory slots system of my game. It's not that I'm against tests, it's that being a solo developer means I either build features or test it on the go. I'd rather have players as testers anyway, and the obsession with correctness is IMO a mistake. Someone said that diamond is pretty but adding to it is hard, while a mudball is messy but adding to it is easy.
Tests are nice to have but ROI is way too low and opportunity cost is far too high for a solo project.

>>105669633
I just started this project 3 days ago and I'm working towards a very simple MVP tho.

>>105669649
If it works for you great, I’m not trying to convince anyone that their approach is wrong, but don’t be surprised when people don’t write tests for video games. Even if you prove that an algorithm or procedure is correct that doesn’t even mean it’ll be doing what you want in game, only that it’s correct from a mathematical perspective. This means that for games real testing beats manual testing most of the time.
I’m not fully against tests though, but that would be mostly for low level library code or implementation of rules like given some character with such stats simulate a number of attacks and see if the average makes sense.

>>105669661
>real testing beats manual testing most of the time
That doesn't make any sense, sorry I just woke up. I meant to say, playtesting beats automated tests most of the time.

>>105669661
Agreed, that's why I only cover engine code. There's a lot more code going on that's not supported by tests.

>>105669470
i dont do TDD if thats what you are talking about. i test things that exist and would be quite a lot of code to go through. so tests if my parsers are correct i will do, but testing if i can convert an int to a string i wont do

>>105669498
they like living on the edge, i guess

>>105669631
maybe in your world, but not in the real world

>>105669832
unit tests for games arent that common in the real world either

>>105669861
idk man, naughtydog even tests their shader code

>>105669881
>stuff looks good
>yup, it works

>>105669938
if thats what you want to believe, then i hope you sleep soundly tonight

>>105670003
You probably also use cryptographically secure prng for randomness since otherwise it might not feel fair to the player?

>>105670025
because what prng i use is related to testing code

>>105670038
prng's are one of the most tested pieces of software. Especially ones used in cryptography.
Because your game is just too good for simple state = _mm_crc32_u32(state, state << 18);

>>105670043
that is not the question you asked me. you asked if i use a cryptographically secure prng, not if i test my prng
i use different prngs for different purposes, including crc32

stop grasping at straws

>>105670079
crc32 isn't a prng though, it's a hash function. You should've tested that before misusing it. Watch out, 4 billion period is not enough for bing bing wahoo randomness...

>>105670089
yes, by itself it isnt. also still grasping at straws

>>105670110
I hope you at least test what matters, like save files, even though good games don't need those anyway.

>>105670111
what matters? no i test if 1 + 1 = 2, 5 million times, just in case a cosmic ray flips a bit

>>105670122
cosmic rays aren't real.

try not to reply to the psycho troll

>>105670111
the time it would take you to reach a high enough coverage for any game or game engine is most likely an order of magnitude higher than the time it would take you to debug errors as they happen
I do more or less agree that some systems are inherently hard to debug (garbage collectors, serializers, things like that) and proving they're correct may be a good idea, but only because the time to write the tests would be smaller than the time to debug them. For most other systems that's not quite true.
Also you brought up ND but they can afford the extra expense. I doubt most solo or small team developers who made successful games ever bothered with tests, starting in the 90s with Rollercoaster Tycoon and up until now with Schedule 1.

>>105670188
Tests in general are worthless and are busywork for job security.

>>105670201
I would normally agree but it depends. SQLite for example must ensure correctness under all sorts of edge cases like power failing during a write, voltage spikes, etc. Then there's the fact that there can be legal liabilities like in finance and medicine.
But I also think that most people approach tests in a very naive way. They take some optimal operational conditions, run the test, and they're happy to see that it passes. Shit dude, you have to simulate power going down while saving your game and seeing if the file gets corrupted, that's where automated testing shines. You have to simulate the player hitting quicksave while another save is going on. People thing they're doing TDD but from my experience seeing how the average programmers write tests, they really aren't.

