/v/ - Video Games

Nice subject line. Sage for no contribution.

>>15689462

>(((how do you do fellow anons)))

>>15689495

not attaching a dick to a woman

>>15689462

Its futa shit, no one should be interested.

>>15689526

>I wanna be able to cater to most if not all audiences

You can't, fetishes conflict with each other.

>except furfags unless a lot of $$ is on the line

Furries always pay tons for shitty furry art, do you live under a rock?

>>15689534

He doesn't seem to be around here, as evidenced by his overuse of double dots.

>>15689524

All of the assets. It looks like an RPGMaker porn game where the dev just used the default character sprites and their favorite porn pictures regardless of the clashes in style.

>>15689560

>self-admitted crossposter futafag /pol/tard

Nigger please. Lurking 8/pol/ nowadays?

>>15689603

Do you even know what sage does and why it's used?

>>15689616

>backpacks, tiles and ohio

>implying you need to lurk more than a few days to know about these

>implying it's even old

Try something else, faggot.

Interesting you avoided answering my question about the use of sage, though.

I made le epic tech program

#include <iostream>

#include <math.h>

#define PI 3.14159265

using namespace std;

int main()

{

double c;

while (1)

{

for(double a = -PI; a< PI; a+=0.35)

{

c = sin (a)*10;

if (c>0)cout <<" ";

else c+=10;

for (int b=0; c>b; b++)cout << " ";

cout<<"waves\n";

}

}

}

>(10) posts in a 22 post thread

>All shit and spam

>Gets deleted

>Now /v/ is being spammed

Really makes you think

>>15689628

Look up control characters for more funky effects on the terminal, such as deleting a single character.

some members of my team left and i must work on the code alone now until i find more people, having this happening after 3 years of devving and 1 year into the current project hurts just a bit

Reposting:

I wanted to try 3D but I couldn't get Ogre or Irrlicht to compile, It's insane that precompiled libraries do not exist, I wanted a library that can do both directX and openGL, mainly the former because I'm on a really old machine with shitty openGL support, but if I wanted to compile any library with that I need to download the whole SDK and whatnot and I can't do that, I simply just gave up, if anybody is using those libraries I'd appreciate linking them,windows 32bit with all the backends and the header, I need it to make the bindings.

If there are other options I'd like to hear them too.

>glBindBuffer somehow causes GL_INVALID_VALUE

I am ready to kill myself, I literally did this a few days ago in a different codebase and it causes GL_NO_ERROR (I am testing it now). They have pretty much the exact same start up.

 GLuint vbo;

glGenBuffers(1, &vbo);

error = glGetError(); //<---- error ==  GL_NO_ERROR

glBindBuffer(GL_ARRAY_BUFFER, vbo);

error = glGetError(); //<---- error == GL_INVALID_VALUE

I am following arcsynthesis, if you have had this same problem at the end of chapter 1, what did you do?

>>15689906

By "pretty much the same start up" I mean exactly the same functions in the same order, just restructured.

>>15689818

What is your project?

>>15689485

>>15689511

>>15689504

What did i miss?

>>15689906

>pretty much the same

Read: different but I can't be arsed to look into how they differ

How is somebody supposed to help you that way?

>>15689959

Looking at the log it seems his posts might have just been deleted.

>>15689952

Okay, they have exactly the same setup.

>>15689982

The only difference is that the HWND is global rather than scoped to the entry point.

>>15689912

(((Anon))) shilling futa shit.

>>15689912

if i get my server fixed in time i will host more matches on the next demo day

but no new weapons, sorry about that

by the way, where does one take a prototype of a game or search for freelance coders? just a month ago everything went full sail ahead, i wasn't ready for this

>>15689982

>GL_INVALID_VALUE is generated if buffer is not a name previously returned from a call to glGenBuffers.

There has got to be some other important thing happening in that function, but if you don't post code: Check that the buffer name after allocation and before the bind call are the same. My necessarily uneducated guesses are that you reuse it somewhere in between or that it's being overwritten by memory corruption.

>>15690163

just record the input whenever you press an input, by the time you make that kind of system, you can implement undo and redos

>>15690212

I've been leaning towards that, but I don't know how an undo would work. I know it's basically just a delta input, but how would you "undo" and explosion that created two or three object types, and destroyed underlying objects to create them?

In any case, that's more or less what I've done, or plan to do. Every time a change is detected, it gets added to a list. When Save() is called, it pushes it to a replay buffer. When Load() is called, it ignores input and reads it from the playback buffer instead. So in a sense I can use savestates by recording actions, but not cleanly undo things

>>15690251

>info about previous state

Ah okay, so that's different than how I'm approaching it. I want my replay files to be fairly small, so every event is just 32 bits in size: 16 for the timestamp which cycles, 8 for p1 input, and 8 for p2 input.

>Performance is shit, try to fix it using this nifty ECS thingy that happens to have a InstancedMeshRenderer component

>Work for two days to get some shit working because Unity devs left it half done

>After finally finishing it, it runs 5 times slower than normal method

I guess I will just do it like I did my grass

>>15690277

You could just simulate all required commands from the beginning (without graphics of course) when the player clicks undo. If games are very long and/or undos are very common, this approach probably needs checkpoints where you store the entire state though.

>>15690292

>You could just simulate all required commands from the beginning (without graphics of course) when the player clicks undo

Yeah, that's what my plan is. Well, there isn't an "undo" command per se, but checkpoint items that are collectible that save the state in case they fuck up.

Because of how I programmed the tile handler, levels have a hard limit of 256x256 tiles, but have a soft limit of 128x128 tiles, and 2048 actors at a time, so simulations shouldn't take long at all on modern hardware.

>>15690319

I see that your inputs are just related to movements on a grid, instead of storing the input you can store a simple number from 0 to 3 (up, right, down, left), and a timer related to the last input, keep in mind that the timer will be 0 when the inputs are held so you can optimize it away for successive inputs, you can easily undo a movement by playing the recorded list backwards (if the original direction is up then you move down…. etc).

