Procedural Black Magic 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.

picrel : https://x.com/PounceLight/status/1950165107163431424

Previsous: >>106065067

>>106075373 (OP)
Wow, black magic indeed.

Does (your) game has such a thing?

where can I find some sources on animation graphs input buffer for user inpfut, good practices for resource usage, rendering architecture, multiple renderer solutions and other topics?
I am about as complex and 2d like wesnoth

>>106075304
real thread: no cringe indian gatekeepers

>>106070823
don't let game companies ruin your fun.
Ruin it yourself by intensely learning game dev.
>>106073229
you'll still need triangles for drawing your shaders on a quad saar

>>106075431
No. Not yet. I'm working on it.
It'll take some time because I'm focusing on making the procedural tools I create to be as generic as possible so that they can be reused and not stay confined to a particular art style for a set of props.
>>106075501
>animation graphs input buffer for user inpfut
you mean like an animation graph editor?
>good practices for resource usage, rendering architecture, multiple renderer solutions and other topics
what backend are we talking about? OpenGL? Vulkan?
you usually abstract your code into application's logic "front-end" and renderer "backend" so that your application can be configures to be able to run on multiple renderers.
I'll just leave these here
https://github.com/Calinou/awesome-gamedev
https://github.com/FronkonGames/Awesome-Gamedev
>I am about as complex and 2d like wesnoth
looks like a turn based combat game. If it'll be stateful then its better to not rely on ECS. Maintaining the entire state of your game in one big arse struct and changing it as gameplay progresses is one way I could think of starting with this.
>>106075558
>

>>106075373 (OP)
>they built their own engine in rust
neat

I know sound effects are important but it is really boring to assign them for all of the game actions, items, and objects

>>106075672
Best AI game dev: https://grok.com
10 games made with grok 4 heavy (only 300$/month)

https://x.com/EHuanglu/status/1944708868766105863

stop gatekeeping pojeet

I think we have a clear consensus that we are going to ignore the existence of the AI schizo

>>106076780
I mean there's nothing to say.

>>106075501
Unfortunately there’s no book tailored specifically to renderer architecture practices. You’re going to have to scour blogs, posts, repositories, and other rendering books to get the frame of reference you need.

FYI Khronos updated their Vulkan tutorial

>>106076764
for someone who couldn't justify why it should be grok 4 and not any of the other chatgpt, claude, gemini variant, you seethe too much.

Whoever made that crap ass addon G.U.I.D.E. for godot I hope you people fuck off forever and delete your code or rewrite it as a proper C++ extension. Most worthless crap I've ever used. I had to go through and strip it out of my code base entirely due to the IMMENSE amount of problems it causes. After removing it my frame rate went up an extra 40-60 frames. Something meant to purely handle input should not be that heavy. You're better off just handling the input yourself to save the headache later.

>>106076964
they all work, grok4 heavy and claude opus work the best. is it that hard to understand? the other models are dogwater

>>106076411
Gay, memelang

>>106077170
>still doesn't justify why it has always been glok4
>is it that hard to understand?
yes.
Despite claude opus being equally good but only grok4 making it to the OP titles makes it really hard to understand.
And both, if not, all AI slops are shit unless paid.
come back when there is no slop and no money involved.

>>106077226
what are you poor? or stupid? claude is only 100 bucks a month. grok4 heavy is goated because it uses 4x agents per response and the most compute per reponse out of any company out there. never doubt elon.

I wanna make a camera gizmo like blenders. How do I calculate the point for the near plane and far plane of my camera frustum?

>>106077246
>what are you poor? or stupid?
Honestly, I'm both
>claude is only 100 bucks a month
expensive doesn't always mean better.
>never doubt elon.
I will but I'm not doubting him rn. I'm doubting you instead.
>>106077248
>How do I calculate the point for the near plane and far plane of my camera frustum?
you can do the good old (ray_origin + t * ray_direction) trick where you can get the edges of your camera from the aspect ratio and by computing t for near and far plane in all edge's directions, you can get the points. Here is more info about this approach from raytracing in weekend series
https://raytracing.github.io/books/RayTracingInOneWeekend.html#rays,asimplecamera,andbackground/sendingraysintothescene
Its also becomes simpler if the gizmo is drawn as a line primitive using an additional draw call.

>>106077248
points*
>>106077369
That tan equation right? I also heard you can use the view projection matrix to get the points

>>106077481
ah, I had a feeling that I was missing something and you pointed it out. I was missing fov and that is the tana equation.
I haven't tried inverting view projection matrix but its simple. So in case you have difficulty finding resources with googling, free tier AI will be able to guide you.

>>106075373 (OP)
The whole area looks horribly sanitized and artificial. Kind of safe pixar-mcdonalds fantasy style.

To be honest, if people are needed to "create" that, then replacing game "devs" with AI makes sense.

>>106077619
Shut the fuck up about AI brainlet

@grok review this guys post history and insult him with a deep cut

>>106077640
This looks exactly like the models that we were shown when MS was demoing AI to us few months ago, desu.

Exactly this flat, soft, blizzard-Asian-pixar look, that looks like ripped from an asset store.

>>106077667
>>106077640

>>106077640
keep ignoring ai and watch the ai enjoyers absolutely destroy you in execution quality and speed.

>>106077732
>watch the ai enjoyers absolutely destroy you in execution quality and speed.
lmfao
wheres your game ranjesh

>>106075373 (OP)
what do you guys do to stay motivated? I always have a neat idea to implement that shouldn't take more than a day but then I get annoyed over a small detail and loose days trying to fix some bug loosing motivation in the process. I thought it was a problem with the language but turns out its a general problem...

>>106077781
My desktop wallpaper is a corvette. Always keep your eyes on the prize.

what are the system requirements for Unity? my shitty laptop ran like shit and hot as fuck. Im saving up for a pc, didn't want all parts to be new tho.

>>106077922
5800x3d+, and 2080ti+. anything less and your watching the loadbar longer than your devving

>>106075373 (OP)
One of the devs of that game gave a technical talk (not GDC) about their custom engine.
https://youtu.be/jusWW2pPnA0?t=334
He also did this
https://github.com/EmbarkStudios/kajiya/
>Experimental real-time global illumination renderer made with Rust and Vulkan
>Done in my spare time at Embark, kajiya is an evolving excursion into real-time global illumination algorithms. Having gone through surfels and morphological inverse cone tracing, it's now fully infused with ReSTIR and irradiance caching.

>>106077928
Fuck, I was thinking about r5 5500 and rx 580 since here is 3rd world country and im also a poorfag. Maybe i'll stick to Godot then.

i made 80 euros so far selling ai made 3d models

any tips of what more i can get AI to produce to sell?

>>106078106
fuck off

>>106078111
what why?

>>106078114
the neckbeards here are unironically too poor to afford a decent ai subscription

>>106078114
not the AI thread

>>106078135
yes but im making 3d models for you devs
so basicly im a gamedev eventhough i know nothing and cant code but im good with AI prompts.

i should try to make a entire AI made game to sell on steam

>>106078165
>so basicly im a gamedev
no you arent, go away

>>106078180
i am i made a game product so that makes me a gamedev kys

anastasia opara is a very clever girl and my gf

>>106078731
whats that? A russian escort?

>>106078747
yes. she make game in op.

>>106078720
how old are you?

I've gone through CS50 and started making a 3D game in Unity, but watching/reading tutorials with lots of code still sometimes leaves me feeling out of depth because it's rarely explained what code does what and why

Are there any books or resources that would give me a stronger foundation for 3D game dev in C#/Unity?

>>106078774
What code don’t you understand and why? That’s easier to answer than general advice for 3D game dev.

>>106078757
why does it matter?

>>106078785
I was watching a video on multithreading and the guy just said what to do but never explained why, why the code is structured in that specific way, why this specific thing gets multithreaded but others do not, why the code goes into this script and not into a separate one, it was my first time seeing callbacks etc

Now I could hunker down and go through multiple threading resources to understand it, but ideally I'd like to have a basis for understanding why things in general are done the way they are in Unity, that would let me make my own conclusions and understand the monobheavior better in order to know how to write my code around it

>>106078787
you act like a kid

>>106078811
do not start with multithreading, implement your shit in singlethread, and only learn multithreading when you actually need it. you will forget everything you learn if you don't use it often

>>106078842
Well I didn't start with it, I was watching a tutorial for something else and that was one part of it, and it immediately made me feel like I have a gap in my knowledge, and I prefer even roughly understanding things in advance to blind trial and error

why do people pretend garbage collected languages are slow? every engine uses it and it happens off-thread with 0 performance impact.