>>105670216
>Shit dude, you have to simulate power going down while saving your game and seeing if the file gets corrupted, that's where automated testing shines. You have to simulate the player hitting quicksave while another save is going on.
The first case is a literal freak edge case that no game accounts for and the second isn't something you need to test for at all

>>105670216
>SQLite
Databases itself are worthless busywork.
>ensure correctness under all sorts of edge cases like power failing during a write, voltage spikes, etc
Does it have a test for me using a jackhammer to drill through the MRI machine it is writing the results?

>>105670222
Shit like this is pure cargo cultism, correct way is to have hardware redundancy, something software cannot ever account for. This even applies for games. If someone cried to me about corrupted savefile, I'd ask them why don't they just replace their dead disk and resilver their zfs cluster.

>>105670224
Yes, it’s equivalent to power failing. It won’t corrupt the database.

>>105670222
Then you’re most likely writing trivial tests

>>105670233
>It won't corrupt the database,
Oh boy I sure am glad that the database will be fully recoverable after I jackhammer the disk too...
This is beyond fucking retarded, anon.

>>105670233
Are you writing tests to see if you can't start one process while another is running? Really?

>>105670239
there's no people bringing jackhammers to MRI machines, meanwhile there's power outages with voltage drops while the generators kick in
SQLite ensures DB integrity under 99.9999999% of expected edge cases

>>105670244
>Game testing to see if it is already running.
This is one of most homosexual behaviors I ever observed.

>>105670253
$50 PSU could keep the power long enough before you start the generator, without expecting software to fix your non-problems, stupid boomer.

>>105670140
you arent real either

>>105670253
hows this relevant to games

>>105670269
Retard user might unplug their computer while game is running, then blame you of course.

>>105670273
Literally no game guards against this

>>105670279
Many games duplicate savefiles just so at least the old save survives.

>>105670279
pokemon solved this in the 90s

Share some progress of what you're working on, anons. Screenshots of your game or of the code.

>>105670433
I've decided to try and learn 3D modelling to make my own assets, but I have nothing screenshot-worthy for that.

>>105670478
How's it working out for you? What kind of assets are you looking to make?

>>105669633
What about tests to ensure the correctness of tests?

Is anybody here working on a game that isn't just pooping around for fun and actually realistically can and will make a dollar at some point?

>>105670433
i am working on threading. currently trying to find a good way of limiting how fast it runs, so atm im reading the docs since sleep is trash and theres no nanosleep like on linux

>>105670505
i hope i will make a dollar at some point from my game

>>105670483
>How's it working out for you?
Barely started, still just banging around shit in blender and trying to get a feel for it.
I'm probably jumping in way too quickly and going straight into trying to make a human.
My eventual goal is something in the same ballpark as Pokemon on the 3DS. Not exactly the same style, but low-poly and vaguely animeish.

>>105670505
No, nobody here is working on Fortnite 2 or World of Warcraft 5 or CS:UNIVERSE.

>>105670505
In theory someone might buy my stupid bullshit but I'm going to have to add a lot to make it sellable

>>105670505
I've released a few games

>>105670639
but did you make money from it?

>>105670859
yeah

>>105670887
how many sales? not to diminish it, just out of curiosity

>>105670908
I only got paid by my publisher so I don't actually know, it was a long time ago
Enough to live off for a few years at least

>>105670924
i hope ill make it like you did

has anyone here made anything yet?
last time i checked it was like gedg#20 or so

>>105670924
Goals. Make game, get money, take sabbatical, rinse repeat.

>finally implement a high resolution waitable timer
>500µs to 550µs overhead on top of the duration to wait
i hate this so much guys, i thought i had it but its not quite there yet... id hate to have to go dig up undocumented kernel functions

>>105671559
>500µs
/gedg/ finds out about syscalls...
Just ignore it, because alternative is benchmarking syscalls at startup and subtracting 767µs before sleeping and then using a busy spinning loop to finish waiting.

>>105671593
yea i know its a syscall but still, i didnt think it would be that expensive. ill give it another shot with an undocumented function, if that still costs a lot, i guess im fucked and have to just spin in user mode or take an L and accept a ~100µs jitter if i just subtract the 500µs from the wait

>>105671655
>~100µs
Any spike in cpu usage could easily make it way more than that.