For long plays you can use check points after some arbitrary number of actions, then you compress records between checkpoints, you only need to make sure that you can compress and decompress your records in time (tweak your values), anything that is too old could be written to a temporary file if even compression isn't enough, this way you can pretty much have hours and days of records, that's how Braid did it at least, and it kept track of a lot of objects.

Does someone know any way to combat procrastination?

>>15690400

put the best anime opening you know in loop and listen to it as you dev

>>15689982

>>15689906

So after 3 hours or whatever, I have found out that it is something to do with the function loading. If I copy paste the function loading function, it just works. I don't know what I did differently, it's hard to tell because there's limited code highlighting in my new editor. Will report back.

>>15690437

Then it's already too late. ~4 months

I will make any simple game as first thing tomorrow morning.

>>15690451

It will be hard. It was several years since I watched one.

Power metal has to do.

I've posted it here before, but do rpgmaker horror games get a pass here? Or is it straight to the gutter?

>>15689811

Is that the priest from the Twisted Metal reboot?

>>15689811

Whats your experience with how much cloth impacts framerate in Unity? I tried setting up cloth physics in the past, but I don't know if I was to blame for it performing poorly or if that's just normal.

Had the idea to give the "fighter/brawler" character for my rpg a gimmick where some of her skills power up if she's drunk, so far that list includes Fire based skills and one that boosts her strength and max hp but cuts defense. (To represent how drunk people seem 'tougher' but really just don't feel the damage they're taking because of deadened nerves)

Has anyone encountered this shit before?

Trying to use glActiveTexture crashes the whole program, always returning -1073741819. Doesn't matter what order I put the other GL functions in or whether they're there at all. While a window does open, I can't get anything to run before it crashes, can't even cout anything. There's a lot of shit I'd have to remove to minimize the program, so I'm asking now in case there's something simple that might work. Spent the past two hours googling and nothing useful came up.

>>15692755

Those tits

the day i see an h-game dev in one of these threads man i tell ya,i am gonna hop on their back

>>15692755

Is it gay if you're more atracted to that fukken impressive rigging or that perfectly smexy topography instead of the girl?

>>15692858

Not at all, the model is an abstract symbol of a real woman, but the rigging and skin weights took time, effort, and technical skill. They're the most human features in that video and therefore are the most relatable aspects. I mean dear god those thigh muscles when the knees bend are fucking magnificent.

>>15691182

because it's a paid feature on wiki hosting sites and I'm a cheapskate. I was thinking of hosting the wiki but I lack the bandwidth

>>15693494

> I'm not developing but want to try modding a game that runs like shit to see if I can improve it somewhat.

Don't do this shit for free, goy.

>>15693494

unity is as good as you are.

>>15691983

Gets a pass from me, anon, but only if it's fun so not a walking simulator, try to give it some actual gameplay and you put some effort into distinguishing it from the typical RPG Maker shovelware so using custom assets over the RTP, modding the UI, debugging and polishing before release, attention given to color theory, level design, and difficulty, etc.

>>15691983

>Is that the priest from the Twisted Metal reboot?

All characters are original, i made them all

>>15692017

So far it had very little impact, but we only apply to hair and a few cloth pieces like the jacket and the scarf of the protagonist

Anybody here published a game on steam? I'd like to know the steps it took to do so, I want to make games on my own and publish them, but I have no experience in legal procedures, it seems that steam requires taxation info and such, you can't just give them your bank account number and let them make transfers? I don't want to make an LLC and deal with random shenanigans, I just want to make games and publish them.

>>15695214

We clench as the demo day draws closer, which is the opposite of what they do, yes.

>>15694670

You don't want to do business without an LLC. This year I bought a couple tablets from alibaba and ordered some metal box frames to put them on a wall for a local aquarium. I made some mistakes with the tax paperwork and generated 3 different fines. If I had worked as an LLC I could have closed the business without paying them. Instead, the fines were on me. Never risk your assets, this is why limited responsibility companies exist.

>>15694670

Personally I will open a small company before submitting anything to steam. I don't know about you but the government here will taxes me depending on my revenues and having a company would limit how far they can ram your ass if things go bad.

If you were to declare bankruptcy (I hope you won't and you'll be successful) then only the company asset can be seized, otherwise it'll go with your personal belongings. There is also other stuff related to laws.

On the upnote, if you get more popular and bigger, you can start hiring/commissioning work from artist and put that as business expense, which gets deduced from your game revenue and limit taxation.

So, whatever happens, it's getting a bit more serious.

>>15694910

Almost the same for me, I have a bit of a slump right now.

I did at least fix a bit my models yesterday (still looks like shit)

For demo day:

- Need to make a few paths for enemy to appear and shoot at player

- Have enemies have the correct orientation when moving around

- Fixing any problem when compiling under windows (developing under freebsd right now)

- If possible, hard select weapon with keyboard (just cosmetic for now)

- Explosions and other animation if I have time

Then I should have a really basic demo: enemies running around shooting at player following a path. No score, no end state, no menus.

>>15695324

>>15695417

What can go wrong when you sell games tho? I know you can be sued for copyright infringement or loose money if too many people refunded, what other dangers are there?

Starting a small company requires a decent amount of cash and paperwork, I live in a small country so there's all kind of laws and bank fuckery. It doesn't help that I don't understand how to deal with a digital product and how can you do taxation work for it, there's also online money transfer and banking which is a separate beast.

I asked around and the best bet was to use itch.io and ask them to send money to your credit card directly, payment there isn't as complex as steam, and there is no refunds, worst case would be to get removed.

I wouldn't touch paypal or any other online service since they have a really bad reputation, paypal can straight up lock your account and hold you money there for months, banks here have some special deals with it, to my knowledge there's only one which accepts it at all, and even then they have a yearly limit (about 1500$) on how much you can draw, is SWIFT the same?

Starting from zero is really painful, you need money to be able to get money, it's easy to get stuck in a negative loop.

