>>103928027
>>103927976
>>103928014
>>103928088
>>103928042
atan can be used like this

atan2(vec_length(vec_cross(a,b)), vec_dot(a,b))

Notice how this isn't very different than using acos. HOWEVER, acos will only give positive angles. Sometimes this doesn't matter, but I thought I should mention it. I think there are also numerical stability concerns.

acos(vec_dot(v1, v2) / (vec_length(v1) * vec_length(v2)))

You can also make acos more stable numerically by doing something like this

acos(vec_dot(v1 / vec_length(v1), v2  / vec_length(v2) ))

= acos(vec_dot(vec_normalize(v1), vec_normalize(v2)) 

atan2 is for low-brow animals

>>103929607
It is also worth noting that many many implementations that I've seen take the acos approach. For example, I doubt I've ever seen a SLERP algorithm that used atan2 over acos.

People seem to vastly prefer vec_dot(v1_unit, v2_unit) = cos(theta) in everything I see.

>>103929640
you're just mad because you've never heard of it

>>103929607
>vec_length used at all
Too slow

>>103929641
>>103929607
atan2 giving bearing angles are defined from the positive x-axis, going anti-clockwise is "positive" (like in most maths) and conversely going clockwise is negative.
however you do not need to limit yourself to such a convention.
For example you could define your angle operator as angle(a, b) = -angle(b, a)

>>103929736
What do you suggest then. You can't normalize a vector without finding its length.

>>103929739
True, its not like atan2 is limiting or anything if you understand how it works. But it does seem widely unnecessary in most scenarios. If you don't care about the orientation of the angle (bearing angle) then there is no point over acos I think. Keep in mind with the atan2 method you are doing an extra operation with the cross product. And I'm assuming (without profiling or any proof so honestly I might be dead wrong here) that in most situations atan2 and acos are comparable in the time it takes the computer to run within an order of magnitude.

>>103929794
No need to normalize anything
>>103928318

>>103929804
CAn you show me this method? I'm unfamiliar with it and would like to see it.

>>103929816
Now let's see Paul Allen's acos disassembly.

>>103929816
the ABSOLUTE state of atan2 niggers

What are good resources for learning game development in Java, like the first tutorial that I see on yt uses passive rendering and I don’t like the magic of that. The idea is to learn from scratch

>>103929826

    inline constexpr TYPE getAngle() const {
        return std::atan2(y, x) * (180 / M_PI) + 90;
    }

    constexpr BaseVector2D<TYPE> rotateNegative(const BaseVector2D<TYPE>& v) const {
        TYPE cosValue = -v.y;
        TYPE sinValue = -v.x;
        return BaseVector2D<TYPE>(cosValue * x - sinValue * y, cosValue * y + sinValue * x);
    }

    constexpr TYPE angDiffFromVector(const BaseVector2D<TYPE>& v) const {
        return rotateNegative(v).getAngle();
    }

Not sure I remember how to use code tags.
Like this as seen in angDiffFromVector. The convention is that 0 angle vector is (0, -1).

progress report
finished cpu set pinning

next up is going to be reducing present wait times

>>103929919
I now understand now what you mean. What you are posting makes sense to me now. I was thinking of something weirdly different.

>>103929867
Depends what you mean by "from scratch".
The most popular game framework for Java is LibGDX. It's pretty simple to use.

>>103929981
Like without libraries plain old Java apis like awt and swing.

>>103930211
Are you looking for video or text tutorials?
These seem to use only use awt and swing:
https://youtu.be/6_N8QZ47toY?feature=shared
https://youtu.be/om59cwR7psI?feature=shared

>>103929919
the fuck

>>103930211
A roguelike with your own version of AsciiPanel is brain dead easy then.

>>103930211
I can share with you a couple of classes you can use to start making a game in java.

>>103929607
>atan2(vec_length(vec_cross(a,b)), vec_dot(a,b)
isn't that more expensive than simply computing arccos of the product?

>>103931307
This should get you going on https://pastebin.com/VcK87dXj study the code and extend it.

>>103931453
I just noticed that it's actually the correct solution but you need to remove the vec_length.

This seems to work.

    const glm::vec3 direction = end - start;
    const glm::vec3 normal = glm::normalize(direction);
    const glm::vec3 reference(1.0f, 0.0f, 0.0f);
    const glm::vec3 axis = glm::normalize(glm::cross(reference, normal));
    const float angle = glm::acos(glm::dot(reference, normal));

>>103931728
>The two-dimensional case is the only non-trivial (i.e. not one-dimensional) case where the rotation matrices group is commutative, so that it does not matter in which order multiple rotations are performed. An alternative convention uses rotating axes,[1] and the above matrices also represent a rotation of the axes clockwise through an angle θ.

i didn't even know this
that's grim
so in 3D the order you apply rotations matter

Are you guys stupid? dot product encodes sin and cross product encodes cos. Stop trying to convert to angles just to plug them back into a trig function
https://iquilezles.org/articles/noacos/

>>103931897
the problem was literally determining the angle though, not the sine or the cosine of the angle

>trigonometry autism and actual programming
I'll take these posts over the usual shitposts, although it's more useful for old/retro machines, there's a GDC talk where a game trig functions were fast on PCs, but actually ended up very slow on consoles, they didn't bother changing it.

I love refactoring.

Added a little tile variation.

>>103931728
>>103931864
>so in 3D the order you apply rotations matter
Yep. Fortunately, rotations in 3D still follow many symmetries. In the matrix representation, for example, they are still orthogonal matrices. In fact all orthogonal matrices represent rotations or reflections depending on the sign of the matrix determinant. When the determinant is +1 you have a rotation (special orthogonal group) and when its -1 you have a reflection.

>>103932071
I am surprised the discussion is still going on. Taking the arcos of the dot product of two unit vectors is a pretty fundamental idea in graphics and computational geometry.

>>103931728
You don't need to calculate the cross-product if you are doing acos anon, axis is unnecessary (unless you are doing something else with axis later in the code).

>>103932099
My two cents
-I think it has too much variation.
-Add variation in hue not light (more yellow or orange).
-It would help if there was not just variation between neighbor tiles but neighbor zones (ill defined amount of tile groups).

Can your homegrown engine output graphics that are at least on par with the graphics that you see in Chilla's Art games?

>>103932276
Not really. Are these translated by the way?

>>103932349
I was thinking about their older games to be fair, they had worse graphics
>are these translated
I think so

>microshart compiler makes atan2 3 times slower for floats
I always used it for doubles so I never noticed.