>AI GENERATED CONTENT IS LE BAD IT JUST IS OK??!!!!11

>NOOOO HOW DARE YOU MAKE A GAME WITH JUST AI IF YOU DONT KNOW HOW TO CODE!!!111

>>106079065
this is the coding thread dumb fuck

I posted this code yesterday and worked 2-3 hours to get a better solution... I hope this time it's better. (I should rename enemiesToSpawn into something els

old version:
void SpawnEnemyWave(int enemiesToSpawn)
{
for (int i = 0; i < enemiesToSpawn; i++)
{
// enemies
if (enemiesToSpawn <= 3)
{
Instantiate(enemies[0], GenerateSpawnPosition(), enemies[0].transform.rotation);
}
else if (enemiesToSpawn <= 6)
{
int randomIndex = Random.Range(0, enemies.Length);
Instantiate(enemies[randomIndex], GenerateSpawnPosition(), enemies[randomIndex].transform.rotation);
}
else if (enemiesToSpawn >= 8)
{
int randomIndex = Random.Range(0, enemies.Length);
Instantiate(enemies[randomIndex], GenerateSpawnPosition(), enemies[randomIndex].transform.rotation);
}
}

// bosses
if (enemiesToSpawn == 7)
{
Instantiate(bosses[0], GenerateSpawnPosition(), bosses[0].transform.rotation);
}
}



new one:
void SpawnEnemyWave(int enemiesToSpawn)
{
for (int i = 0; i < enemiesToSpawn; i++)
{
GameObject enemyToSpawn = GetEnemyToSpawn(enemiesToSpawn);
Instantiate(enemyToSpawn, GenerateSpawnPosition(), enemyToSpawn.transform.rotation);
}

if (ShouldSpawnBoss(enemiesToSpawn))
{
SpawnBoss();
}
}

GameObject GetEnemyToSpawn(int count)
{
if (count <= 3)
{
return enemies[0];
}

int randomIndex = Random.Range(0, enemies.Length);
return enemies[randomIndex];
}

bool ShouldSpawnBoss(int count)
{
return count == 7;
}

void SpawnBoss()
{
Instantiate(bosses[0], GenerateSpawnPosition(), bosses[0].transform.rotation);
}

>>106079101
Bad code
Chat gpt wrote it better

>>106075373 (OP)
thats vibecoded

Hypothetically, let's say I wanted to statically use directx 7 (not 9,11,12) in my game written in c, how would I be able to archive this so it is embedded in the binary I compile? Is the thing I want even possible?

hey bros, do you think an ultrawide monitor is good for game dev? godot editor UI got me just wanting to go super widescreen so I can just properly lay everything out and not be too cramped, and I imagine that splitting the godot editor over multiple monitors with bezels in the way might suck.

although maybe if you had 3 monitors, and the center editor was 16:9 on the middle monitor full screen, then the file browser and inspector are pinned to the left and right monitor just by the inner bezel, and then you could have a web browser taking the other half of the left screen and a video player if you follow tutorials on the other half of the right screen maybe, idk

I think just have a single ultrawide curved monitor with no bezels is going to be ace though. I almost ordered a samsung ultrawide but I've seen too many complains about it being a VA panel and being poor for productivity, so I'm looking at what IPS and OLED options are available, either AOC / Philips / MSI at the moment.

Does anyone here already use an ultrawide monitor for gamedev?

>>106079397
You can't embed it and nobody will be able to play your game, just use OpenGL.

>>106078925
Watch literally any talk about "how we shipped game X" or "how we ported game X to mobile/consoles" where X is a game made in Unity. The talk will include a section about all the gymnastics they had to do to avoid GC pauses.

>>106079397
yes its possible, I think the "easiest" way is to use MinGW not MSVC, and floating around on the net somwhere are the MinGW DirectX 7 SDK files which are DirectX headers ported to MinGW

the compiled program can only depend on MSVCRT and other windows built ins, no .dll needed. So long as windows is new enough to come with the directX stuff built in, which I think WindowsXP and up do?

I found a place that says they have it
https://liballeg.org/old.html
you might not even need those though

>>106075373 (OP)
I declined to buy summerhouse. It was made by nazis.

>>106079494
the summerhouse dev's shout seig heil and nazi salute? surprisingly based... You'd think they'd not do that to not hurt their image and reduce their sales though.

>>106079485
Very helpful post, thanks.
Just not sure how to proceed but I will try to look around for a bit.

>>106079440
ultrawides are dumb for most use cases. really inefficient use of screen real estate but obviously there's exceptions like it being great for video editing or racing games. imo you're better off with 2-3 separate "normal" monitors for the majority of things.

>>106079559
If I use a program to pin godot editor to always be in a fixed location, I'm sure I could go 3x multi monitor. I wonder what it would be like to have 3 monitors, but turn them all sideways...

>>106078754
I first came to know of her from her houduini tutorials and was surprised when I saw that she was one of the two who made the game. A true gem.
https://www.sidefx.com/profile/anastasia%20opara/
>>106078925
>why do people pretend garbage collected languages are slow?
why pretend when its actually the case though?
you need to understand that memory management plays a huge role in performance and a generic garbage collector can't do well in all scenarios. Each problem demands its own way of memory allocation strategy for most efficiency.
>>106079065
AI generated content is le bad because it doesn't give me each asset with the level of detail that each of the them would need for them to feel like all of them belong to the same virtual world. That is very important for the quality of the game.
It takes some artistic abilities to even realize that part, not something a bland AI asset creator, supposedly "game-dev", would notice or even understand so you won't get it.
AI generated content is good when the situation is like "something is better than nothing". Its not good enough for usage in an actual in-game content generator. In the end, you are better off relying on actual artists for creating a good quality game.
>>106079494
>It was made by nazis
not a problem for me, actually.
Besides, the name of summerhouse dev is "Friedemann". I'll cut his some slack just for that name and ignore the natzi part.

>>106079620
do whatever you want, stack three ultra wides on top of each other if you want. you can make anything work even if it looks atrocious.

memory management? just load the whole game into memory!

>>106075373 (OP)
how did they do the lighting ? shit looks like real time cycles

>>106079646
>her houduini tutorials
yes. i think her lake houses series was far and away the best procedural modelling tutorial for its time (might still be).
i think tiny glade is entirely custom tooling in rust. it was easy to tell she'd grow beyond houdini tho even back then.

typing anything programming related automatically generates an AI response on google. so far it has been much better than stack overflow
this is merely an observation devoid of any personal opinions about llm tech

>>106079800
>typing anything programming related automatically generates an AI response on google. so far it has been much better than stack overflow
>this is merely an observation devoid of any personal opinions about llm tech

>>106079696
you mean to load all the levels that the player currently isn't in and all the 3D assets of the props and NPCs that won't appear within the current gameplay level as well? Would be nice if everyone had over 100GB of ram and vram.
>>106079708
>>106077948

>>106079744
her tutorial felt distinct when I came across it.
I wish she went hardcore like Konstantin Magnus and created more tutorials regularly lol
>>106079800
>so far it has been much better than stack overflow
for simple problems, AI seems to get the job done, I won't deny that.
But things go to shit as soon as i tell AI to deal with things that aren't simple and straight forward. It still get something done and that something just isn't better almost all the time. Too much slop in there.

>>106078754
looks like the only thing she made are blowjobs for actual developer

>>106079485
>>106079549

I have found some samples and am trying to learn from them but I have some problems getting mingw to find the command 'i686-w64-mingw32-windres'.

I solved it for finding gcc but not sure what to install on mine to resolve windres.

>>106079896
>looks like the only thing she made are blowjobs for actual developer


NOO I MUST BE MYSAGONISTIC!!!!§!

>>106075373 (OP)
What is more important, art or music?

>>106079933
>What is more important, art or music?

>>106079922
Nevermind it is installed on mine and I just needed to change the shell script to only invoke the binary 'windres' and not try to invoke
'i686-w64-mingw32-windres'

>>106075373 (OP)
Wow that webm is incredible. So smooth.