>>105671667
very true, im just a bit autistic about it for no good reason

>>105671701
It also depends on kernel preemptibility.
In the end it doesn't matter, because all modern games are triple buffered, your monitor is always several frames behind.

>>105671727
this isnt for the render thread but the simulation one, so if the timestep is too small things get fucked up. i could of course just spin, but thats not ideal either

>>105668849
this except as a form of procrastination

>>105671771
So just simulate as many states as you need and replay them on time, fixed timestep simulation doesn't depend on time at all.

>>105671839
if the simulation takes less time than a timestep, you need to do something with the remaining time, you either spin, yield, or something else. im trying to find the solution with the smallest overhead that isnt spinning

>>105671871
No you don't, you can keep going and it's irrelevant how fast simulation is, if your simulation doesn't have user input, it could even be running ahead for thousands of steps ahead of render thread and mostly sleeping permanently because GPU won't catch up if that's really an issue you have.

>>105671905
i think we are talking past each other
what i mean is if you simulate at some fixed interval, then if the time it takes to do the simulation is less than that, you need to wait for whatever is left of the timestep
none of this has anything to do with how fast the gpu is

>>105671998
The whole point of fixed timestep is that you don't have to wait.

>>105672124
if you do 3 seconds of work in 1 second and then right after do another 3 seconds of work in 1 second, then you are 4 seconds into the future

>>105672219
When you're speaking in seconds, you can sleep for seconds and don't need perfect accuracy. Literally just double buffer the simulation.

>>105672235
i think we are definitely not on the same plane. i dont know why you are talking about buffering things all of the sudden. my problem is. things are running too fast and i want to avoid that while also not being forced to spin

>>105672377
Because presimulating ahead lets you sleep less accurately.

>>105672388
ok

>>105670518
Godspeed!

>>105670639
Nice! What kinds of games? Were they solo projects or did you have some sort of a team?

>>105672525
platformers and top-down shooter. solo projects

>>105672551
How does one get into that position? Surely you must have had prior industry experience or a few fairly polished prototypes for a publisher to take you seriously, no?

>>105669255
>Redpill me on your engine.
Engine go Brrrrrrr!

>Achievement Unlocked
>said the n-word

>>105673227
>1/1,000,000,000

1 month, no more seam between lod

>>105673695
ue5 could've fixed that with one toggle

>>105673695
That looks oddly like Unity and Sebastian Lague has a tutorial on this... Anon...

>>105668866
I hate automated testing and unit testing. I like it on principle but there's a time when it can be too much and hinders productitvity.

>>105673695
Same except on a subdivided icosahedron

>>105674852
your problem is your thinking in polygons instead of monitor pixels

>>105674935
Your problem is thinking in real things than imaginary ones. Voxels won.
Software rendering won.

>>105674954
performance should scale linearly with the number of pixels, allowing for unlimited detail in the simulation

>>105674967
If that were the case, we'd all still be using 1366x768

>>105675059
240p@60hz is the ideal for gaming

>>105675070
Based CRT user.

how are you supposed to enginedev if you don't have 40 different test rigs spanning 15 years of hardware generations, vendors, and operating systems?

>>105675211
You wrote like someone who hasn't got a clue about anything.

>>105670433
A lot of code refactoring for the input and rendering system. Then trying to get this shape adding to work so that they look connected by face. Not sure how I'm gonna do it yet or if it'll even work.

>>105674935
>>105674954
>>105674967
Funny you mention these things, last winter I wrote this cpu splatter prototype which uses the techniques from euclideon's "infinite detail" and Ken Silverman's PND3D renderers. Pretty much combines the pros of raycasting and rasterizing, though heavily relies on sequential decision making thus can't ever be ran on a GPU.
I never got to vectorizing the polygon rasterizer nor does it multithread yet. I got distracted by other things. But it already performs around the ballpark that Ken's PND3D did in single thread mode.

>>105675266
>real enginedevs don't test
found the vibratorcoder

>>105675325
What is there to test? If some irrelevant CPU is broken?

>>105675325
>>105675211
You just need to pass the "werks on my machine" test, and then berate the users if it doesn't work on theirs and that it's their fault.
It's good enough for many AAA devs, so why not you?

>>105675417
Because I'm a SS dev.