>>15695663

I wouldn't recommend it for the reason I told you: you'd be making your entire assets, future and yourself liable for anything. I got 3 fines because I fucked up the tax paperwork. Had I been an LLC I could've just closed it and not pay.

>>15695934

Okay, I'll spend more time to research LLCs and taxes, should've studied accounting since it's required for everything instead of the useless shit I wasted my time on, thanks, still want to know about all the pitfalls concerning online transfers and such.

Newfag here, what's the best development kit to build fps games?

>unity is preset garbage

>>15696087

Well, everyone wants Godot to succeed but it'll likely require more work on your part than any other established engine. Unless you're willing to do that you're stuck with either unity or unreal.

>>15690400

Work everyday, never do a "just today i gonna goof a bit", if you do it once it is all over

>>15696986

Seconded. "One day won't make a difference" is my classic pattern for abandoning what I started.

>>15696986

>>15697012

It's not bad if you have a consistent and logical pattern to it. For example a specific day each week. Or every time you finish a significant thing, you get to take a day off.

>>15697662

Nigga, that's kawaii. Also better animations than bethesda.

>>15697734

I think a lot of animations look stiff because of the logic layer that goes on top of the animations afterwards. I'm gonna try to fix it. It's cancelling out too much body motion and it makes her look robotic sometimes.

>>15697662

very cute as usual, nice

*the hat webm last thread gave me a good laugh too**

>>15696169

Only thing I dislike about modeling anime style is that their faces look like alien abominations pre-shading; while doing cel-shading is antithetical to doing any type of 3d work.

>>15695934

wew, good to know

>>15697662

super cute, but why are the donuts made of shit?

That can't be healthy

>>15698199

nevermind I was doing an integer division and didn't notice

>>15698207

what does regular division look like in unity

i'd assume it was that

>>15698217

This has nothing to do with Unity.

In languages like C, C++, Java and C int / int = int. So if you want to 10 / 20 to equal 0.5, you need to turn it into either float / int, int / float or float / float. All those yield a floating point number as result. Just be explicit about your types.

Debug.Log(10.0f/20);

>>15698108

Do you mean splines?

>>15697662

The animations are a bit rough around the edges but it's absolutely great for a one person thing, you're doing a good job.

>>15698246

Why are you explaining this? Spoonfeeding retards doesn't help them, and we don't want them going here for easy answers to shit they could learn in their first typing class. Don't shit where you eat.

>>15698246

ah, i've only ever learned lua from roblox and python, didn't realize math is like that with more serious languages

>>15694171

The only default assets left are the character sprites that I want to replace, but I haven't gotten around to doing.

I'm trying to not make it a walking simulator, but at the moment I am having a tremendous hell of a time trying to implement a sort of LoZ-ish combat system. I found what I think will be a good system, but the instructions are all in Portuguese, so tinkering with it to make it work for what I want going so hot.

D-did you want to see some screenshots..?

>>15698606

It has nothing to do with seriousness, languages that do this either do so because they want to take advantage of hardware or because they were designed by lazy morons. There exist tons of languages that give you a fraction if you do 2/3 for instance.

>>15697662

Nice. The skit-clipping still triggers me though

>>15698105

Are you going to be releasing anything on demo day?

>>15698606

LUA is simple enough and therefore the usage you get from it follows the same idea so you don't have to fiddle with any of that.

More complex languages have to properly specify what kind of variable you're working with to properly allocate memory or use relevant functions that only work with a specific type.

Or you can go with Javascript, where things are what they seem to be at the time. It's like being gender fluid, but it's variable types instead of gender and you don't have to be crazy to believe in it.

It's also kinda usefull if you want to round numbers easily, so if you divide 246/32, you'll just get 7 which is a close enough number. It doesn't make sense to allocate the extra memory if you don't need the extra information afterall.

If you have experience with LUA, you might want to learn a more serious language. You'd be surprised how simple half of the process is (and how hard the other half is) since a lot of the logic behind what you wrote still aplies.

You usually just have to read up on all the cool new shit that other languages let you use to craft a much more powerfull code, like dynamic arrays, enums, inheritance, polymorphism, composition, etc. But since you can read those as you need them, you're never stuck because you lack knowledge, you just find better ways to do the same thing every day. It's pretty great.

>>15698737

>>15698606

>>15698933

More importantly, it's because operations tend to behave very differently depending on the variable type.

Even when you're dealing with floats and integers you will come across differences.

Some edge-cases are handled differently, dividing a floating point by zero results in infinity, whilst in integer land this operation is undefined. (Raising an exception, which crashes the program if not handled, is common across many languages.)

There's differences in accuracy. A single-precision floating point has less accurate results than a double-precision floating point. This might seem like a reason to simply use the most accurate variable, but integer operations are the closest you can get to being "exact". As floating point numbers increase in size, their accuracy decreases (as it stays related to the size of the number). As a result, a 32-bit float set to 2^24 can't be incremented by 1. Only by 0 or 2. But a 32-bit integer can be incremented by 1, all the way to its maximum value.

Another issue is operations having completely different meanings altogether! Boolean types exist in many languages, and for those types, 1+1=1. Because "true or true" equals "true". Some languages even have variables that exists out of multiple numbers, such as complex numbers or vectors. How multiply operations behave for those two is very different, yet they might all be multiple floating point variables stuck together underneath the abstractions.

The main difference between scripting and programming languages is that programming languages try to be very exact and predictable, whilst scripting languages are a lot looser and "get shit to work". Many scripting languages don't even support multiple types of numbers, forcing everything to be a double precision floating point. When you use programming languages, be sure to ask yourself how code behaves around its extremes and do research into what exactly happens on those edge cases.

>>15699038

>>15698933

TLDR: Languages that are strongly-typed don't have ambiguous cases

>>15699427

I remember. Nice to see you back.

Are we going to get a demo this 11/11?

>>15689811

How do cloth physics and rigged mesh mix? Can vertices have smooth mix between being affected by bones and physics or is it 0/1, either full bone or full physics.