>>103932276
dont think many engines can beat ue5 asset flips in graphics fidelity

Give me ONE reason why you don't use Unreal. You are in dev hell otherwise

>>103932611
Last time I used it there was little to no documentation for UE C++ APIs

>>103932193
Thank you. For now, I am just constructing my systems, not really creating a final product. What I am working on is exactly what you are suggesting.
The land will be divided into specific 'regions', each with unique landscaping rules and tile coloring, and there will be more color-mixing options.
And translucent and weaving water.

>>103931981
Yea and what are you going to use the angle for? Because there's essentially no use for an angle in game programming, you're just going to turn it into a vector again

>>103929599
pozzed or underhandedly based?

Fuck it, I'm just installing Unity and making my game there

>>103933158
You can literally use "it".

>>103933900
how do you translate "it" to languages that only have gendered pronouns?

>>103930543
>>103931307
Many tanks bros, will study and share here what I’ve accomplished

>>103933900
People are not things.

>>103933927
>>103934225
IT sounds like a YOU problem

sucks to be able to speak only 0.7 languages burgerkeks

someone posted a link to an article about model outlines a few threads back, can someone post it again?
i cant find it

>>103933900
you literally can't.

>>103935019
Dunno if it's the one you're thinking of, but I saw this on the orange website recently
https://bgolus.medium.com/the-quest-for-very-wide-outlines-ba82ed442cd9

>>103935428
*it* said

>>103932643
Just read the source code.

>>>103880409
PROLOG GAMING PROGRESS UPDATE: I learned how to expose C apis to SWI-Prolog, now I am writing enough bindings so the Raylib hello world example can be built in prolog.

>>103936097
SOVL

I was watching this level design video: https://youtu.be/urTYEBBQbY4 and it got me curious: How would someone without the unreal (unity probably has something similar too) tools, let's say someone developing their own 3D engine, achieve something like that? Would they need to develop all the similar tooling from scratch (well, based of the engine they're making I suppose) or can they mostly skirt around that by using other tools? Like making an entire level on blender or something? I'm genuinely curious, sorry if I sound retarded.

>>103936274
I think this video gives you a good idea of what it's like to design levels without a prebuilt engine. He's not designing levels but the workflow is very similar https://www.youtube.com/watch?v=rbu7Zu5X1zI
You'll see basically he edits code and then he has a viewer to see what the output looks like, so he can modify code and see its effect quickly by refreshing

>>103936274
Tradtionally many gamedevs just made their levels in their 3D modelling program, but in my experience the workflow feels pretty terrible
There aren't really any good generic 3D level editors out there, I made my own for my engine and just making something where you can place, rotate and scale 3D objects isn't that difficult

>>103936097
looks better than real sonic

>>103936326
>>103936332
Thank you frens, I understand it now.

>>103936097
Reminds me of the good old days.

What do you guys think of this code?


https://pastebin.com/rq8TSTSd

- status bar
- something to help visualize the players cardinal direction
- preview on build mode object

>>103939474
That's kind of cute. What are you cooking there, doc?

>>103933158
based. A lot of similar books use 'she' for these exact cases.

>>103940524
ygg engine does everything, mostly

What game design books can you guys actually personally recommend?

>>103940693
the only books I recommend are game programming patterns and game engine architecture

Day 3

Rendering gun's model onto screen now. Right now it's just a cube but when I install blender i'll make some guns, enemies and bullet particle

Changed lighting and UI a bit and fixed some more framerate issues.

>>103921343
When stamina runs out (below 20) movement speed is slowed and you can't jump

>>103940711
>game programming patterns
this is what I was going to recommend too

So I have a transform struct with position, rotation and scale members. I noticed that I calculate the model matrix even when the transform hasn't changed which feels wasteful. A solution would be to turn the struct into a class, have a dirty flag member and control the the operations, but I don't know how to feel about this. Any suggestions?

>>103940992
kek

>>103941277
That's a standard way to do it

C++

GLuint buffer_index;

auto result = buffer_indices.try_emplace(texture_name, buffer_index);
if (!result.second)
    return false;

glCreateTextures(GL_TEXTURE_2D, 1, &(*result.first).second); // WTF

>>103940992
have Sec(x)

>>103941277
generally no, bool flags are considered one of the worst architecture design anti-patterns for numerous reasons, they are bad for your cache, they are bad for complexity, every additional flag is exponentially more state complexity, other reasons you can google it yourself

>>103932611
If I've got to live with the tech I've got, I'd rather live with tech I created and know how best to use it over someone else's giant fragile clusterfuck.

>>103941570
many game engines do this you idiot, it's faster than calculating it every time

>>103940992
>no atan2
have we learned nothing?

>>103941509
lemme fix that for you

auto [it, result] = buffer_indices.try_emplace(texture_name, buffer_index);
if(!result)
    return false;

glCreateTextures(GL_TEXTURE_2D, 1, &it->second);

>>103941624
das it mane

>>103941634
im glad i could help, anon

>>103941591
im not saying to calculate it every frame, i'm saying that adding a bool flag to a class is considered bad practice in general i didn't just pull it out of my ass look it up yourself

>>103941685
It's good practice because it means you don't have to calculate it every frame because you know when it's been modified or not

if I wanted to make something that runs the same on every machine (ie no floating point compiler differences) wat programming language or framework makes sense to use. i also desire internet connectivity / server networking etc. THANK YOU

>>103941685
no it isn't shut the fuck up if you don't know what you are talking about

>>103941794
Richard Russell's BBCSDL

Working on a graphics framework for DirectX. Not a full game engine, it's just meant to reduce the verbosity of managing DirectX objects and initializing things. Each DirectX object (shaders, cbuffers, textures/views, etc.) each have a class whose constructors handle most of the initialization. Each class has getters to the underlying DX objects so the dev can drop out of the framework and handle things manually whenever they want. It does have some helpers for games, like Camera classes and an application runner which calculates the time delta in the update loop for you.

It avoids allocations when possible, but sometimes it has to allocate (allocating a buffer to read the shader bytecode into, for example). In such classes, the dev can provide an STL allocator so they can control how memory is allocated for all allocations in the library. When no allocator is provided, it just uses std::allocator.

Picrel is the code for a rotating triangle a hundred lines or so of C++, minus the shaders themselves (which are dead simple)

>>103932611
>3D
lol

>>103936097
Holy Based.

>>103939474
>- something to help visualize the players cardinal direction
What a happy coincidence.