>>105675303
Nice frame drops, loser.

>>105675545
Our daddy taught us not to be ashamed of our frame drops, especially since they're such good size and all.

>>105675582
Hope that daddy hits the grave soon.

>>105675545
Lets see you draw a scene with 63.9 million visible voxels using just a single CPU core.

>>105675610
Work on your toy engine and jerktracing behemoth that no one uses. Just don't sit on the computer for too long, Timmy.

There are exactly 60 frames in a second.

>>105675695
*leaps over a second in your path*

>>105675667
What is your problem today? It's like you don't anyone to post in this thread for some reason.
Every time I've posted anything here I get mostly seething in return. What's wrong with this general?

>>105675722
Leapseconds don't exist. Checks for >59 do tho.

I finally got normal maps working in Vulkan. Friendly reminder to use unorm and not srgb. Next up is shadow maps and then a background cubemap, and that will cover the basics for a complete scene.

>>105675920
Afterwards it’ll be implementing
>debug views
>anti aliasing
>compositing(maybe)
Then continuing with the gltf format
>multiple cameras with selection
>basic non deform animation
Somewhere along the way I’ll fit in OIT and SSAO

>>105675964
I want to have a solid renderer before taking the plunge into physics and timestepping

>>105675545
>>105675667
Strong ngmi crab energy from these posts

>>105675695
True, but irrelevant as the human eye only sees at 24 frames per second.

>>105676016
Stay in denial of your failures. I don't care about you personally just had to comment on these lost frames old man.

>>105676031
I'm not even that guy you're trying to "dunk on", I just got here and voiced my observation about you

>>105676049
Sure even so doesn't change a thing. You are also good at doing mundane things, just like that anon.

>>105676056
>straight up a compliment: you're good at something
Thanks!

>>105676073
Also good at being disingenuous too. Just getting as much of a ngmi vibe from you.

json["mult"].destruct({
{"r", v.multiplier.r},
{"g", v.multiplier.g},
{"b", v.multiplier.b},
{"a", v.multiplier.a},
{"i", v.multiplier.i}
});

Finally got my shit together and made a json api that's not a nightmare to use.

>>105674852
looks diabolical

>>105673077
Yeah, I released a few free games before I did commericial games

The entire concept of vertices is archaic and should be illegal.

>>105675740
It's literally the same guy who ruined /dpt/ trying to ruin this thread aswell

>>105675740
/g/ threads related to programming are infested with this hyper-aggressive tranny schizo, trying to prove he's the next Terry Davis or some garbage.
It's so fucking tiring. Man I wish this place had some actual form of moderation.

>>105677738
He tends to stay out of this one because he knows absolutely nothing about video games so when he tries to argue about things he gets humilated

Audio for moving fluids. No audio yet for collisions.
>>>/wsg/5906466
Anyone have any ideas for transparency sorting? Doesn't have to be perfect. Right now fluids are always rendered last, so basically anything would be better than that.

>>105677914
Either sort them by their distance to the camera or use dithering
Maybe there's some other techniques but those are the most common

>>105675303
this looks awesome btw, ignore the schizo

>>105677922
Dithering is fine for now I guess. I wasn't using it before because I don't like how it looks, but now I need the proper sorting. Maybe I can come up with something else eventually.

>just want to program a video game
>"btw if you don't possess arcane knowledge of these magic symbols you just happened to hear about on a random dev blog your game won't run on laptops"

>>105678058
Commodore 64 devs really didn't know how good they had it.

>>105675752
"time" is based on how fast we move around the sun, if it speeds up or slows down, someone will add bullshit like leap years and seconds to "fix the calendar" because it's all made up.

>>105678101
damn, I sure am glad the entire universe orbits the sun so time exists everywhere

>>105678104
I sure am glad that I just use rdtsc from my CPU instead of making my game rely on random "real time" nonsense that might jump 5 seconds backwards if ntp suddenly decides to fix up the "drift" because apparently time is so real that we don't have technology to keep track of it.

>>105678113
yeah sometimes I just zone out and next thing I know 3 years have passed shit's weird

>>105678118
You should look up what coma is before you attempt to be sarcastic about things beyond your comprehension.