>>15699464

I don't know about unity but in theory you could mix them. Anything's possible through programming.

>>15699681

https://en.wikipedia.org/wiki/Halting_problem :^)

>>15699702

you cheeky cunt.

>>15699518

It's a cute, albeit repetitive with only the memory game. Are there other ways of earning rings?

I'm doing some tests with NavMeshAgents for the first time in Unity.

If I don't have a specific position I want my character to move to, how should I decide that position?

For example, my character approaches to hit an enemy, after getting close to the enemy and attacking I want him to step back a bit before the next attack.

Is it just fine if I do

agent.SetDestination(-character.transform.forward)

?

>>15700407

There could be some problem with obstacles. For example if your character would fight near thin but long wall, the destination could end up being behind the wall and character would go 20 meters to reach it. Also if I remember correctly, if destination would be inside the blocked area, character would keep trying to walk into the obstacle it certainly works like that if the area is blocked by another agent, not sure how it works with dynamic obstacle component.

>>15698933

i'll consider picking up a more serious language. i only ever use programing to turn math algorithms into functions

>>15700461

That's what I was worried about. How should I go about it then?

>>15700162

I appreciate your enthusiasm, but for the life of me I can't understand why anyone likes head bobbing.

>>15701179

It's a stealth game, you need to know when your next footstep will be. You'll be able to disable it and show a widget if desired.

>>15701210

You should model it after Thief. Head bobbing is unnoticeable when you're running, but when you crouch walk you can feel every footstep.

>>15701256

>Making a Milf

She better be thicc.

Is there a way to start a Coroutine from inside a class?

Googling didn't help, Unity is still telling me I need an object reference for the non static field StartCoroutine. Even by instantiating the class it's giving me the same fucking error

>>15701261

We have a hips slider on the model, so yeah

>>15702775

need to get some antisotropic filtering on that asap my dude

>>15702307

>measuring performance in FPS

>not counting cycles

>not even using frametime

>>15702775

>>15702942

Add some TAA as well, those tiles will look terrible even with 16x.

>>15703054

MSAA exclusively helps with edges of geometry by oversampling them in the rasterizer stage. It doesn't do anything about specular aliasing. If you don't want to get into the fuckery that comes with TAA, you could try mipmapping normals differently. When a regular mipmap of a normal map is generated the normals are averaged, which is physically incorrect. There are a couple of techniques for "smoothing" normals instead, which does diminish specular aliasing as well.

>too lazy to figure it out right now

There's always FXAA as well.

>>15703054

eh, to be fair you will probably not be that low to the ground when you play so it'll probably be less noticeable

>>15703188

>Network: (no players connected)

Working on a multiplayer FPS? Using snapshot interpolation?

>>15703249

No, at least not for the foreseeable future. I'm not sure why it even mentioned anything about a network.

>>15703265

Damn. What physics engine and first person character controller do you use?

>>15703054

>>15703188

Somewhere in vid related the Robo Recall devs show a texturing trick they used to kill specular aliasing without TAA.

>>15703297

They are basically just deactivating specularity by pumping roughness to max on normal map height changes, there is a better way:

https://research.nvidia.com/sites/default/files/pubs/2016-06_Filtering-Distributions-of/NDFFiltering.pdf

>>15703271

Found someone's port of quake 3's physics in JS and I ported that into C#, it's ok for screwing around but I need to get around to making something deterministic for replays to work.

>>15703339

>>15703297

bookmarked

>>15702977

It's just a quick way to see how it's doing, its not a big deal to measure it one way or another like that if you just want to see the basic idea of what's going on. It's not like I'm trying to quantify the effects of a small optimization, but rather, a major rewrite.

Are your demo's ready?

>>15699427

Welcome back anon.

>>15703552

Maybe next year.

>>15703582

fugg

>>15703679

Hard to say without context of the body. As it is it looks fine.

>>15703552

Functionally it is but I need to get some more demo content done before d-day.

>>15703552

Here's a question, are we going to use the 11/11 image we used last year or should we create a new one?

http://8agdg.wikidot.com/demo-days

>>15703827

Trap "programmers" just want to put CoCs in to everything.

I wonder if this is the feeling Japanese students feel right before a school festival

>>15704265

Is this the new sanic game?

>>15704281

no

>>15703877

not to mention they're usually mentally ill

So i think the only core thing i need for demoday on my puzzle game is object interactions. I have 2 days off to work at it… no idea how to bridge the gap from arbitrary text into function calls. I have asked about this before but i am a dumb

>>15704622

You can get the text into functions with Reflection, or you can make a giant if/else

I don't think there is any other way

The Reflection solution changes slightly from language to language, but it's usually something like

Class.GetClass("MyClassName").GetMethod(your_arbitrary_string).Execute();

>>15703552

I wish. I intended to have a demo ready for this demo day, but all of my progress has been on important things that I can't really show off in a demo yet, so I'm going to have to wait until February.

>>15704265

He moves like he crapped himself.

>>15704881

I took the animations from some shitty free package, my brother is slowly replacing them.

Currently he runs on water too

>>15704265

This is a harvest moon kinda game, right? Grass looks amazing, btw.

ITT: Mediocrity

Stay losers, forever.

>>15705299

Stay gone, forever.

>>15705304

Triggered

>>15705299

In the land of the blind the one eyed man is king. Loser.

>>15704881

I will replace the animations after the demoday

>>15705230

Yep, also cutting grass will be part of the game, it will become cow feed

>>15705456

>cutting grass will be part of the game

I hate that this makes me excited because i don't want to cut my own real fucking grass.

>>15705535

>doesnt know motherload

Get a load of this pleb.

>>15703827

fuck off back to >>>/cuckchan/ with that degenerate shit

>>15705299 (you)

>>15703391

>Found someone's port of quake 3's physics in JS

Care to link it?

>need to get around to making something deterministic

As soon as you have a server, you have a single authoritative game state per server tick. As the server already only simulates physics in fixed timesteps, everything is practically deterministic at that point save for a little bit of numerical instability that gets brushed over every 50ms / snapshot interval.