>>103942227
nice to see a fellow dx enjoyer

Should I post here if I'm making a game by piecing systems together? I'll probably reuse code in the future but I'm not making what most people would call an engine

>yonger cousin was always hopelessly addicted to videogames since childhood
>would pee his pants instead of getting up to piss, played like 12 hours a day
>somehow convinced uncle to let him study a 4 year game dev course at a community college a few towns over
>also gets uncle to buy him a 2.5k gaming laptop
>ff to now, 5th or 6th semester
>can't draw 2D, 3D skills are absolutely garbage
>can't program to save his life
>still spends 14 hours a day playing videogames but away from home so his parents can't bug him
>only plays AAA games, never played anything that's not AAA or non FPS (still likes train sim games like any good asian boy)
>never heard of any indie game ever, never heard of any game from before 2010
>never heard of mass effect, stardew valley, prototype, final fantasy, dragon quest, monster hunter
Everyone in the family is worried. We thought he was doing well at univ until we saw his projects that he made with 2 other students it's... Bad

>>103942949

Does he at least have a portfolio, CV, and a résumé? And also, I am assuming there's still saving him from that stint?

>>103942949
the only thing that can save him are bad consequences without any support from others

>>103942977
No, he hasn't done any game dev outside of coursework in the 3 years. He hasn't ever made a single sprite or 3D model outside of coursework i think
He made a 2D platformer with static sprites (non animated i mean) in 4th sem, and recently he made a 3D demo which i highly suspect was just a pre installed lighting example in unreal
He doesn't have a resume or portfolio

guess it's time i start doing some text rendering from scratch again

>>103941685
>adding a bool flag to a class is considered bad practice in general
Gonna need a source on that one. I suspect you've misunderstood whatever advice you read.

>look it up yourself
All I could find was this one random blog post:
https://www.sebastiansylvan.com/post/2014-04-27-flags-are-a-code-smell/
There are tons of articles saying you shouldn't have boolean flags as function arguments, but that's a separate issue.

Main points I see in this article are
>we have significantly more possible states than legal states
Well yeah, nearly all classes are going to have some kind of internal invariant on their fields. Document it and check it where relevant. This is what asserts are for.
>it’s very easy to accidentally end up in an illegal state by forgetting to check a flag in one of the many branches
This is maybe partly legit but also partly a skill issue. Nothing will save you from forgetting to check the precondition at the start of the method. If many methods share the same tricky precondition, then it might be worth factoring out that check.
>As you add more flags the set of possible states explodes
Technically true, but not relevant. Most of the flags will be orthogonal. Even full MCDC test coverage won't require O(2^N) tests unless the code is really pathological.

Even if all of the above points were valid, they don't apply in this case. The class in question here is going to have only one flag, which is fully encapsulated and can't be set or checked from outside the class. In fact there is only one method that checks the flag (the method to get the transformation matrix), and there will be absolutely no change in its observable input/output behavior when the flag is introduced (only a change in performance).

>>103941570
>they are bad for your cache
Anon, we are talking about storing an entire fucking matrix, 24 bytes at the bare minimum, maybe up to 128 depending on what anon is doing. 1 byte for the flag isn't going to make a difference.

>his classes have zero padding
surely not

>>103940611
Is this a gamemaker like?

>>103943774
It shares a lot of the same objectives. The idea is that a given player (client) can login to public or privately hosted servers and they will get unique experiences created by others. Movement type, combat, dialog, quests, minigames, etc are all mutable for the server admin. I will provide some default setups I hand craft for those who want a sort of vanilla experience.

>>103929816
yeah you can run this 50 times while waiting on a single cache line to read from RAM.

Fixed a lot of bugs related to AI. Initially I wanted to add knockdowns but I was restrained by the bugs. Stalker source code gave me many ideas that sits well with stealth games. For example, rather that directly accessing the world, npcs have a memory with the data of enemies and noises around them. When they see something this is just accessing a memory which may be obsolete by now. For noises this is specially useful, running may generate noise but they don't give npcs access to where an enemy is at the present moment.
Tomorrow I'll add knockdowns and possibly guns. Then I want to add graphics

>>103944919
The visuals shouldn't look more complex than this.

>>103944930
excellent crappy soulful dragon garbage quest graphics

>>103939474
>modern-day zzt
Nice!

>>103929599
QUESTION FOR EVERYONE:
First of all, I like progress here, just please for the love of God, make your game so it runs on Linux.
> Question:
I'm interested in how VCMI, OpenMW and similar projects are legal in US and EU, not counting ioquake3 and darkplaces, since Quake was released under GPL license some time after it was published (year later I think).
Sure, they don't distribute art for VCMI (Heroes of Might & Magic 3 engine) like sounds, music, sprites, animations, DEF files etc. But what's Ubisoft stance on that? Can they prevent it?
Ubishit is a terrible company, I was wondering if there are some legal restrictions why they can't sue VCMI team. I browsed, but I couldn't find the real answer to this, aka 'can Ubisoft sue them'?

>>103945354
good god what a cringe post fuck off

>>103945441
(You)
Captcha: YOU

>>103945354
As long as you are not copying the original source code line by line then it's perfectly legal.
Ubisoft likely doesn't care since you still need to purchase the actual game from them to play it.

im gonna lose my mind trying to setup freetype compiling with premake

>>103945573
nevermind, i got it working
guess i just needed to complain about it

>>103929599
This is probably the most retarded question asked on here, but I'd love some insights. I wanna make a game engine but I only know lua, I'd like to learn either C or CPP. Yes, down the line I have use cases for both (ie: cpp for employment and C for tackling on making a simple dos-like operating system.) So I was wondering what way would be more beneficial, C with SDL2 or CPP with openGL

>>103945682
Just learn C, then learn C++

>>103945682
>or CPP with openGL
You will need some way to handle input, window, audio, fonts, etc. So you need SDL2 or similar anyway.

>>103945826
bro your native window implementation?

>>103945506
Thanks Anon.
>>103945682
If you use C language with Raylib, you'll learn a lot about C in general, and you'll have something finished at the end of the day.
Raylib is like saner little brother of SDL2, when you learn C, it won't make C++ easier, I think it'll make it harder, but oh well...

Is there some kind of userland equivalent of CPU interrupts? I want to ask the OS to wait for some number of milliseconds and then have the OS call some function I have in the engine with the thread's context

>>103945922
Is this your first day of programming? Heard of ChatGPT? We used to call it "Google."

https://www.youtube.com/watch?v=anGdYJu_eH4

>>103946064
But I want to hear it from a human being, not a search engine.

>>103946064
What makes you think I didn't search for it before, I didn't find anything. Are you saying that it doesn't exist after all?

>>103946070
I'm reading this thread, and the dev claims (this used to be 5 years ago, so maybe things changed) that with 10k it only needed 200MB, and this was made in java. Impressive https://old.reddit.com/r/songsofsyx/comments/dirkmw/crowd_control/

>>103946064
i like you

Anyone here working with a custom DSL for their engine?
I'm making an RPG and working on abillities. Stuff like elements and status effects is easy to implement with some normalisation, but I'd like a way to easily implement a handful of one-off abillities.

