/v/ - HAPPY E3 EVERYONE

>>14901542

you grab one of the libraries that creates a window for you and fetches input and you roll with that

SFML seems to be the one of choice for that

>>14901542

Both Godot and UE4 support C++. How about some

PROGRESS

I'm currently working on breaking down the curriculums for grades 1-3, so I can implement algorithm generators for each grade.

>>14901542

Make a plugin for Godot that adds occlusion culling and stencil buffer shadow volumes.

>>14901542

Time for enginedev

>>14901647

I looked into shadow volumes recently and I'm gonna have to disappoint you anon. Carmack's reverse is patented.

>>14901738

The XOR method isn't, and a patent-free variation of the original is in the Doom 3 source release.

>>14901747

>The XOR method isn't

>Doesn't support intersecting volumes, aka any scene with two objects occluding eachother

Why bother?

>and a patent-free variation of the original is in the Doom 3 source release.

I tried to figure out how that one works, but I couldn't find anything on it. Nor is the source code of that project something that I would consider humanly readable. Do you happen to have any info on that?

I know it's different enough from Carmack's Reverse to avoid patents whilst being similar to it, but I got no clue.

>>14901765

XOR can be useful if you want a uniform darkness for your shadows. You can get a lot of info on the render methods here: http://fabiensanglard.net/doom3/renderer.php

Go to the frontend section and read on from there.

>>14901765

>>14901857

Specifically, the method used in the source release is the depth-pass method. It's slower, but modern cards should be able to handle it no problem.

>>14901863

Then how does it handle the situation where the camera is inside a shadow volume? Because that is kinda the reason why depth-fail/carmacks reverse is used.

>>14901893

From wiki:

>This approach has problems when the eye itself is inside a shadow volume (for example, when the light source moves behind an object). From this point of view, the eye sees the back face of this shadow volume before anything else, and this adds a −1 bias to the entire stencil buffer, effectively inverting the shadows. This can be remedied by adding a "cap" surface to the front of the shadow volume facing the eye, such as at the front clipping plane. There is another situation where the eye may be in the shadow of a volume cast by an object behind the camera, which also has to be capped somehow to prevent a similar problem. In most common implementations, because properly capping for depth-pass can be difficult to accomplish, the depth-fail method (see below) may be licensed for these special situations. Alternatively one can give the stencil buffer a +1 bias for every shadow volume the camera is inside, though doing the detection can be slow.

>>14901647

Why is /v/ obsessed with shadow volumes?

>>14901919

Because they're much sexier than trashy soft shadows and have a tiny memory footprint.

>>14901931

But what about voxel cone traced shadows? Personally I think the look of shadow volumes is wildly overrated, cascaded variance shadow maps looks way better.

>>14901946

I don't like using voxels for much. I'm quite partial to meshless geometry.

>>14901917

It has been years and I still don't know what game that is.

What do you guys know about dwarf fortress(or other roguelikes) way of randomly generating conversations with NPCs that is not vid related ? I have a few ideas myself, but they all come down to having 10k lines of pre written text where some things in the sentences can be slightly modified to fit the character's personality.

>>14901965

i think its called coropata for DS

>>14901914

I was mostly curious as to how Doom 3 did it. According to ya boi it isn't far off from Carmack's reverse: https://twitter.com/ID_AA_Carmack/status/137189212519792640

That leaves my guess to be similar to the cap that's mentioned in your pasta, but I honestly just wanted some actual confirmation about what exact technique was used.

>>14901919

They're exact, unlike shadow maps.

That said, they have plenty of issues and aren't as fast as shadow maps, so it really isn't that good of a technique. You should only bother if you want sharp shadows.

>>14901965

See >>14901969, it's basically The Incredible Machine (DOS game) with a blend of weeb.

>>14901973

>>14901969

Thanks, I didn't expect to get the actual title of the game.

>>14901973

I'm fairly sure it uses depth-pass with +1 bias method. Doing it with caps could potentially leave an edge case somewhere, and I imagine Carmack wouldn't be too happy with the code in that state.