Grass generation is quite a bit faster, and the scythe is back for drawing swastikas on the grass.

I will try to get the grass to instantiate during edit time and cache the result for instant loading during playtime, but I might not be able to finish it tomorrow.

>>15703827

Hang yourself

>>15708529

I was kinda excepting the potion seller pasta fam.

>>15708716

I'll probably get something out, but I'm afraid I'll have to cut some of the stuff I wanted to show off because it's not fully compatible with my reworked AI yet.

>>15708716

>working on the sabbath

REPENT SINNER

>>15701754

>>15701673

Good progress, anon. Keep it up.

>>15701754

are the bullets going to be actual projectiles or hitscan?

>>15709735

Thanks anon. Hopefully I'll have something for demo day.

>>15710121

Kind of a hybrid between the two, but effectively projectiles. The external ballistics and parts of the terminal ballistics system are going to be pretty in depth.

Let's say you have 2,000 particles and you need a method to see if one particular particle happen to be adjacent to another one. What is the best way to store/poll them?

That is, I have it implemented as a large array of structs for performance reasons, and the object itself is fairly heavy so I can't recklessly copy or pass them around. Because of that, they aren't indexed according to position (and besides two could occupy the same position, so a hashmap/dictionary won't work here). Not sure how to do this

>>15710504

Oh actually I think I have a good idea. Since I'm pre-sorting the array every frame (via Array.Sort() which is very fast for some reason), I can then iterate over it and build a simple int:int hashmap that links a position with an array index, perfect.

>>15710601

No. That's average frametime.

If one frame takes 500ms and the other 59 take ~8.4ms, you'll still have 60fps. But the game will run like fucking shit.

On top of that, you can quite easily query the GPU for how long it took to complete a frame by simply slapping a query timer around it.

>>15710910

One thing I did was keep a rolling average of the last 300 timestamps (5s @ 60 fps) and see if it was running too fast or too slow. Is that sufficient?

>>15710913

FPS and frametime are different statistics that have different meanings and can be used to diagnose different issues.

Keeping track of frametime tells you whether an individual frame notably slower than others; which can cause framerate hiccups.

FPS (or your roling average frametime) shows average frametime, and tells you whether the frame timing (when you actually start on a frame) is good.

>>15711141

>saying only f-zero when f-zero is fundamentally different from f-zero X and GX

Even the handheld games similar to the original snes f-zero game is also different from the X and GX games, instead of drifting to turn you would blast turn, it is present in X and GX but has less of an effect, so the snes/gba f-zero, pressing the accelerator allows you to corner a lot better without losing too much speed, I really liked that but found shifting along the horizontal axis didn't help steering too much and was used exclusively to line your vehicle along the ideal racing line, which is the only gripe I had with the original F-Zero games. Please avoid retarded racing game gimmicks like Fast Racing Neo's gimmick, some stupid flying option, or the weapon shit which detracts the game from a racer.

I would like to see some of the churayzee tracks from F-Zero X and a track editor like F-Zero X 64DD, that would be fucking rad

>>15711240

I was copying F-Zero X, i never liked GX much.

>instead of drifting to turn you would blast turn

I had that, but after i made the driving more physics based i removed it and instead you have to deacellerate and/or drift to take tighter corners.

>weapon shit which detracts the game from a racer.

What do you dislike about weapons?

>>15711404

Weapons isn't racing, it's fucking gay, at best I would like to be able to enable/disable it. I have heard that the Sega All Stars Racing has been praised for doing a great job on weapons in a party racing game.

>>15711427

>at best I would like to be able to enable/disable it

That's a good idea. I personally like weapons but those who don't like them can avoid them this way. I'll also take a look at Sega All Stars thanks for the tip.

>>15711491

Will there be more equip slots later?

Do one ring per finger!

>>15711503

and a cock ring

I'm only currently considering jewlery, otherwise I don't think I wanna have separate shoulders, belts etc

>>15704904

>Currently he runs on water too

It's a feature!

>>15711491

>Though the armour is still just a cosmetic item

Doesn't it take like 5 minutes to add a float to the armor objects and subtract that float from damage dealt to you?

So tomorrow is that day.

Does anybody here is hopeful about it? Because I don't.

>>15712283

Do you have a game screenshot?

>>15712283

They do it slowly. Seeing how many NEETs here still don't get shit done despite all their time, they probably aren't much less productive.

>implying prostitution isn't productive

>>15712283

>I don't know how anyone does anything productive without being NEET though.

That's because being NEET is cancerous to all kinds of motivation.

t.NEET

>>15712283

>all these new-age ideas where kids aren't taught the discipline of a hard day's work to make a reasonable salary

>muh slavery

>being paid money is slavery

everyone who thinks they should not have to work for a living is either a gypsie, a jew, or a nigger, and nothing precludes them from being all three.

>forced to work

by the same logic that people are forced to work in order to subsist and survive, it could be said that you are "forced" to drink water and eat food. Your body needs food to survive, how oppressive it must be to be organic and consume.

If you really didn't want to deal with modern "problems" you'd go live innawoods and fuck off forever like /k/ did about a year ago; but then you'd realize that in order to live you need to do "so unproductive" things like harvest and build an actual home and protect your body from the elements.

What a hard life it must be to have a desk job where you don't have to weather the elements and everything is handed to you and you still have to complain that you spent ~8 hours working.

sage for offtopic, the bait got me good.

>>15712505

>seriousfagging against MS Paint strawman comics

>>15712566

>Meme War, the hyper-gamified forum

>Mr. Boomer […] Millennials like you

>the comic was meant seriously

If this is a joke it's a damn good one, what the fuck.

>>15713171

seems that my post shows up on the main board page while being invisible in the thread itself. nice.

>>15713132

Can you fuck the goblins

>>15713235

No, they're dead now.

>>15713334

Yes, and?

you faggots better get ready to start the actual demo day thread.

>>15713702

Yeah, so is it getting posted in 7 hours or what?