>>103946266
Are you an asshole nodev too?

>>103946348
>DSL
no sorry dial up modems is not interesting

>>103946405
im an asshole, but im not a nodev

>>103946096
Read Chapter 4 of Game Engine Architecture. If this is indicative of your usual ability to search, you might want to search for resources on how to search better. Because googling "programming wait milliseconds" instantly informs you of the existence of a syscall called "sleep" and from there you should be able to get to linux API docs for nanosleep or a stackoverflow post where someone has already explained the implementation of that system call far better than an original response you will receive asking here.

>>103929599
What the fuck kind of Waldorf School name is "Tynan?"

I need to fix my UI aesthetics, I don't like any of them at all but I can't hold up progress because of it either

>>103945922
>Is there some kind of userland equivalent of CPU interrupts?
Signals, on unix

>I want to ask the OS to wait for some number of milliseconds and then have the OS call some function I have in the engine with the thread's context
Don't use signals for that. You're extremely limited in what you can do in a signal handler context. For example, it's not safe to allocate memory (might deadlock) if you're running inside a signal handler. About the only thing you can really do is set a flag of some kind for your main loop to check next time it runs.

Instead, you should set a timeout when you query the OS for input events in your main loop. E.g. use SDL_WaitEventTimeout instead of SDL_WaitEvent. When you have a timer pending, set the timeout such that it will wake up at the point when you want the timer to go off, if no events arrive by then.

>>103946348
I think most people just use Lua for that kind of thing. Might be fun to build your own simple scripting language, though. Helps if you've taken a compilers class or otherwise know how to build a decent parser and AST, and you can write either an interpreter or a compiler to C++ or whatever you've used to implement the engine.

Another option might be to build a little Lisp/Racket library you can use to generate C++ code for these abilities, and hook that up to your build system so it has no run-time overhead.

>>103942747
Yes

>>103946581
You didn't get it, I need the OS to interrupt the normal execution flow of my thread and put the thread execution into some other function. That is, I want the OS to interrupt the execution of some function X in some thread A on a timeout and move cpu's instruction pointer to the start of some other function Y in that thread A
>>103946711
I wasn't planning on using that for input events, you're right though

>>103946656
I've been wondering about this for a while myself.
Actually, I've been wondering about creating UI elements using extremely high poly models in blender with simple single-color materials and then just taking a screenshot of them in appropriate lighting

>>103942949
Damn. Reminds me of 2006-2007, I was going to NYU and 2 of my roommates played World of Warcraft every waking hour (not with each other, it was just a coincidence they were in the same dorm room). One of the guys broke up with his gf because she was taking too much of his time from WoW. The other failed all his classes and had to retake them. Not a cheap school to do that at either, guy was probably out 50k in 2007 dollars over that.

>>103943037
That’s fine with me, one less competitor

>>103944930
Zelda II vibes

What program should I use if I want to make a higurashi quality like visual novel?

>>103945354
My game ran on Linux but after reading this I put an explicit check in the code to close the program if it detects the user is on Linux

>>103946064
Real ones used to call it altavista

>>103946405
>being a nodev
Go back to /v/

>>103946408
DSL is not dial-up

>>103946348
Well I am working on Prolog bindings for Raylib. With prolog you can make your own DSLs and macros

>>103946963
i dialed your mother yesterday
dick sex location

>>103946819
!remindme 1 day

>>103946986
Damn I walked into that one

>>103945682
>lua
https://love2d.org/

>>103946997
hello rEddi`t

>>103946348
why not lua?

Any recommendations on books on code organization, software engineering or programming patterns in procedural (or more generally anything that is not OOP) languages?
For example I know C pretty ok and feel comfortable reading it at least, but I have no fucking clue how to organize a midsize to large codebase in something like C, Odin, or Zig well. When it comes to creating codebases of nontrivial size all I know is classes and shit. Work and university has poisened my mind.
I want out of OOP.

>>103946919
Thank you for your contribution to humanity, your decision shall be remembered and respected.
>>103947056
Simply either write smaller project, or keep everything local (no global variables and unity builds).

>>103945354
>make your game so it runs on Linux
My game will ONLY have a linux build.

>>103947056
Try programming in prolog, it will change how you think https://youtu.be/G_eYTctGZw8?si=ItHIAM8J7V2XLM0g

>>103947056

Functional core, imperative shell.

Organize things by lifetime, write everything twice and only abstract the third time around.

>>103947229
Based, I'm slightly optimizing loading time.
It needs 1 second for parsing 1097 game scripts and loading 1793 game PNG images (which are merged into 1 texture on the GPU), with GLFW and OpenGL/AL initialization and stuff...
Maybe it's not that bad, because when I pass '-Ofast' flag, it does everything in 0.641276s, but it needs 3 fucking seconds to compile, which is kinda bad to wait on... Screenshot is '-O1'.

>>103947322
>It needs 1 second for parsing 1097 game scripts and loading 1793 game PNG images (which are merged into 1 texture on the GPU)
How much of that is the scripts vs the images? Basically I'm wondering how much you would save if you pre-built the texture atlas at compile time instead of assembling it at startup. (Or if you cached it on first run, which IIRC is how Factorio does it)

>>103945354
>make your game so it runs on Linux
I'm also making the game on a fucking X220 that can't even run Blender.

>>103945354
>>103947229
>>103947419
Also make sure your builds act as malware when you compile for Windows. It's the only way they are going to learn ...

>>103947263
the fuck is this used for

>>103947532
Professional Games Programming

>>103947548
oh I see so silky smooth 24 fps ghosting smeared TAA running at ai upscaled 2k taking 250 Gigs of disk space and always online so your data is continually stolen?

font atlas generation is working
my pants are full of jizz

>>103947570
anon...

>>103947570
forgot to mention that those 2 things arent related

>>103947565
No it’s programming in the language of pure first order logic, no typing bullshit, no OOP, just pure reasoning

>>103947548
post proof

>>103947413
> Or if you cached it on first run, which IIRC is how Factorio does it.
This is actually a very good idea, merged sprite sheet is around 17 MiB, 4096x4096 texture.
Loading PNG images takes 80-90% of time of script * image, so that is likely solution, "caching".
It does create problems with dynamically loading game data tho, which is what I want to have...
Adding completely new item to the game is just placing 1 PNG file in folder and writing:

(my_sword)
name   = "My Ultra-Mega Sword"
slot   = main_hand
frames = 1 # animation frames per second
rarity = 3 # how rare is the item
effect = (increase_offence_by_3, increment_conjuration, increment_knowledge)

>>103947419
>>103947485
Kings. (:

>>103948097
>Loading PNG images takes 80-90% of time
dont tell me you're making the texture every time you run the game

If you care about loading times don't load PNGs, use a raw format

>>103948128
> Don't tell me you're making the texture every time you run the game.
Depends how exactly you mean.
I load ~1800 small PNG images when the game is ran and merge them into 1 texture.
Why? Because I can easily extend the game content that way, or replace artwork...
It would allow for easier "modding" of the game, which doesn't exist yet but whatever...
>>103948212
I also care about the size, I fit almost entire HoM&M3 artwork into 17.8 MiB.
Some animation frames are missing, maybe an item or two, but I really care about size.
Including sounds for the game, I want to make it below 30 MiB in total, for everything.
Why? I don't know, it's kinda fun to fit 1 GiB game into 30 MiB, I'm not a real programmer.