>>14901982

Except that requires you to detect whether the camera is inside of a shadow volume which isn't trivial either.

>>14901988

Yeah, that's true. Maybe he found some way to set caps perfectly every time. I've tried the retail version and the source release on the same potato, and there's hardly any speed difference.

https://blogs.microsoft.com/blog/2018/06/04/microsoft-github-empowering-developers/

You don't have your code on Github, do you anon?

>>14902605

Press F to pay respects

>>14902250

>Just how dynamic do you want your conversations to be?

When my character is going to recruit a mercenary in a tavern or speak to an NPC, I want the NPC to have weapons, food, drinks, armors, animals, cities, towns, monsters that he likes, dislikes or scares him, the ability to form a friendship or become a lover, thereby unlocking the possiblity of him telling you more about the place he comes from, his family, what motivates him. I want him to be able to comment about the things around him/things that happens to him or you and do it it a way that's not repetitive and I want it to be expressed in different speech styles, lower class, higher class, "no fucks given class", whatever reflects his personality the best. And I don't think it can be done with a few variables and this :

std::cout << "Hey " << char_name << " this " << item_name << " is " << opinion_on_the_subject << std::endl;

And yes, I'd like to keep it simple and elegant, at least as much as it can be.

>>14902605

I do but I need a resume to get a job so I'm going to have to put up with the faggotry until the expected embrace, extend and extinguish happens.

>>14901542

Source engine and UE4, anon.

>>14902605

Already moved my game to gitlab. Github has been going to shit for a while now particularly since they're openly sjws.

http://archive.is/mPcxS

>>14902878

what does gitlab do better?

>>14902883

They're not political.

>>14902884

besides that, i mean

>>14902883

Besides >>14902884, they also have more features, including kanban boards, unlimited users for free, it's open source and therefore can be self hosted, and lots of other goodies.

https://about.gitlab.com/pricing/gitlab-com/feature-comparison

Really though, the biggest thing is that they won't censor you. (unless you're actively harassing people https://about.gitlab.com/terms/#gitlab_com)

They don't make any commitment to be inclusive or other such nonsense.

>>14902900

Private repos are also free

>>14902926

is it normal for gitlab to take hours to load?

>>14902733

For fucks sake…

GITHUB AND GIT IS NOT THE SAME THING

Git is a version control system. Github is a host whom provides git project hosting, amongst other things. You can use git without ever involving yourself with github. Here's how simple it is:

>cd /path/to/project/root

>git init

>git add -A

>git commit

>Type a commit message into the editor and save

There, now repeat the add+commit every time you've made progress.

I have been thinking, to accelerate any game project I may have that is not sex themed, maybe it would be interesting to use the art direction MGS1 had, that is, sprites with less than 500 polygons and medium to low texture detail when it comes to faces, what do you think people? would you play something like "Star Wars Battlefront Zero" with said art style or stuff like that?

The only downside would be not being really limited by what the PS1 had on its time, I could try to imitate a lot but some stuff can't just be done like that.

>>14903259

Low poly and low resolution textures doesn't necessarily make things easier. It can be harder to indicate details with such small textures, for instance, and you still need a good sense of how to model and texture things in 3d.

>>14903322

7.5 billion dollaroos in stock. Which makes it a even more retarded deal.

>>14903285

>lying on the internet

>>14903791

I love that limit color look on the hands.

>>14903413

news.microsoft.com/2018/06/04/microsoft-to-acquire-github-for-7-5-billion/

>>14903903

He's joking, he was the one that pointed it out.

Bitbucket's home page now focuses 100% on how to import your code from Github in a few easy steps. LOL.

>>14903346

i hate this business culture of no plan but to get bought out but this really takes the cake

the only thing github actually made was a trash-tier desktop application that no one used. the real work is all in git which they didn't contribute to at all and now they're making bank on it

>>14902605

Just deleted my account today, kind of stuck wondering where I should host my public projects now.