>>15713886

I hope not. I'll be asleep by then

>>15711511

You could keep armor for general mobility vs defense stats and keep all enchantments on rings

This way you can have more flexible gear since all special effects are contained in general equipment (jewelery)

Also please don't do Item sets, they fuck with versatility and make most item set gear unoptimal without all the parts

>>15714291

>Also please don't do Item sets, they fuck with versatility and make most item set gear unoptimal without all the parts

My dude, that's the idea behind the whole item naming convention thing.

>>15714360

Add a torch or two

>>15714556

If you're feeling lazy, sure

>>15714603

>lazy

yeah, let's call it that

>>15714614

Just kidding man, I highly like what you do

>>15714614

Just kidding man, I highly like what you do

>>15714630

no offense taken, my dude. Crunch time is what it is, gotta get "creativite"

>uploading windows and linux 7zips to mediafire and mega.nz

>170 fucking mega each

shit takes forever

>>15714823

i fucked up that sentence, i meant:

Don't worry, the next girl*

>>15714820

mistakes into miracles

>>15708716

>Friday Night Magic

You can't do that when you want to get anything done on the weekend you tubo-nerd.

I've tried to do that in the past. It doesn't work, the time you spend recreational on Friday adversely affects your performance on Saturday and axes any hope of doing work on Friday.

Format? Deck?

Actually don't answer, you should be working on your demo.

Do we have a baker for the demo thread, or should I do it? I promise I won't forget the subject line this time.

>>15714858

ughthiccer.png

>>15714861

>he time you spend recreational on Friday adversely affects your performance on Saturday and axes any hope of doing work on Friday

>not drunkenly programming at 5 AM on Saturday when you get home

where is alien moon rpg

>>15695663

You might want to check more in details with people that know about this, but I don't think the amount of work is that huge.

I lightly checked with someone I know, and the most I will have to do for the company upkeep is filing revenues/expense for the company once per year. And there's accountants that specialize in small companies that can help you for cheap and they take care of most paperwork.

Got past my slump a bit late, and got surprised for some specific cases related with script behavior.

Everything work now, but I need to reboot and check if compilation under windows won't give me trouble (Murphy's law is being applied right now)

Going to sleep and hopefully I'll submit a small something later today.

>>15716319

give me the linux port

>>15716319

>FPS: 10142

you best be providing an option to cap framerate

its not your job to melt people's gpu's

Demoday threads are up:

>>15716503

>>15716503

>>15716503

>>15716503

>>15716503

>>>/agdg/32567

>>>/agdg/32567

>>>/agdg/32567

>>>/agdg/32567

>>>/agdg/32567

>>15716333

See: >>32569

>>15716344

It can cap framerate fine, I was just showing off. Taking that screenshot was enough to make my computer get pretty loud. Although lately the program has a bug, where the capped framerate caps it twice as much, on some computers. It didn't used to have this bug, but timers are pretty bothersome I guess. I'll fix it eventually. If you notice it set the cvar "r_fps" to twice what you want.

>>15716386

Just wait until I get competent enough with Vulkan to do some real damage.

>>15716567

I didn't realize I had posted on /agdg/. See >>15716576 for a post on /v/.

>>15716901

Yours is probably the only upcoming game I'm hyped for, indie or otherwise.

>>15718659

What is this, bitwise AND logic?

>>15717782

So just implement your own Quicksort function for it.

>>15718698

Hasse diagram solver. Transitive reduction. You input number of projects and their respective values in X indicators and it makes a matrix with their size comparisons. First one is that. Second is the same, squared. Third one is the first minus the second, or M∧!M². Cuts all indirect connections in the graph and orders by covering relation. You can input a number of possible investments and any number of indicators of cost and gains in currency unit and weight them for time and it will tell you the ideal order for the series of them to optimize the gain.

I actually did use booleans and bitwise for the matrices, a first now that I'm on C++. I'm working of drawing the visual diagrams for each step on fucking winforms.

>>15718794

>>15717782

Turns out I fixed it by making a non-retarded way to get relative angles and used Linq's OrderBy.

>>15720588

There's also an open source equivalent, LMMS https://lmms.io/

I don't know how it compare to it, but people seems happy with it.

Managed to make my code work under windows, but after the event…

Hopefully next time I'll have at least one complete stage as demo.

>>15721543

I've tried both FL studio and LMMS and at the time and I preferred the former by far, but that was because I was used to it and had a fast process I couldn't replicate on LMMS.

>>15721562

Do you use custom shaders?

>>15721780

Yes, but not there. The thing never finished, 3 hours in and I just cancelled it

>>15722042

Normally custom or older shaders have problems with GI. See if your unity version has the "progressive" lightmapper, if yes try using it instead of "enlighten". Best way to tweak GI or graphical things in general is to make a small demo scene with like 3 objects and gradually change/add shit to see what exactly is breaking it.

>>15722691

well, that will take forever, will do after we rework the city

>>15717782

Since it seems to work I don't like to ask this, but: What exactly is this supposed to do?

>>15726483

Looks comfy.

>>15726858

Do you use instancing? Might be why. Well even if not, if a blade of grass is one tri you can easily do 10+ million tris on a decent GPU. Fun fact, most "modern" open world AAA games on a PS4 use around 8-10 million tris per frame.

>>15726870

From what I remember of their project:

Sort vertices created by his automatic terrain system so, when they get drawn, they all have the same winding order and thus the same normals for backface culling.

>>15726870

>>15726932

That's correct. I am not the most math and geometry savvy, so i have to use a lot of workaround strategies. I am pretty sure most of the stuff I do is unneeded if you know what you are doing

>>15726916

I am using instancing, but each "GrassTile" only draws about 8k blades of grass, made of two triangles (to be able to be trapezoid when you cut it). Each GrassTile draws however many grass should be drawn in it's 8x8x8 cube. This map have 1165 of these, some as low as 4 grasses on it. Those with zero gets removed.