>>106079933
what is most important is code bloat and the size of your executable

>>106080054
>Wow that webm is incredible. So smooth.

MAME and ReactOS devs are literally very involved in using old APIs and libraries on modern hardware. It goes deep huh.

>>106079397
embed it as a binary with the resource compiler, then at runtime create a temporary file from that

>>106079933
both are important.
But their importance is prioritized based upon the requirements.
For games like osu, music is very important but art... not so much. Numbers and circles alone are more than enough for the gameplay.
For visual novels games, art is very important and music... its fine if the music sucks here but good music plays huge role for the player to emphasize with the characters or the environment, which makes the visual novel more intuitive.
For games like pong, neither of them are important. The gameplay will be totally fine with poor art and music but good art and music do help with making the game look more polished and distinguished.

>years of history in c-like scripting, maya mel/houdini script mostly and some actual c programming
>because of ai begin to dabble with python
>been over a year now and i still fucking hate python, can't understand lists and loops and how to iterate them properly, because they are not clearly defined (for me)
>made my own ai model client + text adventure interface with rooms, it works but need help with any more complex string manipulation I encounter
I'm way past of looking at simple exercises like apple + apple and most of the string related stuff seems to be like this on internet at least. Python is horrible.

>>106082296
plus every chatpajeet ai answer almost always includes REGEX by default unless asked to change this. What the hell is that about? This is not excel for fuck sake.
In essence every piece of advice I have puzzled together always conflict and contradict somehow.

Imagine when we can design our own houses like this and build drones make it happen.

i reopened a project i started 11 years ago to learn c++ in depth
all i did was follow a tutorial but im still proud of it

now i will not continue that project and will simply move on to ue5, thank you

>>106079484
>GC pauses
this literally cannot happen unless you run out of memory unity faggots don't know how it works

>>106082844
>UE5

>>106082844
Sometimes early work is your best because you were enthusiastic back then. Now? Just a cynical old husk.

>>106080249
are we really turning into the old faggots we hated that hate on younger people?

>>106083562
im expecting him to work and release his ue5 improvements

>>106083593
it was more about how to manage a project than straight coding in c++, it was really just a tutorial teaching the basics
but its really fun just making shit work and happen from scratch even if im not going to be releasing anything from it

>jai
No more excuses gedg

That's it I give up on my engine it's not worth it

>>106079744
>it was easy to tell she'd grow beyond houdini tho even back then.
Houdini is some of the slowest dogshit software I've ever used so I'm not surprised.

>>106010717

polyhedron plural polyhedra == poly-hydra

rewrote the main data structure to use double edges, here blue edges are unused and red edges point towards their face. you can see that they obey a certain direction/flow, this will make doing edge slide and edge loop selection easier
it also supports Ngons and the triangle generation becomes trivial

also implemented a basic UI container system that positions their children and grow to acomodate them. not sure if i should do another recursive pass expanding the children to fit the parent's new dimensions

>>106077948
tldr?

i cant be bothered to watch some onions spastic babble about "mesh shaders bad" as if anyone except armchair reddit developers ever thought they were a genuine pipeline advancement. stunning hot take, i know. wait until his target audience find out most artist and cad tooling still runs off gl draws without so much as shared memory mappings / user provided buffers.

>Experimental [...] made with Rust and Vulkan
ah ok so nothing of value is worth digging out of this talk? got it

>real-time real-time real-time real-time
Congrats, your program runs with frames per second instead of seconds per frame

>>106086228
the talk and the github link are 2 different things, the talk isn't about the experimental renderer

>>106079485
mingw is a terrible compiler to use because it lacks address sanitizer, and you are going to want to use vcpkg (for other libraries such as devil, and you can build libraries statically without a lot of manual autism, and you probably want dynamic libraries for development because it should speed up linking time), and vcpkg works with the visual studio project file (the one good thing about visual studios project file is that you can add and remove code files from your project without modifying a text file, otherwise cmake is the best alternative, linking to libraries without vcpkg is very painful).
If anon was using C++, I would agree MSVC visual C++ runtime is annoying, because I don't like shitty defaults, the runtime should have been part of windows or something (this can be fixed by changing /MD to /MT in your project/cmake + your libraries), but since anon is using C it doesn't matter.
Also I don't know which flags you need to build mingw so that it doesn't link to mingw specific libraries. I know that there is 1 DLL for C++ (no need since using C), 1 DLL for pthread (no need since win32 api works), and 1 DLL for SEH exceptions (libgcc_s_seh), which is part of C but I just hate shitty defaults.
Like OK, that's just 1 flag, but on msvc, it's also just one flag (/MD -> /MT, it's very easy to change on a visual studio project, AND you can also add a new target for address sanitizer for debugging).
I also don't use DirectX, but I have a feeling that there are DirectX specific tools (similar to RenderDoc/nvidia Nsight), and those tools probably depend on msvc debug info OR it's only a visual studio Plugin (the DX/opengl functions should be easily detected since that's external, but I have seen tools access your source code for locating the location of the DX/opengl call).

>>106086532
>but on msvc, it's also just one flag
Actually you don't need a static flag on msvc for C, I don't think there is a single reason for C to be build with the static CRT. Maybe windows XP / performance, but that sounds pointless to me.
Also I believe DX 7 is just DX 9 but with less feature levels. It's sort of like in Opengl, you could select GL 4.2, but effectively you could be only using optional extensions that most GPU's implemented in opengl 3.2. I think DX 10 (which most people skip and use DX11) is when they actually break the API like Texel sampling being off by 0.5 and probably more, correct me if I am wrong however.

>>106078925
its not slow but its not good for games because when the gc kicks in its causes a lag spike

>>106086532
>>106086667
What compels the zoomie to babble this much over topics out of their reach?

>random capitalization
>DX7 is the same as DX9
>mismatched capitalization between OpenGL and DirectX
>i need address sanitizer
>wtf this is shit. why isnt there clangs address sanitizer here
>babbling about "1" configurations for "DLL for C++", "1 DLL for pthread" and another configuration for "exceptions"
>...."which is apart of C"
>"i dont like shitty defaults"
>babbling about visual studio
>...visual studio capital-P "Plugin"
>...
>etc
over a fucking ancient gcc distribution. kys

>>106086791
Everything you posted is opinion, but could you expand on DX7 not being the same as DX9?

>>106086833
dx9 supports shaders, dx7 is fixed pipeline

>>106086844
DX7 supports a limited form of shaders, but it's the first version of the shader model, obviously shadow casting and other fancy effects couldn't be done in the first version, but the API is essentially the same (if you wrote a DX9 game, there would be ZERO excuse to not support DX7 backwards, because you don't need to implement a new backend, it's the same API, similar to Opengl versions, obviously special effects and details would be lost).
I was kind of hoping that you actually knew about DX7 and you could correct my view so I don't repeat it (because I will).

>>106086877
apparently dx7 doesnt have pixel shaders but dx9 does so that's a pretty significant difference

yeah my whole argument was wrong :)
But a dx7 GPU it might support shader model 1 such as the GeForce4 MX 440-SE (2002), but the Geforce2 (2000) does not.
(Also side note, why does the GeForce3 Ti200 (2001) say it supports DX 8 on techpowerup, but the newer MX 440-SE (2002) says it only supports DX7?)
Obviously I am wrong however, and I'm moving the goalpost, I didn't know DX7 didn't have shaders.

"What compels the zoomie to babble this much over topics out of their reach?"

>>106086844
>>106086877
>>106086910
tl;dr
>>106086953
>I didn't know DX7 didn't have shaders.
Mental illness.

I want to start doing combat stuff soon
- First the more chaotic survivors game loop
- Then the turn based combat
- A party system for managing multiple "characters". But what about controlling hordes? Imagine 1 zombie god vs 6 players, where zombie god commands the legion of zombies. I think minecraft already did it but whatever.

>>106087072
where are the procgen dungeons ?

>>106087072
Bro is it you? YGG dev?
I haven't been on 4chan in ages to see you again is huge how is your game man

>>106087036
>Mental illness.
Anon, I'm just trying to learn more about DX7, you don't seem to know much about it either because it seems like you have not used it. If you did, you would have already shown me examples that would break your code.
I know that opengl's fixed pipeline technically had shaders, through ARB shaders (I still don't know what to call them since they have a lot of names, it's basically shader assembly, used in games like quake 3 and doom 3).
When you used shaders with the fixed pipeline, I don't think it's actually that different.
I might be wrong, but I think the GL 1 -> GL 2 transition is actually the same as the DX9 -> DX10 transition, but I don't know, since I have not actually looked at DX code.