>>105678058
>DXGI_GPU_PREFERENCE_HIGH_PERFORMANCE
>VK_PHYSICAL_DEVICE_TYPE_DISCRETE_GPU
its that easy in the modern era

>Modern OpenGL Guide.pdf
// hello world triangle
float vertices[] = {
0.0f, 0.5f, // Vertex 1 (X, Y)
0.5f, -0.5f, // Vertex 2 (X, Y)
-0.5f, -0.5f // Vertex 3 (X, Y)
};

>glVertexAttribPointer(posAttrib, 2, GL_FLOAT, GL_FALSE, 0, 0);
>the stride,
>or how many bytes are between each position attribute in the array.
>The value 0 means that there is no data in between.
>This is currently the case as the position of each vertex is immediately followed by the position of the next vertex.
Son of a bitch.
stride = 0 works because it's a shortcut, not because there are 0 bytes between 2 vertices (or between this vertice's 2dposition field and the next vertice's 2dposition field)
stride is size of a full vertice (struct), here it's 8 bytes.

>>105678058
>>105678175
rare time when Windows is actually superior to Linux
>egl? but discrete gpu can't be setup in justwerks manner, use glx
>but glx is just a shitty extension that for some reason has imaginary problems happen in real world all the time
>but don't worry, wayland, on which nvidia doesn't even work, fixes egl, I promise!

>>105678192
You should try reading opengl guide written by someone who actually sold his game, oh wait, that never happened, stick to reading khronos manual.

How long until GPUs just go full blown fixed pipeline raytracing machines and 3d programming disappears entirely?

>>105678220
This will never happen, because "raytracing" is just compute shaders that for once aren't slower than a dead dog.

>>105678220
2 more weeks

>>105678220
lol sorry you get ai frame generation instead

>>105678209
nah, I've been in tutorial hell for too long and appart from this nitpick this one seems ok.

>>105678220
raytracing is always going to be slower than rasterization so not unless GPUs increase in strength by an order of magnitude

>>105678379
that's what they said about raytracing in general and then you had rtx

>>105678403
And? Normal rasterization is still faster.

>>105678403
And it's still true, they're using RTX for things that you can't do with rasterization

does anyone here engine dev with linux? how?

I don't even want my own game engine I just want to make a basic 3d simulation but basically everything online is for windows and visual studio.

>>105678505
You just install SDL or glfw and use either opengl or vulkan, it's not that complicated.

>>105678505
>basically everything online is for windows and visual studio
Use vscode and also what this guy said >>105678538
You can use emacs or (neo)vim instead of vscode if you don't like microsoft, but it will take longer to get started.

>>105678632
>but it will take longer to get started.
no it won't

>play old game
>shadows are sharp and well defined
>play new games
>shadows are a blurry mess

>>105678809
>>play old game
>shadows don't exist because nobody gives a shit

>>105678809
Depending on how old we're talking, they probably only did shadows for a few objects like the player character using some brute force CPU algorithm.

Why does it write like that?

>Furthermore, the recommendations in
[NVIDIA08] are from 2008 (first generation geometry shaders), so things should have improved.

>>105678538
>>105678632
>>105678769
thank you

what's the best linux distro from engine devving?

>>105679037
1.0

>>105679037
Any

>>105679048
even tumbleweed?

>>105679037
Linux accounts for less than 1% of the gaming market but will account for more than 99% of your debugging time.

>>105679037
Gentoo. If you're an actual programmer, you don't have time to deal with cucksoomer slop distros, you need a distro that supports debugging and profiling on a level of whole OS including the kernel.

I know this might not be the right place to post, but I wanted to share something. I’ve been wanting to make a little RPG game and after finding RPG Maker is a little piece of shit, I decided to start learning C# with the C# Player’s Guide, and now I’ve been going through the official Unity tutorials (like Create with Code, etc.)

And I’m actually having fun! Once I finish the tutorials I plan to make a simple little game just for myself. you can create a character, choose a type of enemy to fight from a map, and then go into a basic turn-based combat screen to battle. just something I can build and enjoy on my own lmao.

Wish me luck /g/iggers.

>>105679062
Did I fucking stutter?