>>103948893
>I load ~1800 small PNG images when the game is ran and merge them into 1 texture.
that's a waste of resources, make a system that detects changes in the directory of the PNGs and regenerates the texture when it needs to
then you can save the texture as one big png and load it directly on startup when no changes are detected
and voila, a 90% decrease in loading times

>>103948923
I'd have to use fstat-like functions (which invoke system calls all over the place, I dealt with them in x86_64 assembly), and to keep a record of that data in a text file...
For a reference, image related, with '-Ofast' it loads everything in 0.665 seconds approximately, 1097 scripts and 1793 images. And all that content fits in 17.8 MiB (compressed ofcourse!).
--
I'll keep in mind what you (and other Anon, if not the same guy) said, because someone on slower PC might run this game, and see lazy bullshit I wrote.
> make a system that detects changes in the directory
Note physically taken in my notebook, thanks, wouldn't think of it myself after struct horrors of FASM I had to deal with for using 'fstat', in C it's fine.

>>103948982
every game should have a two phase resource load system, first phase performs pre-processing on resources (combines all your PNGs into one), second phase loads them for use in the game
first phase only needs to be run when resources change

>>103948097
>It does create problems with dynamically loading game data tho, which is what I want to have...
Yeah, IIRC that's one of the reasons the Factorio devs went with the dynamic caching approach, instead of pre-generating the big texture and shipping it with the game.

>>103948982
>I'd have to use fstat-like functions (which invoke system calls all over the place, I dealt with them in x86_64 assembly)
Wat? Are you actually writing your game entirely in assembly? You mentioned -O1 vs -Ofast which I assumed were C compiler flags.

Anyway, idk how glibc does it, but in asm `stat` should only be a single system call

>and to keep a record of that data in a text file...
Just hash together the mtimes of everything in the image directory and the mtime of the directory itself. Then store the hash in a file, or even just try to load `cache-<hash value>.png` and see if it exists or not.

Modifying one of the files will change that file's mtime, and adding, removing, or renaming any files will change the mtime of the directory. You don't even need to sort the readdir results by name - the ordering isn't specified, but it's usually based on some on-disk structure and will therefore be stable across runs. (And if it's not stable, no big deal, you just end up regenerating the cache an extra time when it's not really necessary.)

>>103949106
> Are you actually writing your game entirely in assembly?
No no man, I wrote other projects in assembly, this game is 100% C, not including INI-like files for game entity data (script that I mentioned being parsed).
>>103949013
I'm also noting this, I thought about it, but I want to finish more important gameplay things first, like actually commenting out written battle system and checking how well it works.
--
Again, thanks a lot for advices, I appreciate it because I'm spending more time on other "sub-systems". Good night Anon.
PS: I'd write a game in assembly if IA32e had better mnemonics.

opinion about godot? I'm starting to program and I'm using Godot because is easy, the entire internet says is a good engine but I want to know 4chan opinion

>>103950082
It's an underbaked engine, Unity is better

>>103950082
It’s a great engine. If I wasn’t making my own engine then I would use godot. The best part is that it is very fast and GDScript is one of the best languages

>>103950365
>The best part is that it is very fast and GDScript is one of the best languages
You blew your cover

>>103950082
It's pretty capable

If you write your game in say C++ with SDL or something like that, would you say that you "wrote a game engine"?
There is a difference between writing a game without a larger engine like godot, unity, unreal, etc. and writing a game engine which you can then use to make a variety of games. What would you call it?

>>103950454
the term "game engine" has become pretty muddy

>>103950082
Godot is very comfy for solodev, and if you're new to programming GDScript is similar to Python - meaning it is one of the easier languages to pick up on.

Unity is good and has a billion resources out there for you to find any solution to a particular issue you're having.

Unreal you can guarantee will be heavily supported into the future and has some of the best engine tech in the industry. But you being new to programming (and assuming dev) you will not be utilizing anything that Unreal truly offers.

Try them all out.
3D?
UE5 > Unity > Godot
2D?
Godot > Unity > UE5 (don't do this in UE5)
Don't want to ever think about licenses and royalties?
Godot >>>>>> UE5 > Unity

>>103950454
anon, SDL is an OS abstraction, a framework, it doesn't handle animations by itself, nor map files, nor GUIs, nor cutscenes, or anything high level that's normally found in an engine.
Even Raylib is more advanced than SDL because it comes with a GLTF file loader

SDL is like Love2D, Pygame, Allegro and other "framework", it's no different than using a sound library or an image loading library, you only get barebone primitives, it's up to you make something out of them, and nothing's stopping you from going lower for certain things.

>>103950416
GDScript is 10x better than C#. GDScript’s only flaw is that it isn’t Python

>>103950082
It's not hyper optimized, but it's good for what it is.

>>103950717
GDScript is slow, Python is also slow

>>103929599
Is that book any good?

Also, I'm going to try running Anarchy Arcade's Hammer in Linux. It's not going to work, is it?

>>103933158
>>103933900
"He" has been used this way since forever. It shouldn't need to be explained away like this.

>>103951538
It's time that "we" fought back.

>>103950540
That's a great answer to a question nobody asked.

>>103942747
Anon isn't that what engine dev is all about?

>>103952426
engine devs can't into making games.

>>103947056
Look at projects on GitHub to see how they structure their projects and read the code.

>>103950454
A game engine is a collection of reusable code that can be used to make video games.

>>103952488
that's a framework

>>103947056
Who is she?

>>103952556
Answering myself. Korean fighting game Battle Queen.

>>103952509
No, a framework by itself is not enough to make video games. An engine is more than that. It should provide ALL the code that you need to make a game. And you can modify it further when you need to, but fundamentally if you have an engine all you need on top of that is content and possibly scripting.