>>106087339
DX7 is the equialvent of OpenGL 2
DX9 is the equiavlent of OpenGL 3

>>106087072
Did you try the new RuneScape game? Did you like it and take any inspiration?

>>106087148
Good point, also coming in the list. I actually had some procedural generated buildings and other neat things in my last iteration. Should be quick to put that back in place since its just loading to a database.

>>106087238
I am pretty happy with the current direction and state. I took a break while working on other projects for network automation. But I am working hard on next refactor of things, and can do all kinds of chaos.

>>106087383
I watched about an hour of gameplay, looks well received by the players. I try to keep it all in mind. For my own vision, I am hoping as also add "game modes" or game changing mechanics that something may strike a note with people. Maybe Ygg Engine is a farming with friends game? Shop keeper game? Habbo hotel clone? Once I have the core systems in place I can continue to experiment with things other than the traditional questing and rpg experience.

Also maybe worth sharing this sprite stacking working concept, I told myself if Ygg Engine becomes full time for me I will proceed with the alternative camera views like this.

>>106087354
Shader Model 1 (DX7~/DX8), did not use hlsl, it used assembly, that's why I would compare it to Opengl 1 + ARB shaders (which is not the same as Opengl 1, I know that at some point opengl added a feature to store vertices instead of constantly writing, but without a modern VBO syntax).
Opengl 2 added VBO's, and glsl, so I would consider it the same as DX9 (and I remember having an intel core 2 duo that had DX9 and opengl 2.1 support).
BUT I don't think you can compare DX9 to Opengl 2. I think opengl transitioned to a new API syntax before directx9 (breaking the code by forcing large changes, sort of, GL 1 looks nothing like GL 2-3 code), while I ASSUME DX9 continued to use the same syntax as it did in DX7 (and maybe before). And then in DX10 they broke things.
But like I said, I don't KNOW because I don't have experience or time to actually compare a DX5/7/8 vs DX9, BUT I have a gut feeling that you could combine both into the same backend (that's why I said DX7 is pretty much the same as DX9).
If you wanted to support Opengl 1 + Opengl 3, you would probably need 2 backend implementations because the API is too different (glEnableClientState with a VBO GLint?).
while DX 7/9 could be implemented through simple feature test branches (similar to how you could support GL 3.1 and GL 4.2, it could be very different, but you don't necessarily need new structs to support the old versions, usually).

use case for pixel by pixel lighting system?

>>106088171
specular

>>106088171
it looks better in all cases

audio test https://i.4cdn.org/wsg/1753932337143307.webm

>>106088411

>his game doesn't use tcp over localhost in order to allow interactive programming.

Problem is that my workflow needs a robust autocomplete system, and nothing has satisfied me. Geiser is bugged and won't show autocomplete for symbols defined before connecting over tcp.

I need something that can read the symbol table I think.

Guess I'll have to learn how to write my own minor mode in elisp so I can finally work on my game ...

>>106089017
How do I get in?

>>106089021
Get into what? If you mean the tcp connection, its over localhost kek.

nrepl is a chicken egg for development its not recommended for exposing over the internet kek.

Or are you asking for details on how I did this?

>>106089034

>>106085137
and yet still somehow the best dcc

>>106085137
Houdini is the best.
Blender is coping real hard with its geometry nodes just to mimic a fraction of houdini's powaa.
Houdini is also not slow. It mostly uses CPU so performance will be poor for old CPUs but they do plan to use vulkan and rely on hardware acceleration so it'll only get much faster.

is CLion worth learning now that it is free? or is it not worth messing with even if it doesnt cost anything?

I'm fucking writing my game to use Amazon SQS and Amazon DynamoDB so that no Stop Killing Games fag can actually deploy the released server binary without going bankrupt

>>106078165
What ai do you use?

>>106079933
Art by far

>>106075373 (OP)
how the fuck

>>106075373 (OP)
these are always very cool, but when these people finally make "games", they're can't get over themselves and make the procgen too prominent and end up making toys, not games
many such cases
that is if they ever end up making a finished product at all

>>106091112
are games not toys now

>>106091112
What is wrong with toy games? "Cozy" games are an entire genre. Not every game needs some kind of purpose or conclusion.

I don't see how basic mesh editor is a "game". That's like selling a unity FPS prototype as a game.
For me this is just laziness and incompetence being rewarded in current eceleb hype economy.

I would imagine tiny glade as a town building game prototype. Imagine you buy warcraft 3, but it's only level editor. That's modern game industry.

It's $15 and really just there as a kind of tech demo that you buy as a gesture of support and to help fund bigger projects. Absolutely nothing wrong with that.

midwits complaining a 2 person dev team where both people are orders of magnitude better than them.

>>106091075
blender geometry nodes but you don't need to model the assets and uses lumen instead of cycles

>>106090673
I'll add to the question. Does jetbrains rider live reload work fine with c++ or I shouldn't bother switching? If it's good I'm finally ditching visual studio and jumping ship from windows.

>>106091242
>are games not toys now
>>106091254
>Not every game needs some kind of purpose or conclusion.
it really depends on your definition of "game"
if you're not retarded, they're not toys and they do have goals. by definition
but this is a discussion that has been going on since before you were born, probably
and this isn't about quality or validity or maturity, it's just that I, personally, like games more than I like walking simulators, sand boxes, interactive fiction, etc.

>>106080249
Not a zoomer I just thought it looked cool :(

>>106090673
>is CLion worth learning
if you are dealing with C++ shit then it does feel comfy and apart from that there is no reason to use CLion. VSCode with clangd is good enough too.
>Does jetbrains rider live reload work fine with c++
the last I used it with unreal, it was the best and live reload did work fine.
>>106091075
Procedural black magic
>>106091112
>and end up making toys, not games
that is exactly what minecraft was and still is.

>>106091980
Edit and Continue does not work with address sanitizer. Address sanitizer is just more useful compared to Edit and Continue in my opinion.
>>106092096
>VSCode with clangd
You forgot to include the part where on windows, VScode's clangd extension does not work with msvc, but it magically works on clion because clion has a special version of clangd that works with msvc.
TECHNICALLY vscode's clangd works with msvc, but it will give clang (clang-cl/clang on windows) diagnostics and warnings, but I just use intellisense on vscode because vscode does not handle multiple build folders well (I set separate build folders with cmakepresets, vscode clangd works fine with 1 build folder but it gets annoying with multiple build folders, unlike clion/VS studio), and clang is not compatible with the /MDd + Address sanitizer flag, while msvc works fine with those flags, which means clion's version of clangd fixes that issue.
Clang-cl / clang on windows doesn't support address sanitizer as well as msvc (clang will not stop the debugger, and some minor JIT debugger stuff which needs VSAsan global variable settings to enable, and I would prefer not using /MDd but I like the exception/assert dialog popup and debug iterators has caught me lacking once).
I don't like clion that much however, I think the "non-commercial" part can be problematic, and for a bigger project (50k LoC+), clion eats 16gb of ram easily (but it depends on how many files you have open in the IDE, and if you have a multiple web browsers open with 100 tabs).
If I didn't use linux / emscripten I would probably use Visual Studio's IDE (The reason I don't use VS with cmake is that PVS studio extension does not support cmake, but I wouldn't need it if I had automation so that pushing commits would check with PVS...).

>>106092619
>VScode's clangd extension does not work with msvc
did you try generating a compile_commands.json file?
https://clangd.llvm.org/design/compile-commands
you can generate it with build systems like cmake, ninja, tup, scons or bear.
clangd works as fine as intellisense with that one file in the project's root directory. It worked well for multiple directories for me. I even get struct's size and its field's padding info popup on emacs. Hopping to-and-fro from definition to definition became very effortless, at least in my experience. Ram usage was quite high compared to other LSPs but its still decent when compared to intellisense.

>>106092014
>I, personally, like games more than I like walking simulators, sand boxes, interactive fiction
I actually do like walking simulators and things like visual novels. So yeah, it's all just subjective in the end.