As long as your ide and compiler runs on it, your distro (or Windows or macOS) does not matter.

>>105679062
>>105679037
It honestly makes very little difference, unless you're someone who particularly likes to use much newer versions of libraries.

>>105679089
Godspeed anon. Be sure to share some progress from time to time. We have too little of that itt

>>105679237
>As long as your ide and compiler runs on it
yeah it doesn't at least not without fixing a bunch of shit. why doesn't anything just work?

>>105679260
Just run Ubuntu LTS. Unless you're doing some very esoteric low level microprocessor shit. Better yet, use a MacBook with macOS if you want shit to just work.

>>105679265
>just dev on metal if you want vulkan to work

hol up if all the transforms are done in the vertex shader then how does collision detection work without duplicating all that work on the cpu?

>>105679280
There's not a single game that uses triangle perfect collisions.

>>105679280
On the CPU you're transforming some bounding boxes of select objects that matter for the simulation, on the GPU you're transforming the tens of thousands of vertices that make up the model

>>105679280
You have collision meshes on the CPU, collision meshes usually aren't the same as the one you're rendering

>>105679280
>he's using model meshes for collision detection
u fukin wot m8

>>105668849
why do you refactoring if you're a solo dev?

>>105679446
thats an odd question

refactoring is for team projects and maintenance

>>105679449
refactoring is for team projects and maintenance

>>105679453
its got nothing to do with being on a team

>>105679458
I don't think you know what refactoring is.

>>105679460
>>105679465
then for what?

>>105679469
improving code, making it clearer, faster, easier to expand

>>105679478
Just write it good the first time. Smh ngmi

>>105679484
whats the point in making retarded posts like this?

>>105679478
if you should go back and make your code base clearer, you're failed in the first place
>>105679484
this

>>105679494
Shut up, tranny schizo. Nobody asked for your input.

>>105679280
you do collision detection in the vertex shader too duh

>>105679494
Nobody writes good code the first time

>>105679497
that's funny coming from the tranny schizo

>>105679497
good morning sir
>>105679499
i do tho

>>105679485
To first think about the code you're going to write before writing it. Unless you're a nodev vibecoder, of course.

>>105679507
please stop ruining the thread

>>105679513
Sorry

Time to give my map generator some more love and make my levels look less boring.

>>105679248
I'm definitely going to do this. I'm currently on Chapter 4 of the Create with Code tutorial and after that i plan to follow a 2D tutorial to understand the differences between 3D and 2D things in unity (like rigidbody 2D/3D, etc.). Then I will jump into making my own little project.

And yeah I'm honestly really excited. It's the first time in maybe two decades that I've found a hobby that truly interests me something I can sink countless hours into while learning on my own. It feels like a mix of self therapy and finally discovering the right thing.

>>105679534
based rider

>>105679540
Hell yeah anon, you got this. I'm rooting for you and your project.

>>105679544
JetBrains is so comfy. I wish Junie support would release soon so I can vibecode even more efficiently.

>>105675740
Don't mind them. They are making slop in Godot while you are making something they never could.

>>105675740
>>105679561
Slop in Godot sells, your game will be ran by nobody.

>>105679668
>gaydot
gay people don't actually play video games

>>105679677
Gays aren't people and they're among the most likely of humanoids to play your indie game.

>>105679534
Not perfect, but better.

>>105679697
This is same shit with more rooms, you can do so much better than that.

>>105679711
It will look better once I add enemies, props, decals, doors etc and implement autotiling. Cba to do that for the MVP.

Next up is refactoring my old camera system and then work on the actor/turn system, before adding some basic enemies and fog/FoV.

>>105679730
No it won't, it will forever look shit. Do better.

>>105679738
Supporting prefabs (including genning a map with only connecting prefabs) is already on my to do list.

What, you want me to work on this right away?

>>105679750
Yes, if world is shit, things in it don't even matter. First of all, "you're stuck in brick walls" shit has been done for a billion times. At least use sandstone like you're in some pyramid, or organic jungle that cannot be crossed with tools you're stuck with, or some underwater cave, or even just a normal cave. Dungeons are dated and boring slop. Do better.