>>14904070

Sourceforge is doing the same

>>14904107

Either Bitbucket Cloud, Sourceforge, Gitlab, or your own private server.

You can also get a local Bitbucket/Gitlab installation and set it up on something like Amazon EC2, which gives you the most power but requires the most work and money. However, it's also the least likely way to get pozzed.

Alternatively you could just set up a Bitbucket/Gitlab server on your own machine, but then you lose all the redundancies and security that a service like EC2 offers.

>>14904100

They also made the cancer that is Electron, bringing all the bloat of the web to the desktop. I hope those shares become worthless.

>>14904107

gitgud.io

>>14903903

Thanks anon.

>>14904107

I've used Gitgud for public stuff and it's been fine.

>>14904413

We can probably be generous and assume he meant "model" instead of "sprite."

This time.

>>14904535

Breddy cool though houses carved from the inside of a tree are comfiest

>>14904618

>Metal

>not Vulkan

What a bunch of cunts, fuck 'em.

>>14904618

>>14904627

Isn't Vulkan related to DirectX/NVidia shit?

>>14905257

Oh no, it's basically a sequel to OpenGL in spirit, good stuff.

>>14904535

That looks fun but the more i stare at it the worse it gets. That bridge is so disproportionately large compared to those steps and that doorway.

>>14903259

That's what I've been thinking. I wanted to try and create an engine that was focused on physics and destruction, rather than looks. The thing with low-poly though is that you have to go stylised (think super hot). Don't go retro look without retro gameplay otherwise you have no market.

>>14905508

the bridges are supposed to be big, the ropes are at almost head high

Developing physics-heavy, 2D game in C. I'm relatively new to all of this stuff.

Vulkan vs OpenGL.

I feel as if I won't be knowledgeable enough to make use of any advanced features of Vulkan and that openGL has been around enough to have a massive E selection of good tutorials. But, apparently Vulkan is more orientated towards C.

>>14905931

gcc 8

i'm writing my own raytracing kernel that'll use pdacrt, then i'm going to vectorize it for POWER9, since my main workstation is now PPC64LE.

>>14903404

I don't think they want to completely extinguish it. Just cripple its use by normalfags, so the majority of people use Microsoft's software over and open source alternative. Just look at windows. The only reason to use it is because Microsoft cornered the market and gained exclusive support of 90% of user apps (including games), despite their shitty practices that fuck over devs and consumers alike It's kinda sad when people on /v/ think this is a good thing.

>>14906176

Idk man. I don't know Vulkan much, but I don't see how you get much more C oriented than OpenGL.

In Godot 3, how come this works in GDScript…

func _physics_process(delta):
    var collision_info = move_and_collide(direction.normalized() * speed * delta)
    if collision_info:
//do thing

but this doesn't work in C#?

public override void _PhysicsProcess(float delta)
{
KinematicCollision2D CollisionInfo = MoveAndCollide(Direction.Normalized() * Speed * delta);
if (CollisionInfo)
            {
//do thing
            }
}

The if statement here just throws a syntax error.

>>14906524

My assumption is that GDScript treats any non-zero value as "true". Whereas in C#, an if specifically needs a boolean return value (eg CollisionInfo is a number of object, check if it's >0 or null)

>>14906176

You won't be able to utilize Vulkan. I don't think people realize how much slower it is to write Vulkan code and how much you need to multithread before you're ahead of OpenGL.

>>14906524

>>14906567

Shouldn't it tell you that when you try to compile?

>>14907720

yes

>>14907720

Separate render no, device context yes. (this is part of window)

Assuming you are on Windows and using OpenGL:

>CreateWindow returns the HWND (Handle to Window) of the newly created window

>GetDC returns the HDC (Handle to Device Context) of the window

>wglMakeCurrent takes a HDC and HGLRC (Handle to openGL Rendering Context), makes the OpenGL Context current for the current thread and makes the OpenGL Contexts' default render target point to the surface of the specified window

Of course, keep in mind that

>OpenGL API is completely single threaded, so if the windows share an OpenGL context, you can only render to one at a time: no multi-threading allowed!