>>106092871
I don't have issues with clangd working on vscode, I prefer using vscode on linux (and it works fine with multiple directories, clangd does crash once in a while, but it happens on clion as well I think).
BUT msvc (or clang-cl) + address sanitizer won't work well on vscode + clangd because msvc will compile but clangd is going to think it wont compile because it thinks you are using clang, unless you use clions version (You could try to remove /MDd, but I do like debug iterators, and it's a major conflict with vcpkg, I think conan has a more simpler but cryptic workaround to allow you to build your binary with debug info but link with release libraries, but I have not tested it).

>>106093056
eating feces is one of those "subjective preference" too. there is a moral value attached to any "subjective experience". some are degenerate, some are progressive.

>>106093101
wakatta

>>106090673
Jetbrains IDEs are generally the best, so yes they are absolutely worth it.
>>106092096
>if you are dealing with C++ shit then it does feel comfy and apart from that there is no reason to use CLion. VSCode with clangd is good enough too.
How the fuck can you suggest VSCode to people? Dumbfuck tripfag, as usual.

>>106093318
kys moralfag

>>106093521
>How the fuck can you suggest VSCode to people
it is good enough though.
Travis Vroman shows how to set up VSCode in his kohi game engine series and that thing's got even a debugger with breakpoints working.
>Dumbfuck tripfag, as usual
ayy

>>106093687
>Travis Vroman shows how to set up VSCode in his kohi game engine series and that thing's got even a debugger with breakpoints working.
Wow I only have to watch one game engine series to be shown how to set up something you get out of the box in any Jetbrains IDE? Holy shit sign me up for the electronslop!

visual studio is obviously the best ide

hello, anons. noob here. for programming, is there a prerequisite or soft dependency for learning certain languages? as in, would one need to make headway into the innerworkings of one language to better appreciate/learn another?
i'm wanting to start modding games i like as a hobby, and it looks like C# would be the language to start learning given the tools of said games i want to mod. i assume there's no issue with just jumping into the language i'll be needing to learn and not worrying about anything else yet?

>>106075373 (OP)
>been making games since I was 9 made a bunch of different stuff never actually stuck to one project

how do I get a game idea I want to stick to

>>106094937
You can grab anything and it will give you a picture of how things are done, except for memory managed languages, which are a different kind of beast, but those would give you the universal understanding of other languages by letting you see past the abstraction and realizing they are all shit.

>Odin driving me nuts
>can't even initialize default values in structs
I really like Odin's syntax and most things about it, but some things are just fucking obtuse. Why the fuck would you not have default value initialization in structs? Oh so me declaring "random_global := 5" is fine, but me doing that exact same thing (except not global) inside a struct is a big no-no?

>>106075373 (OP)
neat, seems they've added a bunch of stuff since I last checked up on that project

>>106095119
>Why the fuck would you not have default value initialization in structs?
And now you know why people who actually make games just use C++, despite it being "bloated", "shit", "the big OOPs", and whatever else. All these memelangs are just straight up hostile to the user. Initializing stuff in structs might be unclear. Way better to have you jump 5 flaming hoops to do what a simple x := 1 would have done. So much easier to follow some convoluted automatic struct initialization system than just go-to-definition and reading a number.

Same reason there are no namespaces in Odin last I checked. Why let the user easily organize functions when you can just force him to prefix his functions manually to know what the fuck that is_valid checks validity of? Why call them functions like everyone has done for decades when you can be a pedantic sperg and make it just a little more annoying by calling them procedures instead?

>>106095067
i will keep this in mind and just start banging my head against it, thank you.

>>106094418
I use nvim btw, but for gamedev on windows I don't know why you would use anything other than visual studio. Not VSCode, but visual studio.

why would anyone use windows

>>106095798
The main reason to use any other IDE other than Visual Studio is that you can use vscode on linux and windows, and apple maybe?
It's nice to have the same IDE on different platforms.
And clangd is a thing.
My reason for wanting clangd is so I can hover over structs member variables and seeing the size and offset of each variable, it's very convenient and it helps give you a hint into what's inside external code (like how large is this STL container / iterator).
Also there is a slightly different style for hovering over variables to get the type between clangd vs intellisense (I think intellisense is more verbose? which could be good or bad, might be a flag option).
clangd also has refactoring stuff but it doesn't always work (it can botch the refactoring and fuck it up, I think clion will give a warning before undoing all the changes in all the files, and I fallback to manually replacing all the text with ctrl+h).
But Visual Studios is unavoidable, I use it for JIT debugging and reading core dumps (sometimes it's possible for a crash to only happen without a debugger attached, BUT such a crash might be once in a blue moon on a release binary, so it's hard to reproduce it on address sanitizer or valgrind / dr memory to check if you are doing something bad with invalid memory or uninitialized memory, so all you have is the JIT debugger or coredump on a release build with debug info + disable inlining). Technically clion / vscode can read core dumps, but they can't be a JIT debugger, which replaces coredumps, and VS is the least intrusive JIT debugger unlike windbg/etc, because it displays a prompt before opening.

For whatever reason, I get the idea that C++ is just a better language to learn for graphics programming and/or game dev.
It feels like Rust is just not good for this. However, I feel more comfortable (slightly), at least at the moment, with Rust than with C or C++, so in part it makes me feel like I should stick to Rust.
Keep in mind, I am not an experienced programmer and my opinions could be wrong or limited.
Can anyone please educate me if I'm wrong?

>>106096074
idk, there has to be a reason. It seems like everyone just fall into using Visual Studio and save headaches with libraries or dependencies

>>106096339
Rust is fine for graphics programming/game dev OP is made with Rust.

Rust is for trans

>>106096339
you got the wrong idea. for learning, C# is the best. best development tools, best debugger, best expressivness, best features. no need to fight undefined behavior or borrow checker.

>>106096074
my opengl game doesn't even work on linux because the mesa drivers don't work for bindless

>>106097251
Are the drivers you're testing with like 15 years old or something?

>>106097272
https://gitlab.freedesktop.org/mesa/mesa/-/issues/12036
linux literally does not work with amd, lmao!

Didn't know these existed.
https://clang.llvm.org/docs/LanguageExtensions.html#vectors-and-extended-vectors
https://clang.llvm.org/docs/LanguageExtensions.html#matrix-types

>>106098042
std::vec3 would be funny

>>106094215
>one game engine series
one video, the first one, of one game engine series, to be precise
>>106094987
>how do I get a game idea I want to stick to
you don't. That's some advanced game dev shit.
You start out by picking an existing idea and working on it.
>>106095119
>got filtered by v-tables
>so me declaring "random_global := 5" is fine
nothing's stopping you from declaring constant structs and using them to initialize your struct though
MyStruct :: struct {
x : i32,
y : u32,
z : f32,
}

INIT_MY_STRUCT :: MyStruct {-4, 3, 2.0}

main :: proc () {
s := INIT_MY_STRUCT // s will by of type MyStruct with values -4,3,2.0
}


>>106095286
too much shit in here, I won't bother correcting.
Just know that Odin has picked up a lot from Pascal and that's the reason for its non-oop nature.
>>106096384
yes and so is every other systems programming language and even scripting languages in that regard.
Manual memory management is important for game dev and Rust sucks in that department. It'll take more time to get shit done compared to other languages like C/C++ or Jai/Odin.
And once done, maintaining a Rust game will be a literal nightmare.

>>106096074
>>106095119
In Beef it's just
struct MyStruct :
this(int32 x = -4, uint32 y = 3, float z = 3);

static void Main()
{
MyStruct s1 = .();
MyStruct s2 = default;
}

>>106095119
you vvil write the constructors

>>106098967
This is okay too.
MyStruct s;

Hey guys it's me again
>>106026139

I finally finished the Hello Triangle chapter, and I solved the exercises!

I'm not sure if I can say I actually learned something. I mean, I know I did, but it all feels so foggy. It makes me feel insecure about moving on... But oh well. I guess tomorrow I will start the next chapter

>>106099557
Good job. It's very rewarding.

>>106099557
could you explain to me what these database entries mean?

https://rentry.co/baboony-loguments

>>106099844
>some random pedos vpn info
??? who cares ???

>>106099557
i feel the same way learning math, bro. we just need to persevere until the lightbulb flicks on and it all comes together

>/gedg/ try to use proven technologies challenge IMPOSSIBLE
Why can’t everybody here use C++ and Vulkan on Windows like competent devs

>>106100130
And what is wrong with using Linux?