>>105679761
Please stop posting

>>105679778
Stop coping. The fact you posted an "update" with 0 improvements shows that you're ngmi. Do better.

>>105679793
I didn't post shit, get the fuck out of the thread you psychopath

>>105679796
Stop crying faggot. All I told you to do is to do better. At least read a book about dungeons instead of generating slop.

>>105679793
NTA. Where's your progress, anon?

>>105679800
Go back to making registry dumpers

>>105679804
>progress
I already did everything I wanted to.
>>105679808
Please take your meds.

>>105679815
Take yours and maybe you'll stop being the worst poster on this board

>>105679817
If you took yours, you'd have noticed that registrydumber stopped posting as much once I've proven that he doesn't even know as much as compiler behavior in regards to tzcnt.
Now stop deflecting, and make a dungeon generator worth posting on this thread, loser.

How about you both behave. Go outside and touch some grass. We're talking gamedev in here.

>>105679829
Why are you talking about yourself in third person

>>105679836
So let's talk about who will be impressed by square boxes overlapping with no story behind why they're here.
>inb4 autists get stuck on how story isn't important once again
Using L-systems, you could build rules for dungeons built by specific ancient factions and have some variety in your soulless crawler, but no, you're too stuck up in your ass to take criticism like every single faggot.

>>105679853
I'm not the one you've been arguing with lol NTA. I've just been posting good vibes and sharing some progress (no matter how insignificant it seems to you).

Be better.

>t. rldev

>>105679865
Don't reply to him, he exists solely to drain the energy from others, he doesn't make anything

>>105679865
https://sci-hub.se/10.1007/978-3-030-03314-9_32

>>105679880
I accept your concession.

>>105679882
Cool, thanks!

>>105679668
I'm not making a game. One evening the universe just revealed to me, what the techniques behind PND3D could be, and I had to try it out. There's only like 3 examples of this octree splatting tech on the internet currently.
Optimizing software renderers is such a fun challenge too. I can't pass up a good excuse to write SIMD intrinsics.

>songs of syx
how do I make my ui this good bros...

>>105680793
pic?

>>105680818
there is basically one shader used for everything. it's a geometry shader that takes start/end of the rectangle and uv and turns it into a quad. I guess it submits rectangles to a buffer, which I'm not sure it would actually draw it back->front like that. the whole game is like 3 draw calls. meanwhile I'm doing 3 draw calls to draw a single button and one per font character.

>>105680932
what do you even use to make normal maps like this

>>105680979
- Photoshop/Gimp/Krita all have filters for creating those
- You can just draw them by hand for pixel art
- You can generate them from 3D models

>>105679534
Based Rider bro.
Currently trying to improve mouse support on my shitty GUI implementation.

>>105679853
>Using L-systems, you could build rules for dungeons
what do L-systems have to do with castle architecture?

>>105681528
Everything.

>>105681528
dont reply to the troll

>>105681559
Okay, I won't reply to you anymore.

>>105678883
Because you want it to answer that way.

>>105681676
No I don't, I called it out on cargo cultism behavior and it doubled down.
GLAD is a piece of shit.

>>105681697
how mentally ill do you have to be to have conversations with AI?

>>105681762
Meds.

>>105680932
...anon, this UI looks horrendous

>>105681796
show me yours. oh right you just have imgui.

>>105679697
How do you have a Windows style taskbar on Ubuntu?

>>105681834
This is our current UI atm.

>>105681918
neither of them look good desu

>>105681925
can we see yours?

all three of them look bad

>>105681918
are those just 9 slices? looks really good

>>105681918
Use real textures instead of the noise filter, add some emboss / shadows, and change the repulsive shade of brown and this will look good

>>105681999
Yeah a lot of them are I believe.

>>105682026
>Use real textures instead of the noise filter
But anon, if a noise texture isn't a texture then what is?
>add some emboss / shadows, and change the repulsive shade of brown and this will look good
All of these were made, textured, and rendered in Blender so there is shadows and depth from 3D objects. So I'm not sure how else to satisfy your headcanon.

>>105681918
sovl

>>105682046
Noise is a texture, it's just a terrible one
Making excuses won't improve your art

>>105682054
What excuse exactly?