>>103952582
>It should provide ALL the code that you need to make a game.
That doesn't make any sense

>>103952582
go back to the vg general already

>>103952595
>>103952599
Look at the Dark Engine. Do you think LGS had to add any extra shit when making Thief 2? No, the code was all in place, it was just a matter of content and little scripts. When you have a tailored in-house engine the engine should be doing everything you need already.

>>103952632
Ok now look at Unity

>>103952645
I don't give a fuck what slop devs are doing.

>>103952766
You're retarded

>>103947056
what's the issue exactly? Just use structs instead of classes and use inheritance

>>103952773
Slop dev identified, kill yourself.

>>103952509
an engine are the systems that your game runs on
if you make tetris, you already made an engine

>>103952961
>>103952974
An engine is a reusable code base
It differs from a framework in that it provides an architecture for your game, but you still have to write code ontop of it to make the actual game

>>103946954
Yes, nodevs back to /v/.

What do you guys think of better enums?

>>103950791
Skill issue

>>103953408
Language issue

>>103953408
>>103953409
issue issue

>>103950791
All relies on optimization. You can write code in C all you want but if your code is shit then you might as well be writing in python

>>103953486
Not really, shit C code still runs much better than good Python code

Yesterday I was thinking ways to implement hierarchical pathfinding. The simplest implementation would be to store the bottom and right side of each sector, in two bits, whether the border is open or not, which can be detected by walking from corner to corner until you find a non obstacle tile. The approach is both fast and lightweight but the obstacle might exist in the middle of a sector, then what?
A second approach would be to compute a path from the center of a sector to each border and store if such a path exists, which would take more time to build, though only rarely you'd need to update the table anyway and changes would be focalized.I read some articles where the idea of portals are used, for each sector you store where exactly the connection exits. This would require one bit per tile, so if each border is composed of 10 tiles, it would require 20 bits or so, and much more time but overall it gets amortized. Last but not least, I wonder how one would efficiently store information from border to border within a sector.

>>103953537
Someone posted an article on how Factorio does it in the last thread which is pretty good
I take the sectors and I do flood fills to find the sub-sectors then build connections between the sub-sectors, but this isn't done in real time and Factorio's method is faster

>>103953548
I read the article, their approach is interesting. They compute some some "abstract nodes" but use this information for the heuristic of the old one:
> So what we end up with is this: We have two pathfinders, called the base pathfinder, which finds the actual path, and the abstract pathfinder, which provides the heuristic function for the base pathfinder. Whenever the base pathfinder creates a new base node, it calls the abstract pathfinder to get the estimate on the distance to the goal. The abstract pathfinder works backwards – it starts at the goal of the search, and works its way toward the start, jumping from one chunk’s component to another. Once the abstract search finds the chunk and the component in which the new base node is created, it uses the distance from the start of the abstract search (which, again, is the goal position of the overall search) to calculate the estimated distance from the new base node to the overall goal.
>Running an entire pathfinder for every single base node would, however, be anything but fast, even if the abstract pathfinder leaps from one chunk to the next. Instead we use what’s called Reverse Resumable A*. Reverse means simply it goes from goal to start, as I already said. Resumable means that after it finds the chunk the base pathfinder is interested in, we keep all its nodes in memory. The next time the base pathfinder creates a new node and needs to know its distance estimate, we simply look at the abstract nodes we kept from the previous search, with a good chance the required abstract node will still be there (after all, one abstract node covers a large part of a chunk, often the entire chunk).
I wonder if this is faster than computing first the sectors, then subsector pathing.

>>103945682
>>103947026
This. Just use LÖVE.

>>103953610
>I wonder if this is faster than computing first the sectors, then subsector pathing.
It depends
In Factorio, you can change the map, you can build structures on it, you can replace and remove water. So nothing is fixed and the pathfinding has to be fully dynamic. For a dynamic approach, this is faster
If your map has static elements then you can do some precomputation, calculate all the subsectors and their connections at compile time

>>103940759
Day 4
Gun model, basic enemy logic, and UI changes

Glory to the Glasgow Haskell Compiler.

progress report:
finished initial multithreaded rendering and have gained 25% more performance just like that

ill add a bit more before taking a deserved break

I'm starting my project to modify my game framework to run in the web browser on WebAssembly. It shouldn't be that hard, r-right?

>>103929640
yeah, cool kids are already at atan3 with atan4 coming out next month. Get on my level gramps.

I saw this in another thread
https://youtu.be/j1aEU1okCAQ?si=-Yg2HH8OHf4sW6Ii

Never forget that Sony robbed us of the Lisp renaissance.

>>103935991
SUCCESS!

>>103945922
On Windows, you can use window messages, mutexes shared between processes and a bunch of other things.

On Linux, the usual way to do this is doing a blocking read on a file descriptor

Which books to read for game engine design and for implementing lighting that doesn't look like shit (unironically)?

>>103935991
>>103956369
Godspeed. Interested to see how this turns out.

Sup lads.
Haven't been here in a couple months or so (I'm not sure how many) because I went down a sonic autism rabbithole recently and experienced a fuck ton of adventure-era sonic media and watched movie 3. In the midst of this, I, in an act of infinite wisdom, scrapped a lot of my existing project ideas (such as a potential city sim, a potential fps, and a potential life sim) and kept musing about the idea of a 3D sonic fangame.
Now, I've decided that that would take too long for a fangame (I've estimated it to take 2 years to make based on my spec), so I'm considering making a distinct "sonic-like" indie game instead.
Picrel was my gameplay ideas from when I was going to make said fangame if you're curious.
I'm also going to try brainstorming some of the ideas I scrapped, now that the high from sonic movie 3 has worn off.

inline constexpr std::uint8_t fastSaturateIntToChar(int val) {
    constexpr std::uint32_t mask1 = (~255) ^ (1 << 31);
    return ~(((val & mask1) ? 0 : ~val) | ((val < 0) ? 255 : 0)); // 5250u with benchmark overhead
    // return ((val & mask1) ? 255 : val) & ((val < 0) ? 0 : 255); // 6790u with benchmark overhead
}

I have absolutely no idea why the first variant is faster (and the other is as fast as the naive approach), but here it is.

>>103956752
Thanks! It's motivating to know some people will find it interesting.

>>103956369
I can't believe in what I'm looking... Keep it up Anon.
>>103956560
Read Doom 3 or Tesseract source code, not a book.

>>103957238
Dude, literally the same second...

>>103957239
>>103957250
>>103956752
Awesome, I'm glad more people see the value! Now that I've got raylib's hello world example working, it's just a matter of expanding the bindings to the rest of the library. Then I'll port more of raylib's examples to prolog. Off to the races now!

>>103929599
General questions:
-- What are some things you'd like to see in indie games, regarding content and mechanics?
-- Do you consider a game being open source and modifiable good or bad, and why?
-- Are there some old games (any genre) that were never replicated again, mechanics-wise?
-- Is there incentive for you to sell your game on Itch or Steam, and do you think it'll sell?