>>106099557
>I mean, I know I did, but it all feels so foggy. It makes me feel insecure about moving on...
that is how it is supposed to feel. You are doing great.
Remember, consistency is da key.
>>106100130
>C++
That shit is the reason behind Java, Jai, Odin and a lot more usable, alternative solutions to it.
What's wrong with using them and maintaining sanity?
Using pure C++ is worse than rust.
Modern C++ is very complex and I'm sure (You) probably don't know most of it and just rely on C to get shit done.

>>106100201
technically nothing its just a sign of being a cringe nodev

>>106098556
>nothing's stopping you from declaring constant structs and using them to initialize your struct though
And what happens when I'm passing around types and can't pass around a constant struct? "zero initialization is always valid" doesn't fucking work, it's nice that it initializes to 0 by default instead of garbage memory, but a lot of things like scale vectors and quaternions don't work like that. The latter will in fact crash your program if it is initialized to zero.

Relearning what I programmed, made some slight changes to things. Will handle the auto-completion debacle later.

webm part 2
scale-rotation-translation test

Maybe a good thing to do tomorrow or tonight (if I have the time) is to try to create an update component for the node ecs system. That way I can create update logic in scheme.

I want these cubes to move around, rotate, etc.

>>106100201
It is easy to learn and free, but not common yet. Also for some people (mac users especially) things only have value if you spend money on them.

He'll change his mind when more people change their opinions on it, that's how these things go.

I almost figured out how to implement prefabs

>>106101348
>things like scale vectors and quaternions don't work like that
ah, you found another thing to pick on and cry about.
well, everyone knows things don't work like that but every language somehow tries to makes things work. This is where bloated math libraries come to the rescue for languages like C and C++.
Just because Odin has built-in array programming does not mean you get to trash talk it for not behaving exactly like a math library that you have in mind. Its a you problem, keep seeting.
Here is how Odin makes it convenient to deal with scalar parts
m : matrix[4,4]f32 // all elements are zero
m = f32(1) // makes the matrix an identity matrix

v : [3]f32 // is already an identity vector
v = f32(1) // makes the vector (1, 1, 1)

q : quaternion128
q = f32(3) // sets the scalar part to 3

>>106101453
noice progress!!

>>106075501
>rendering architecture, multiple renderer solutions and other topics?
https://wickedengine.net/2021/05/graphics-api-abstraction/
https://floooh.github.io/2017/07/29/sokol-gfx-tour.html

the sokol blog and source code is especially useful, it's a high quality library and very pleasant to use and learn from. wicked engine blog is also nice.
there are also blogs about graphics analysis of famous games
https://simoncoenen.com/blog/programming/graphics/DoomEternalStudy.html
here's the list
https://www.adriancourreges.com/blog/

bump

>>106102328
>ah, you found another thing to pick on and cry about.
Not being able to initialize a quaternion for rotation or a type you intend to you as scale is a pretty good thing to pick on.
>schizo rant about math libraries
You're just showing a bunch of sometimes-applicable solutions to a problem that shouldn't exist.
>have a struct called Tank
>Tank has a scale member which is a vector3
>it is 0 0 0 by default
>now I have to, somewhere in the code (likely wherever I create Tank), initialize scale to 1 1 1
>either I have a specific create_tank function, or I need to put a check in a more generic function to catch it
>all this could have been solved by setting its value in the struct, ensuring it will always be 1 1 1 without worrying about it ever again

Anyone here used unreal engine 5 with Linux and how have you found it?

>>106105456
It's functional. It crashes a lot but I think the same is true for Windows so not sure how specific that is to Linux however the Linux version's interface is markedly worse due to how laggy it can behave. It's overall just a worse experience than on Windows but I wouldn't consider it a deal breaker if that's the engine you want to use.

>>106105547
Cheers boss man

>>106105417
nothing stopping you from writing your own quaternion class and avoid the pointless seethe though.
You can spin up your own default proc for a quaternion that creates, initializes and returns a custom quaternion struct with the function pointers in your quaternion struct assigned appropriately for usage as methods or go back to C++ and chill.
For someone who got badly OOP brainwashed and finds initialization to be a huge roadblock, Odin probably is not for you. Beef might be more of your thing.

>>106100346
>that is how it is supposed to feel. You are doing great.
>>106099931
>>106099814
Thanks a lot anons. I will try to tackle next chapter today.

Loading collada files directly in my engine turned out to not be a good idea. It takes several seconds to parse, hanging the program in the meantime. Maybe because I abused regexes to parse the file.

So I definitely need to cook* the assets before using them.

*Is cooking the right word here?

>>106107827
>parsing with regexes
Please go learn something about parsing and never touch regexes again

https://www.youtube.com/watch?v=evhBepR92yw

>>106086668
Or you could, you know, plan and preload everything you need in memory and only do cleanup in loading screens/sections and gc will never really kick in but okay.

>>106108427
I asked ChatGPT how it would do it and it recommended me regexes, that's why I used it.

>>106102255
i also implemented prefabs (from json) but ended up scrapping them recently, its too big of a headache. i just make my prefabs in code/script. much easier imo.
>>106107827
yeah cooking is smart, you can also implement it transparently to the rest of the code by checking to see if you need to cook the asset when you try loading it, cook it, then save it to disk for when you load it the next time.

>>106108596
that only works if you have regular loading screens

>>106108637
>I asked ChatGPT
jesus christ fuck off

>>106108637
>I asked ChatGPT
I kneel...

>>106075373 (OP)
>check devs of pic related
>some female name
>great, another troon rust developer
>actually a female
>very cute
>incredibly cute name as well
How does this happen. Where were the engineering baddies back when I was in school?

>>106108862
>tfw will never have a cute gf to make a game in a custom engine with
its ogre, why even live

>>106075373 (OP)
SKG won’t pass. Let other devs know because this incessant fear among the community is sad and only fueled by entitled and uneducated consumers (gamers).

>>106078754
Is she mexican?

>>106108958
I only want it to pass to fuck over corpocucks slop factories

>>106109013
It won’t pass because of another post I made here >>106108908
Tldr; too much on the line for the saas/live service model for existing European tech companies that pay huge taxes in these countries. It’s not going to happen, period

>>106108749
Honestly, chat crappity knows more than you. nta but I would rather listen to an AI than your stupid ass.

>>106108862
>How does this happen.
AI.

https://youtu.be/PBMKUJ6pQbU

Delete the learn-c.org from the wiki. Fucking AI generated website, put the real https://www.gnu.org/software/gnu-c-manual/gnu-c-manual.html chad tutorial, ffs.

>>106110108
how are you even intelligent enough to use a computer?

>>106110187
I've been using troonix since 2004. You're just a fuckwad who acts holier than thou but honestly you probably know jack shit. Chat crapass AI likely *does* know more than you and you're butt mad that someone would rather take the advice of a mindless black box than from your stupid ass with a bunch of incorrect preconceived notions inflated by your ego.

>>106110217
AI only knows what it scrapes from github or online tutorials

>>106110222
I mean considering an online tutorial alone probably has more decent information than you then it must be doing pretty good.

>>106110162
Just add it to the wiki yourself ....

I'm about to finish my 2nd indie game in UE5 and god damn I'm tired of using it (very slow to iterate) and working in 3D. It's a much better idea to just go 2D and have more fun with your own engine, especially if you are working solo. I have literally become an environment artist as a result of this project.

>Go on /gedg/
>230 replies
>Near zero footage of actual projects being made
>Go to /v/ GameDev Thread
>130 replies
>At least a dozen screenshots or webms of actual progress
Thoughts?

>>106111475
this thread is shit now, it's just AI kids

>>106111475
I am with you, we need more progress posts. Sometimes engine progress doesn't translate to a webm though, its just code changes or architectural.

>>106110986
3d is always more about graphics and animation. This is tiresome for a single person and every little thing or chance takes ages to implement. I have a background in 3d (job) and would still despise making anything on my own.

>>106111581
>we need more progress posts
Yup. There's a perception that engine devs don't finish anything, posting footage helps newbies get into it and people mid-project to stay motivated. Cool webm btw, this does make the prospect of engine development to suddenly feel within grasp.