I am unsure of what are the biggest bottlenecks in instanced rendering, but supposedly, memory bandwidth is a big issue, that's why my grass position is a single uint with it's first 24 bits being the x,y and z relative to it's GrassTile center (meaning a max of 32x32 (current grass patch thickness) grass per 1x1x1 tile).

This unfortunately, means that the biggest blocks I can have is 8x8x8 if I want to keep the thick grass patches.

I could rework the position uint for 12/12/8 and remove the unused 'grass type' that was reserved for the last 8 bits, this would enable me to have 128x128x8 blocks. Since flowers would end up using a different mesh, I might as well go with this, and draw other types of grass in different tiles. Or just use more memory per grass blade

But for now, i will just draw the GrassTiles that are less than 32 units away from me and be done with it until much later in development.

>>15726858

>>15727011

I may be able to help optimize, but need some info.

Do you use compute shaders to build your data once and done, or is it per frame?

If it's per frame, run me through the compute shader stages (thread count per stage, basics of what you're doing with data), and if you transfer from GPU -> CPU, and so on (every frame?).

Vertex shaders for drawing right?

Also, what graphics draw function do you use?

Would be a lot easier if i saw the source code, but I understand that's it's proprietary

>as long as I send them in 1165 batches, the performance is exactly the same.

Better to batch with smaller instance count and more vertices (find the sweet spot of your GPU's max data throughput).

Not to patronize, but, GPUs are meant to process a huge amount of data at once, due to having so many cores/threads, and it's better to throw it all at it at once; than small chunks at a time.

Obviously there is still a maximum throughput of processing power, and data it can processes per tick; however, it's generally astronomically higher than most think (as we're used to CPU limits).

>>15726858

>>15727011

Try drawing grass front-to-back. That way, the fragment shader of the grass that's further away wont run as much due to early depth culling. Of course this requires you to turn off features such as blending and messing with the fragment depth that would prevent early depth culling.

>>15727362

For context, I am using Unity and whatever they let me use

>Do you use compute shaders to build your data once and done, or is it per frame?

It's built once with compute shaders, then saved to disk. Later runs read from dist, but I never bothered checking if that's really a big of a increase in load time or not.

All the data is stored in ComputeBuffers

_position;//uint, stores relative position in 8/8/8/8 as x/y/z/type(unnused)
_bend;//how much it should bend (float, I probably should change to byte)
_bendDirection;//the direction it should bend (float, I probably should change to byte)
_length;//missnomer, actually stores when the grass will be fully grown in game time
normalRotation;//not yet used, will store the ground normal. This being not done yet is why the grass disappear if you run uphill
argsBuffer;//unity stuff that is required for Graphics.DrawMeshInstancedIndirect

>Better to batch with smaller instance count and more vertices

Each blade of grass (instance) have 4 vertices (for now), to make sure it's not the bottleneck

>Also, what graphics draw function do you use?

Graphics.DrawMeshInstancedIndirect

>Would be a lot easier if i saw the source code, but I understand that's it's proprietary

Although it wouldn't personally bother me, I don't want to even get close to risk losing the exclusivity of even one of our games selling points until I sell a whole bunch of copies either. I bet my brother thinks at least a bit similarly. It's a bad strategy no matter how I feel about it, at least for now. So I won't share the whole code, only the relevant bits

>>15727620

>Although it wouldn't personally bother me, I don't want to even get close to risk losing the exclusivity of even one of our games selling points until I sell a whole bunch of copies either.

What's the worst case scenario for how many copies you sell?

>>15727620

>Later runs read from dist

Well if it's only once then it's nbd

>All the data is stored in ComputeBuffers

mainly I wanted to know the size of the buffers per entry (stride size)

16kb stride size, matches cache size of modern GPUs, massive speed increase, yes use padding or multiple compute buffers.

Since the actual compute shaders aren't a performance hit, then my original advice I was thinking of won't really help as it's compute shader based.

Seems this is just a general, CPU -> GPU and GPU -> CPU issue.

The goal then is minimizing the amount of calls (this is the bottleneck, the data traveling from cpu -> mobo -> pci-e bus -> gpu).

As the lag time is implicit overhead per call; thus, fewer calls, less implicit overhead.

Thus, throw everything at the GPU at once; instead of doing ~2k individual calls with ~2k sequences of implicit lag.

>Each blade of grass (instance) have 4 vertices (for now), to make sure it's not the bottleneck

I'm not entirely sure how you're laying out your grass data, like per cell, or something or another.

However, if you have a singular instance it will dramatically reduce the overhead.

All data should be written into a single buffer; it should be done in a single draw call.

e.g. in args buffer, it should be:

 
var argBuffer = new int[4] { 4 * instanceCount, 1, 0, 0 };

So, a single instance, u can reconstruct the vertex index in the vert shader via SV_VertexID

Though, armed with the above knowledge you should be able to figure our a solution that works best for your situation.

>I don't want to even get close to risk losing the exclusivity of even one of our games selling points until I sell a whole bunch of copies either.

No i understand completely, I'm the same way about my code.

>>15727874

>What's the worst case scenario for how many copies you sell?

00000

my livelihood is at stake on the success of the game, I will not take unneeded risks

>>15727983

>I wanted to know the size of the buffers per entry

You mean the size of the buffers per Graphics.DrawMeshInstancedIndirect call?

The instance data is spread over multiple buffers, is this a performance bottleneck?

>16kb stride size, matches cache size of modern GPUs, massive speed increase, yes use padding or multiple compute buffers

You mean to drop all of the instance data into a single buffer (currently would have a stride of 20 bytes)? And keep the total buffer size a multiple of 16kb?

Also, I have a single buffer that stores a whole lot of cached rotations, so I can have my grass use a single byte to represent a rotation. These kinds of "static shader data" should be stored how?

Also, I am pretty sure those ComputeBuffers are always in the GPU memory like a texture would be, so there shouldn't be cpu -> mobo -> pci-e bus -> gpu downtime. That's why I would like to think that there shouldn't be any issues with many draw calls, the data should already be there waiting to be used. The memory bandwidth I cited above was meant to be GPU memory -> GPU cache or something

>>15728118

Math is mostly going to be useful for translating 3D coordinates or for drawing shit. It's pretty important, depending on how complex/hand holding the engine you use is.