>>103956369
you fucking mad man

>>103957022
what is the assembly like?

>>103929599
>spend years engine hopping and jumping from project to project
>finally start making excellent progress in Unreal on my 3D beat em up with blueprints
>been going good for around 3 months now
>actually feel a sense of accomplishment for once
>decide it's time to rewrite most of it in C++ for the questionable performance gain
>now I've been stuck for weeks trying to rewrite some of the most simple systems let alone the complex ones
>starting to feel that urge to start engine hopping again

I'm never going to finish anything. Learning all these Unrealisms and macros is making my brain hurt. I just want to have fun programming a game that doesn't require digging through forum posts for hours trying to figure out how undocumented bullshit works. Then even after all this some asshole will just call my game unreal slop.

>>103958420
When was the last time you had fun programming, as in writing actual programs, in static, compiled and linked programming language?

>>103958420
>>decide it's time to rewrite most of it in C++ for the questionable performance gain
With few caveats, I have vowed to never rewrite something that already works. Here is screenshot from a piece by Peter Norvig, who has been very influential to my own formation https://www.norvig.com/Lisp-retro.html

>>103957022
How do these variants perform?

inline constexpr std::uint8_t fastSaturateIntToChar_Naive(int val) {
    return val < 0 ? 0 : val > 255 ? 255 : val;
}

inline constexpr std::uint8_t fastSaturateIntToChar_NoBranch1(int val) {
    uint8_t too_big = val > 255;
    uint8_t ones_if_too_big = too_big * 255;
    uint8_t too_small = val < 0;
    return ((uint8_t)val * !too_small) | ones_if_too_big;
}

inline constexpr std::uint8_t fastSaturateIntToChar_NoBranch2(int val) {
    int neg_if_too_small = val;
    uint8_t zeros_if_too_small = ~(uint8_t)(neg_if_too_small >> 31);
    int neg_if_too_big = (int)((unsigned int)255 - (unsigned int)val);
    uint8_t ones_if_too_big = (uint8_t)(neg_if_too_big >> 31);
    return (uint8_t)val & zeros_if_too_small | ones_if_too_big;
}

inline constexpr std::uint8_t fastSaturateIntToChar_Naive2(int val) {
    // From https://stackoverflow.com/a/46876282
    if (val < 0) val = 0;
    if (val > 255) val = 255;
    return val;
}

Also, what kind of input values are you using for benchmarking? It might be worth measuring once with sequential inputs and again with random inputs to see the effect of branch prediction

>ask ai for some advice on coding problem
>the answer is actually exactly what i wanted
it's over, programmers confirmed deprecated

>>103958586
It answered about your JavaScript "problem", the one which billion jeets solved before you, which you could find on fucking SOverflow as well.
Ask it to write something semi-trivial in Ada or F90 and have fun looking through compilation errors, I tried it, it was so miserable to see AI do it.
I'm talking about literally placing C++ comments in Ada and F90 code, or thing that tries to be one, because there's little open source data for those 2.
Go to /twg/ for doom-posting, this isn't the place for that. Cheers fag.

I'm gonna be the very best
Enginedev there ever was

>>103958816
it answered a c++ template problem but do keep seething :3

Can't decide whether to do a
>3D fighting game
>3D platformer
>TPS
>life sim
>jrpg-like rpg
>morrow-like rpg
>mouse based crpg
>city sim
>fps
This is the problem when I have so many basic ideas.
I get overloaded and can't choose; I get decision paralysis.

>>103958586
over half of the shit i ask the ai for is horribly wrong

>>103958897
There are multiple reasons why you don't want to use templates in C++, but I'll leave that up to you, if you don't know why, learn it.

>>103958949
lmao, you are so fucking mad right now

Your game needs more cowbox

>>103958949
retard take

>>103959018
l2cubemap

>>103959018
i have a spinning uv cube, does that count as a cow with cool fur?

>>103958964
To be honest, I was waiting for you to rage so I can post this image, but I ended up raging...

>>103959039
there are very few good cubemaps available in the format I use (DDS). I have to manually build my own from old CS:S community resources

>>103959071
>there are very few good cubemaps available in the format I use (jpeg).
this is what you sound like right now

>>103958929
decision paralysis just means none of the above, it isn't real just an illusion

>>103959084
>none of the above
but I want to make 1 of those kinds of games, not 0.

>add spdlog
>it only flushes after program exit even when i call flush()
>check github
>other people also have this problem
I thought spdlog was mature and stable?

>>103959083
what do u mean anon

>>103959083
I wish the government would release more rare Pepes in .gif format...

>>103959092
glTF either encodes textures as raw base64 data in the file, or as URLs to images. Check your export settings in blender, and inspect your gltf file to see if it has a long buffer of base64 data or if it just has something like `file://path/to/texture.png`

>>103959100
what you want aint always what you need, usually you start with something you don't want to figure out what you really need
https://blog.crisp.se/2016/01/25/henrikkniberg/making-sense-of-mvp

>>103958578
Ok, so I changed the setup a few times and weird shit was happening, like explicitly casting float to int instead of automatically to uint in a single benchmark was giving a good result but affected some other separate benchmark functions negatively in a deterministic way, wtf.
Randomized data worked best for naive approaches, then std::clamp, then no branch1.
Executing several times per single randomized data point with added constant offsets is best for naive, then variants of my solution, then std::clamp and no branch 1 and 2 (and same as my commented out variant).
Naive 1 remains as the best option overall except for very specific cases it seems.

>>103959151
But I WANT to design/plan out a game before I start development on it.
I don't want to blindly develop a game not knowing what I want/need to make

>>103959133
this might sound crazy anon but you can convert any image format to any other image format and there's in fact a command line tool specifically provided to do exactly that for dds and most image editors support it either directly or through plugins

>>103959166
you can't, and i can say that you can't because you already admitted that you can't

>>103959176
Yes I know that's why I said I have to manually build my own

>>103957022
try this insane thing

return !!(x & ~0xFF) * (((-(x >> 8)) | (x << 24)) >> 8) + !(x & ~0xFF) * x;

Would it make sense to make a simpler game to test the engine your "real" game will use, and release that "test game" first?
Ala how Virtua Racing was initially made by Sega as a tech/engine demo for the Model 1 and it became its own thing released before Virtua Fighter

>>103959325
thats what i have planned on doing

>>103959286
This C++ code converts an unsigned 32-bit integer to a signed 32-bit integer, but only if the original value is greater than 255. If the original value is between 0 and 255 (inclusive), it leaves the value unchanged.

c# bros... dont tell the other langs about our embarrassing gc times please )': (6% of the frame budget at 120 fps)

with vsync enabled, do you guys set deltaTime to be exactly 1 / monitorHz if you can assume the game is hitting vblank? the FPS jitter is probably becuase of the operating system or whatever. the way its measured is just with a timer between frames.