>>105682097
Sorry I meant give the borders and buttons and windows shadows and bevels

>>105682151
I don't quite understand, they have all that they are 3D scenes.

>>105682281
Your 2D GUI needs 2D shadows

>>105668822 (OP)
>Build a Roguelike-edition
I'm almost at the point where I can't keep working on the engine and really need to start making a game

>>105682316
It has shadows, and lighting. Sorry you don't like them.

>>105682386
Shadows on the 2D GUI you crybaby

>>105682333
Start making your game is when you learn which parts of your engine you need to refactor

>>105682411
I don't remember crying. Also something like a shadow behind the UI window is literally a preference/opinion. You're basically asking for a border around a border which I think is goofy and the games I'm influenced by didn't have.

>>105682458
Shadows / embosses / glows on GUIs are ubiquitous, especially skeuomorphic ones, they just look better and make things more readable

>>105682471
>and make things more readable
Yeah, which is why anything that's a font has a shadow... The UI has borders and expand/shrink on usage or have faded out/highlights showing activity/usage. Having a border and having a shadow around the border makes no sense. Also regardless a snip with hdr captured isn't showing anything properly or in action so none of this matters.

>>105682543
You can make all the excuses you want but having high contrast borders is key to having a readable GUI and if you look at any GUI that isnt flat they will have them

>>105682563
There is no excuses, you're the only one worried about this lol. And honestly if you have trouble using pic related it sounds like a you problem and I'm completely okay with that.

>>105682623
You've made about 5 excuses so far, I'm just trying to help, if you don't want it to look better that's on you

>>105682645
What excuses? I'm not held to some higher power lmao. I can make and do whatever I want and currently this is what I've been happy with.

>>105682702
being able to take constructive criticism would make you a better artist

>>105682711
You're not doing constructive criticism, you're saying "do this style instead". It's like telling someone making a wood sculpture they should be using clay.

>>105682766
It's not a style, literally every single non-flat GUI has high contrast edging, including whatever your references are
It makes it more visually readable

>>105679534
Rider is so fucking comfy

>>105682623
NTA. I think your UI looks cool, anon!

>>105682787
whats rider

>>105682790
Thanks

>>105682787
Rider is absolute kino.

What Unity shenanigans are you making?

>>105682795
rider deez nuts lmao

https://www.jetbrains.com/rider/

>>105682798
>>What Unity shenanigans are you making?
currently just doing unity tutorials (Create with Code). but aiming to start with my own little project after this.

>>105682892
Ah you're that CoC guy. Nice. Which Macbook you're using?

And already got some ideas for your RPG you want to make?

>>105682948
s/CoC/CwC/

>>105682787
Rider feels like old school software to me, I don't know how to describe it but it's like the distinction between programs and apps. It just hits different

brehs, which modern game engine would you recommend for a game with a lot of "visual" physics? For example explosions pushing things around, particles that react with the environment, (but not physics as a core mechanic). It should also be light on the graphics, intended to run on older systems.

>>105683221
anything really, unity, unreal

>>105683221
C programming language.

>>105681918
>>105682623
as always, this looks sick as hell anon

>>105681918
Would look better without those high contrast LGBTQIA+ bars.

>>105683221
Unity. Alternatively Godot.

>>105683265
No.

>>105683253
should I look into older versions of unreal, or UE5 could be made to run fast?

>>105683366
Well then it won't be light in any way, goodluck retard.

>>105683404
Just use URP (Unity) or Compatibility (Godot) dipshit. If they can make 60fps on mobile, it's definitely lightweight on desktop. Unless you want to spend 200 hours writing your own OpenGL engine in C.

>>105683422
By light, I assumed it would run on battery powered Intel Core 2 Duo, but I guess by old hardware, you meant something like i9 Intel desktop CPU overclocked beyond what should be possible.
You can stop posting now, faggot.

>>105683444
It's not the 2000s anymore gramps

>>105683501
Fuck off subhuman faggot, go use your mcslop engines, don't be surprised when it's a slow piece of shit that doesn't work on older hardware that still works just fine.

>>105683524
stop posting

new bread

>>105683601
>>105683601
>>105683601