>>106111626
>There's a perception that engine devs don't finish anything
its true

>>106108583
The Legend.

Will one of you assholes write a game in J or APL or something for me? Thanks.
>just like dip in real quickly to see if there are any sharks in the water

>>106111814
Just have "AI" do it. Surely it can handle that simple task?

>>106110986
>environment artist
future technical artist?
>>106111475
what that you can understand how hard making a game engine is over making a game.
I've been here for a while and still don't have the balls to start my first game engine. I got all about the GUI shit(immediate and retained mode) figured out but a path tracer, OpenUSD and creating a good game object design are due on my to-do list.
Maybe I'll be able to start by the year end.

>>106111993
*with that

>>106111993
But this begs the question whether solo engine development is a programming exercise or an actual path towards game development.
Because of course you get more freedom but in a way game design was always about creating smoke and mirrors around limitations. In game design it's absolutely normal to change the game according to, let's say, hardware limitations.
What is the percent of projects that actually need freedoms that engines don't offer? That an engine can't be altered to offer? What percent would actually be inferior for the player if made in unreal or unity?

>>106112037
Making your own engine is a good idea for 2D games
For 3D games you will never finish anything in a reasonable amount of time

>>106077948
i thought some chick made that game or is it another case of a chick taking the credit and a guy doing the work

>>106112037
>or an actual path towards game development
Casey says and proves that it is in Handmade hero. If all the game devs have no idea on how the tools they use acutally work underneath their engine's API, then its a total disaster.
You don't need to create a game engine to make another flappy bird clone but creating a game engine helps when making games like terraria or noita
>What is the percent of projects that actually need freedoms that engines don't offer?
You'll never be able to figure it out if you are looking for percentages. Its completely irrelevant. You just get your GDD done first. Depending on how elaborative the initial GDD was, any and all changes will end up being minor.
You need to realize that all game engines are unique. Every engine has its own strength and weakness. You'll have a hard time figuring things out if you see both unreal and unity as just another game engine. If your game needs to target mobile platform and it doesn't need realist graphics then choosing unreal will be a very bad choice. You need to get the requirements for you game straight and choose an engine by comparing them all for the most compatibility with your GDD and your resources(dev's experience and team size). AAA studios will do things however they see fit so its better not to take references from their development pipelines for our solo route.
Anyone will have a hard time if they did a sloppy job with GDD and chose the wrong engine. You just figure out what the job is through the GDD and choose the right tool(an engine) to get the job done. In case you failed to find the right engine for the job or don't have the resources to pull it off with that engine then going all in on a custom engine totally makes sense. Percentages can't do shit about getting shit done here.
>>106112167
chick's legit.
me vouch.

>>106112175
>Casey says and proves that it is in Handmade hero
10 years and no game

>>106112175
>In case you failed to find the right engine for the job or don't have the resources to pull it off with that engine then going all in on a custom engine totally makes sense. Percentages can't do shit about getting shit done here.
I definitely agree with that. I talk about percentages because I think most projects probably do fit some existing general purpose engine, or that maybe you need to reasonably alter the GDD and still retain the "point" of your game.
So, if most projects can be made with an engine, you should ask yourself: does my project _really_ requires a custom engine? Or am I doing it because otherwise I'd be considered an inferior developer or because I want freedoms I'm not actually going to use or that aren't actually important? Is my point

>>106112195
>no game
and?
that should give you a general idea on how hard making a good game engine is and why /gedg/ is not as active with progress as /agdg/
Anyone can make a game with just a decent compute but not everyone can make a full fledged game engine. It take lot more than just a compute to pull it off.
>>106112240
>you should ask yourself
nope.
I'm too retarded to get any answer out of myself.
I'll just do whatever the fock I want and understand how shit works.
Me too tired to give a shit about being inferior or what others would've thunk.

>>106112262
Casey's a really bad example because he makes it look much harder than it really is
Some dinky little 2D game you could program the engine for in a few months but he wastes so much time and ultimately never finished it

>>106112262
Well that indeed sounds like an exercise in programming rather than game development, which is perfectly fine, but someone who has an idea for a game he wants to release should in my view ask himself these question.

>>106111492
>it's just AI kids
it's a bunch of "game devs" here trying to draw triangles on the screen in this thread, lmao. the stupid ai kids are making slop games nobody will play. get your facts straight triangle fag

>>106112272
>he wastes so much time and ultimately never finished it
true
and I won't bother finishing it now.
I've already gained a lot from what I followed through and I think that's good enough.
Keeping if it is finished or not aside, the content is indeed solid and I would recommend it.
>>106112281
>someone who has an idea for a game he wants to release should in my view ask himself these question
ah, I see. In that case, I'd say the answer here would totally depend on the questioner's work experience.
If they are an amateur, they might make the wrong judgement by assuming they will need a custom engine before trying any other engines but there is no way to know if the existing engines are useless unless they implement their game idea in any of them.
This is a question that is very hard to get right without a good technical experience of game engines and to get good knowledge of that, understanding how different game engines work and building your own engine is essential.
To make up for the lack of experience, its better to stick to making simple games and build it up(kinda what /agdg/ does) but ultimately, engine dev experience is essential for making accurate judgements and will help a lot for creating not-simple games with unique ideas.
>>106112298
>get your facts straight triangle fag
making triangles in /gedg/ is the norm saar. No need to explicitly mention it.
Even people who made several games will struggle(a lot) with making a triangle using a graphics API for the first time, saar.

Does raylib belong here?

>>106112587

>>106075373 (OP)
getting to the end of the actual engine work on this prototype puzzle game i'm working on, i was worried that the mechanic sucked and wouldn't work but i realized the solution was just giving the player multiple powerups to help keep the board clear and tidy, kind of like how arkanoid extends breakout by giving you shitloads of powerups.

part of me thinks it's still lazy game design and i'm gonna still be dissatisfied even when i implement all the ideas i have, but at least i'm at the stage of development where i can test new ideas without having to spend a day or two writing more pieces of the engine to make the idea work

>>106112587
i'm using raylib too, it just makes stuff so easy man

>>106112587
>>106112677
it can't 3d in 2025

>>106113288
https://www.raylib.com/examples/core/loader.html?name=core_world_screen

>implement optimization
>its slower than naive brute force

>>106075373 (OP)
wait is it really that easy to make games nowadays?
is this why games in the last 15 years are all uncreative slop shit?
what software is that?

>>106113347
How good is it for a 3d game? Like Beyond All Reasons, for example.

>>106114134
not good

>>106113979
>what software is that?
>>106077948

>>106114134
>How good is it for a 3d game?
its OpenGL with a generic API so for anything serious, its not a good choice.

>>106113288
>>106114134
It's fine for prototyping, testing, etc. I use it right now as a way to quickly get stuff on screen in 3D. However it's very high level and handles way too much for you, so I would not release a game with it. I fully intend to either switch over to Sokol or just do OpenGL directly eventually.

>>106114405
>Sokol
How's that? Looked at it briefly and it didn't seem much higher level than plain OpenGL which makes me wonder if it's even worth it unless you wanted to use Vulkan.

What do you lads use for 2d animations? I've been using photoshop with the video timeline and it's radioactive garbage, mostly because it's laggy as hell and clunky in some aspects. is CSP better?

>>106111993
It'd be best to make a game that implements a lot of different systems than an engine that tries to do everything.
When the game is done, rip out those systems and put it in the engine project, rinse and repeat.

Half of the time when someone is making a game, they have very specific needs:
>Oh, I need a mirror here that reflects the room and player
>Sorry X engine can't do it out of the box, you need to go buy X asset from the store or write your own

>>106114856
Do you mean frame-by-frame? I use Krita for that. You can import references, have multiple layers, the only thing is missing is scenes, but just use different files for that.
There's also Pencil2D and OpenToonz but that's too overcomplicated and the UI feels archaic.

>>106114909
>the only thing is missing is scenes
you mean whole canvases? yeah, I already do this in photoshop, that would be on a different layer. If Krita has layers than it should also be possible. But it doesn't matter as I don't do this even in photoshop, i just animate different scenes in different files.

you ever tried CSP? how does it compare to Krita? I see a lot of content, brushes and stuff for CSP on /ic/, haven't seen much stuff for Krita.