>>103959502
you do know you can use arrays/pools and other memory management techniques to avoid triggering the GC at all in any language, right?
are you being retarded and making thousands of objects per frame for no reason?

>c#
can't wait to use dynspy on it for the free laugh and code :)

>>103959502
no i don't do that, I poll inputs at 1000hz but I only update the game state at 60hz and the renderer is completely detached from the update rate, I just let it pump out frames as fast as it can by interpolating the game state between ticks.

>>103959286
Nice, it's slightly better than mine in both cases.

>>103959415
are you sure, all my tests are passing

>>103959721
ha thats mental. how fast is it compared to the others?

>>103959715
why interpolate and not just render the latest state? to reduce felt jitter? how do you store the state?

>>103959502
Very pretty and sovlful

>>103959844
If you do exponential smoothing with deltaTime equal to refresh rate then you have to use power functions to restore frame rate independence. If your deltaTime is independent of refresh rate then you can use linear interpolation. It's just stored like normal data I don't know what you mean.

>>103959502
how this is yesdev

pretty sure you can query for it the monitor is ready to blank or not

>>103959811
It's mid tier for randomizer data, and in the great tier for corelated data.
I found the original test and I was transforming random numbers inside of the loop which might be throwing off branch prediction. In this case it actually beats the otherwise top tier naive function by 10%, but mine cuts it down by another 10.

>>103960057
would this cost you an extra frame of input lag then? but the jitter error would be very small im imagining. very cool

>>103960064
i wait for blanking, but still just because of os shit or whatever even if you hit the blank your timer used to measure time between frames is not gonna be exactly 1/monitorHz, even tough thats what the timings between blanks actually is. i think with modern apis dx12 or vulkan you can get timestamps when something was shown on monitor, giving an exact measurement between blanks to be used for deltaTime. im not even sure if these micro jitters are perceptible anyhow...

>>103958816
solved - that means memorized in hindi?

>>103960290
>would this cost you an extra frame of input lag then?
0.5 frames average actually

>>103959502
>with vsync enabled, do you guys set deltaTime to be exactly 1 / monitorHz if you can assume the game is hitting vblank?
There's a really good talk from one of the Talos Principle devs on frame pacing. IIRC they wound up doing something like this, with some heuristics to drop to 1/2 the vblank rate if the machine is overloaded. But they were also working on getting APIs standardized that would allow you to do some better thing.

>>103959715
nigger wants his gpu to die faster

>>103959121
>spdlog
All you need is 40 lines of code brother
Here is some code taken from ... somewhere.

typedef enum
{
  LOG_TRACE,
  LOG_DEBUG,
  LOG_INFO,
  LOG_WARN,
  LOG_ERROR,
  LOG_FATAL
  
} LoggerType;

#define LOGGER_CHARACTER_COUNT 2048

#define ANSI_COLOR_RESET   "\x1b[0m"
#define ANSI_COLOR_RED     "\x1b[31m"
#define ANSI_COLOR_GREEN   "\x1b[32m"
#define ANSI_COLOR_YELLOW  "\x1b[33m"
#define ANSI_COLOR_BLUE    "\x1b[34m"
#define ANSI_COLOR_MAGENTA "\x1b[35m"
#define ANSI_COLOR_CYAN    "\x1b[36m"

static char loggerBuffer[LOGGER_CHARACTER_COUNT];
static const char* loggerNames[] =
{
  "TRACE", "DEBUG", "INFO",
  "WARN",  "ERROR", "FATAL"
};
static const char* loggerColors[] =
{
  ANSI_COLOR_BLUE,   ANSI_COLOR_CYAN,    ANSI_COLOR_GREEN,
  ANSI_COLOR_YELLOW, ANSI_COLOR_MAGENTA, ANSI_COLOR_RED
};

void Log(LoggerType t, const char* format, ...)
{
  va_list args;
  va_start(args, format);
  vsnprintf(loggerBuffer, LOGGER_CHARACTER_COUNT, format, args);
  va_end(args);
  printf("%s[%s]: %s%s\n", loggerColors[t], loggerNames[t], loggerBuffer, ANSI_COLOR_RESET);
}

>>103958586
>Ask AI to help solve linking problem
>None of the answers are correct/work
>Search google/read stackoverflow
>Same thing
>Dig through documentation
>Success
I think we're fine.

without tonemapping / with tonemapping

anyone building their game for HDR monitors? whats the workflow like?

>>103960995
anon you don't just check the aces box and call it a day, you're supposed to design your textures and lighting around it

>>103960995
what's an HDR monitor? is that 64 bit color or something?

>>103961044
local dimming, mainly OLED and to a lesser extent miniLED

other displays that claim to support HDR are full of shit

>>103960995
It looks washed out. Can you crank up exposure?

>>103961036
its not aces, what does that mean though? is there a guide somewhere?

>>103960995
I looked into it and I realized there wasn't much point considering hardly anyone has a HDR monitor (except mac users)

im making my own hash map in c++ for the first time, i hope it works out

>>103961289
dont bother trying, its impossible

Alright I got a window/event loop up and running and converted one of the DirectX9 SDK examples to C/said window framework
What issues will I have to deal with to actually have DX9 as an option alongside OpenGL 2.1

>>103961326
now i want to do it even more
we be using concepts
>>103961354
sorry that i like using namespaces and RAII

Work in progress, refactoring script parser...

>>103961470
i wish i could read that

>>103961478
Here it is...

>>103961384
Namespaces are good. RAII on the other hand ...

Realistically speaking, I want to make a (specifically 3D) game and have it out by mid-late 2026 or early 2027.
I wonder what kinds of 3D games would be less time consuming to make (and thus would be ready by that time).
Arcadey style games perhaps? Such as Virtua Racing, Star Fox, Tekken, etc?

>>103961671
If your only requirement is "3D" then you can just make any shitty game with low production value in Unity

>>103961686
Obviously that's not my only requirement, no way I want to make something that's ass

>>103961707
If you only have a year then you're going to make something that's ass unless you're a very experienced dev

>>103961726
OK, how about 2 years then?

I played fortnite 14 hours instead of working on my engine

>>103961871
its not like you would have gotten anything done, right?

>>103961871
>fortnite
underage b&

>>103961732
ass

>>103961882
I was finishing up skeletal animations

>>103947056
>on code organization, software engineering
practice and fucking up a few times your projects.