For coding, there's a few different paradigms:

>Language: Shit like C# or C++ or Python. Everything else is built on top of one of these, and then gets compiled down into machine code

>Framework: Something like SFML or SDL which is built from a language, and adds additional functionality (eg SFML loads resources into an OpenGL context for you and is very lightweight)

>Engine: Stuff like GameMaker and Unity. It's usually an all-in-one package that can include editors and more

Generally, you'll have better habits if you pick up a strongly-typed, compiled language. Something like Lua or Python might be easy to start with, but you might develop bad habits because of how they let you do anything (but they are often slower).

C is basically useless (it's too low level to be productive, unless you do it as an academic exercise, like Sigmadev and his low level engine), C++ or C# is probably the sweet spot

>>15728139

>00000

I see someone's capable of learning

>>15728118

There are ten million ways to do it. My personal favorite book is SICP because it teaches programming itself rather than a language, but it's not an easy read. Something like Python is a frequent recommendation to beginners. It has lots of books and good libraries you can use for graphics and shit; some people like that sort of visual feedback while learning and you can get started with basic games very early on.

>I'm pretty shit at math, does that matter?

School math isn't really that important, but programming requires a similar kind of logical thinking as proving theorems, so if you have some experience there, that helps. It's not required though.

>>15728162

My brother might or might not think otherwise, we have the same ID

>>15728151

>GameMaker and Unity

unironically the two worst engines available

>>15728219

Sure, but they're the most widely known and beginner friendly. While I wouldn't recommend them, I used them as an example.

>>15728247

>this doesn't particularly matter, as long as there isn't too many sequential calls

I mean this in the context of writing compute buffers of data to the gpu per frame; tho as mentioned you said you don't do this.

>>15728247

>No that particular part doesn't matter

Also this may confuse you, as, the overall size of buffer is implicitly a multiple of 16kb if you have 16kb entries.

To clarify, the reason I said "that particular part doesn't matter" is because it's not the real reason we're doing it (overall memory size).

We're doing it for the reason of the GPU cache line, and it being 16kb size per entry; so it only matters what each entry's size is, which has the overall size being a multiple of 16kb as a consequence not as the main reason for doing so.

ANOTHER clarification

>>15728247

>>15728252

>>15728262

What is an "entry"?

>>15729326

entry = array[index];

>>15731007

dude, each blade of grass is 20 bytes.

I have no idea how to turn it into 16kb.

Although I might figure something out, I remember once being able to "create vertices" out of nothing in the shader, geometry shader I think. but I can't use Unity's surface shader with it.

Alternatively I could attempt to figure something by making the actual mesh be actually 819 blades of grass and index them with uv1 or something, then business as usual

>>15731044

>dude, each blade of grass is 20 bytes.

>I have no idea how to turn it into 16kb

By doing 6400 of them at the same time.

>>15731044

i was on 48hrs no sleep last night, my bad.

I meant 16bytes per entry, so like, 4 floats (4 bytes per). This is for loading memory from cache into the registers in most efficient manner possible so there's fewer cache misses.

GPUs have 16kb L1 cache sizes so my tired brain hopped to that.

>>15731334

so either reduce from 20 to 16, or pad to 32?

>>15731118

>By doing 6400 of them at the same time

16 * 1024 / 20 = 819.2

>>15731370

definitely reduce to 16 per entry.

tho that's not really the biggest deal here and it's just a quick optimization.

The largest optimization is doing fewer batches. The reason being, is that the gpu is generally memory bottlenecked, and not computationally bottlenecked.

Hence the, "throw everything at it at once" line.

>>15734315

> red nailpolish

slut

>>15734585

it's a fox girl, of course she's a slut.

>>15720467

Free software has come a long way in making music. LMMS is still good for doodling around. Ardour is pretty nice from what I've heard for recording, and Musescore is great for writing musical notation (it's up there with commercial software like Finale or Sibelius).

As for learning methods, I can't give much advice since the books I use are in spanish, but the one thing I can recommend is to learn one thing at a time and put it into practice as soon as you learn it so you can make sense of it and also have fun, along with a sense of achievement/progression. For that you should get an instrument if you don't have one. The obvious choice would be a Piano (if you have the money) or electric keyboard (hopefully with MIDI through USB so you can use it with your PC), although Guitar is pretty easy to get into as well. Have fun making stuff up, mate.

>>15734729

Where is his dick?

>>15734729

Everything below the neck looks neat, the face is pretty fucking strange though

>>15734729

lookan good, although the index is almost longer as the ring finger

I need help setting up a dictionary in C# or something that maps one or more states to an index.

Basically I have my Tiles and Actors that have an EntityInfo associated with it. This has the meta/common object information like typeid and name. I have two fields that are common across Actor/Tile instances, which are State and Direction.

What I'd like to do is assign a sprite index to a certain state, eg the player has a particular sprite used when facing left. However the problem is that i have two indexable values and a simple tuple/packed value won't work because in some cases i might only want to worry about the direction and find the sprite without caring about the state, etc

>>15735327

You mean for instance:

You are attacking to the left, therefore you need the "attack left frame 1" sprite?

You probably should make it a hierarchical system, where the sprite you want is named (state+direction+frame), and if not found, defaults to ("walk"+direction+frame)

>>15734729

really good torso and head, but canonball shoulders make it a bit top heavy

>>15734585

>>15734681

>slut

She is a loyal milf that only starts accepting the idea of dating the main character if you find evidence her husband is dead and you are not dating her daughter.

>>15735390

>finds another man after her husband is dead

Slut.

>>15735368

It'd be like "player has a sprite for each direction, defaults to facing down if missing", or "this rock always uses this single image" or "this gem has 3 different subtypes and it needs a different image for each one", that sort of thing. I didn't think about hierarching it