>>106114896
>It'd be best to make a game that implements a lot of different systems than an engine that tries to do everything.
that's not at all doable. Everyone would be doing that already if that is truly feasible. Go back to /agdg/
you can't make a game with lot of different systems without using an engine that actually supports lots of different system in the first place. You are grossly underestimating a game engine's architecture. The interoperability to transition one system of a 3D game engine into another is a hugely retarded task that is practically never practiced. Unreal's got its actor-model and Unity's got its DOTS. You can't make anything you create in one engine work on another without a full rewrite from scratch in that another engine.
You are better off relying on a game engine only if you are going to create a clone of it with its exact same architecture. The rework won't be worth it as it'd take a lot of time and effort to get things working right.

>>106075373 (OP)
Woke up, 4 hours of straight progress [spoiler]debugging shit[/spoiler]
Good start into the day

>>106078925
They are not off-thread, they are stop the world completely unpredictable.

>>106086668
>gc kicks in its causes a lag spike
what do retards mean when they say this? this only happens if you completely run out of memory on your computer. the gc is "kicking in" constantly

>>106091075
I imagine they're using some variant of wave function collapse. very impressive.

>>106115454
procedural meshes are simpler than they look

>>106115362
They mean that there's there's more garbage accumulated for whatever reason, when the garbage collector kicks in it can cause that frame to take too long.
Anyway, looks like we finally got a Jai leak, is anybody checking it out? I started reading the thousand pages of tutorials and documentation, it's pretty comfy.

>>106115166
debug ur tag usage

>>106115671
>He doesn't use spoilers on non spoiler boards
[spoiler]Coward[/spoiler]

>>106115661
>is anybody checking it out?
I am. Its meh.
overall, apart from the metaprogramming part, Odin aligns well with my programming philosophy so I really don't have a strong enough reason to shift.
Guess imma be sticking to Odin and chill.

>>106115454
lmao

>>106115661
Been trying it:
>the metaprogramming is amazing
>most of the syntax etc is a lot better than C++
>quite a few nice features
>as a whole doesn't feel as polished as Odin (unsurprisingly, considering their release status)
>but Odin has a bunch of limitations, some purely autistic and pedantic that make it unsuited for developing complex games (e.g. not a retro 2D platformer)
If you're making a simple game or a non-game project, go with Odin. If you're making a more complex game, the investment in learning Jai will absolutely pay off.

>>106116276
>simple game or a non-game project, go with Odin
>decides a programming language is bad based on its syntax and simple nature
lmao, here comes the Odin coper
Just because Odin isn't as complex as Jai doesn't mean Odin can't deal with complex shit.
Odin has JangaFX products to prove that you are retard so keep on coping.
Jai has some features that makes reading the code very ambiguous and hard to read. Odin OTOH with its limitations makes it easier to understand exactly what the code is doing. Not like it matters to a hardcore OOP fag. Just dump Jai and stick to Beef.

Guys, I made my entire game using floats. Now I'm paranoid that straying too far from the origin (by many orders of magnitude, which my game sort of needs) will cause strange float imprecision bugs. Should I be worried?

>>106116395
Why are you seething so hard over a self-declared "C-like" language being more verbose and tedious than languages with higher level features when it comes to complex projects like a bigger game?
>Just because Odin isn't as complex as Jai doesn't mean Odin can't deal with complex shit.
No shit, so can ASM, but there's a reason most games are made in C++ and not C: because you'll be writing A LOT of boilerplate in a language like C (or Odin) compared to C++ (or Jai).

Also what's the obsession with Beef? Is it because you can't eat it you fucking hindoo?

>>106115661
I played with it, the compiler is very fast, as advertised
the nicest part is all the documentation explaining the thought process behind the design of certain modules
but the language itself doesn't seem that great; certianly not a 10x improvement to want to deal with compiler bugs and all the awful shit that comes with new tech

>>106116456
>like C (or Odin) compared to C++ (or Jai)
wrong.
Everyone re-writes their vec3 class again and again in C++ but not in Odin. Array programming is built it. At least find something valid to justify your seethe lmao.
>Also what's the obsession with Beef?
It came from your retarded seethe over how Odin doesn't handle initialization like an OOP language though.
Is it because you can't eat it you fucking hindoo?
I eat beef though. In fact, I ate it last night.

>>106116276
It would be very stupid to publish a game using a pirated, unreleased, and not to neglect to mention unfinished, compiler.

>>106116395
>Odin isn't as complex as Jai
Oh cool a claim from someone on the Internet that I will have to believe due to having insufficient familiarity with both languages.

where's my close range isometric cell shaded medieval sim village to care for?

>>106116518
>Everyone re-writes their vec3 class again and again in C++ but not in Odin.
No, not everyone does that especially not "again and again", it's also an incredibly simple thing and not at all the kind of extra boilerplate that will pop up in a complex project I am referring to.
>It came from your retarded seethe over how Odin doesn't handle initialization like an OOP language though.
I had to search that through the thread. You are evidently emotionally attached to Odin, because why would you think that we're the same person? Because we both mentioned extra boilerplate/tedium? That's exactly what every low level language does, it's not some sort of personal opinion only held by a few.
>I eat beef though. In fact, I ate it last night.
Good for you, eat some more so you get your test up and stop acting like a hysteric woman.
>>106116532
>It would be very stupid to publish a game using a pirated, unreleased, and not to neglect to mention unfinished, compiler.
Obviously, it's just a hypothetical comparison. That said, if someone is thinking about writing his own engine in any language at this moment, by the time he's even close to something resembling a public release Jon will either have released some sort of Jai public release a long time prior or he'll be homeless/dead on the streets of San Francisco.

Honestly the main issue with Jai is it's not complete, compiler access legitimacy isn't a problem, even if you have something to show before a public release, you can probably come clean to Blow directly and show him your non-trivial real project and he'd happily get you a beta invite.

>>106116532
>due to having insufficient familiarity with both languages
one example I can give you is the "if" statement
In Odin, an if statement always expects a boolean evaluation.
But in Jai, an if statement "tests the truth value of whatever you put in". You can put whatever after an if statement in Jai, like a string, an int or a float and it'll return false if its a default value.
But in Odin, you'll have to give an evaluation expression that checks if a string is nil or an int is 0. This not only tells clearly what type the if statement is dealing with but also makes it easy to understand the code while reading it.
Some people will complain that this needs more typing and is very bad but it does come in handy on the long run.
>>106116624
>especially not "again and again",
everyone knows C++ devs write everything themselves. You can stop coping.
>why would you think that we're the same person?
why wouldn't I? both are coping for the exact same thing.
>stop acting like a hysteric woman
stop assuming I'm you saar

>>106116412
its only a problem at > 1km

>>106117123
>imagine making a small game world

>>106116412
Why did you not use larger floats if you knew you'd need long distances?

In any case switching our f32 for f64 isn't exactly hard. Stop whining and do it.

>>106117289
t. retard

changing a datatype doesn't solve precision and the exponential drop off. you either transform the world around the player at a shit resolution like some space and n64 games or you start using local chunk coords. going from float to double solves nothing.

>>106117607
You knew from the start the game would have large maps, if you didn't build it poorly it would be mostly about "just changing a datatype".

Surely you weren't going to have a game that was big enough to start having precision problems without implementing things with spatial dimensions in mind? Surely you already had quadtree/octree/hashgrid culling etc?

Why do people use Revshit when engines like the one in OP's run circles around it?

>>106117648
>everyone on 4chan is the same person
>even when they type different
>...
>sperges out in "muh datastructures" he's heard from gamedev grifters
take your "etc" and shove it up your ass, you incompetent mong

>>106113288
>it can't 3d in 2025
It can, you can't.

>>106083520
>>106079484
>The talk will include a section about all the gymnastics they had to do to avoid GC pauses.

The journey is difficult, take this.

>>106117289
>Why did you not use larger floats if you knew you'd need long distances?
Yeah, I have a bunch of ongoing projects using my engine by now. The one I'm tackling right now happens to need big worlds.

I think the best route of action at this point is to replace every single instance of float with double. I'd rather not complicate things needlessly by having doubles and floats coexist. Now I'm going to be paranoid about performance instead.

>>106117289
>In any case switching our f32 for f64 isn't exactly hard.
Yes it is? It's a massive pain in the ass to adjust everything in your shaders which have to use f32. Not to mention math libraries using f32