>OpenGL Contexts' are only valid for one window at a time, so you can't render to multiple windows at once. (However, nothing prevents you from rendering everything to render buffers, and then copying these buffers one by one to each window.)

>>14905508

It's a heavy-duty treehouse, the bridge is for cars

>>14907720

>>14907790

I also want to add that if you want multi-monitor support, you should just make a single big-ass window that spans all the monitors combined, rather than use multiple windows. There's a bunch of reasons why that's better (vsync, windows prioritizing processes/threads of the activve window, no active window switching during gameplay & all the events that come with it, only one window to work with which reduces code & also increases render performance), so only make multiple windows if you actually need multiple windows that you expect a user to move around separately.

>>14906759

That's my opinion too. I looked through some Vulkan documentation and startup tutorials and code. So much more setup stuff to write. More control in the hands of the application developer rather than the driver developer.. but that's more work for no gain if you aren't already a graphics & multithreading expert. Also looks hard to debug. OpenGL drivers (atleast NVidia's) take care of a lot of things in the background for you.

I'd recommend OpenGL 4.5 so that you can use bindless textures and object access over Vulkan at this point. Also Vulkan drivers aren't all that great across all hardware yet, and much harder to debug (not like OpenGL is easy though..)

>>14907790

Good explanation.

As an experiment some time back I used GLFW to open two windows, I created two GL contexts using GLFW and GLEW "experimental" multithreaded mode. Each GL context ran in its own thread. The first GL context had to be attached to the main thread. Worked, no issues. But as you know, can't share textures, objects across contexts. The GL drivers appeared to render only 1 context at a time. The "primary" context seemed to get priority. Neither scene was all that complex so performance was actually lower multithreaded+multicontext than the much simpler and memory-efficient single-context+single-threaded version.

>>14906759

>>14908034

DirectX has always been that extra work, and you should be multithreading stuff anyway if you're not a shit programmer, which you shouldn't be if you're enginefagging in the first place. But don't pretend proper Vulcan is somehow unfeasible unless you're a AAA studio.

>>14908034

>But as you know, can't share textures, objects across contexts.

Actually, wglShareLists exists. But you shouldn't do it since driver devs just slap one big fucking mutex around shared objects.

>The "primary" context seemed to get priority.

Primary as in main thread, or as in whichever window is active?

>Neither scene was all that complex so performance was actually lower multithreaded+multicontext than the much simpler and memory-efficient single-context+single-threaded version.

Windows seems to prioritize the active window above everything, and you really notice it when you start to do things like this. My game runs noticeably worse when I run two clients on the same machine (when doing network testing), but has zero issues otherwise.

>>14908083

You can't multithread rendering in OpenGL and DirectX 11 or lower.

>>14908732

I did not imply so, I said you should be multithreading stuff. Once you know how to do it for one thing you generally know how to do it for anything, barring some specifics.

>>14908083

>multithreading stuff

thanks for the laugh. This little gem buried in a hostile post on an anonymous board. Good work. I do multithread "stuff". Unlike your post, I was getting into specific cases and code paths and my results of multithreading and multicontext experience. Please STFU unless you have something meaningful to say.

>But don't pretend proper Vulcan is somehow unfeasible unless you're a AAA studio

Clumsy strawman. Are you 12? Who mentioned AAA? Who said unfeasible? All you.

>>14908093

Thanks. I hadn't looked into wglSharedLists before. At least it's a way to save GPU memory if nothing else.

>>14909494

What color/size limitations are you following?

>>14909507

As far as color limits I'm not really following any. For size I want the spritesheets to be compatible with rpg maker vxa so either 32x32 or 48x48.

Hope that makes sense, I'm rather new to this.

>>14909536

I asked because it looks like a resized 16x16 sprite. Try drawing the character with the same proportions on paper with all the details you want, and make a real 32x32 sprite based on the drawing. Also check out this topic (it's for general pixel art but it's all applicable to sprites): http://pixeljoint.com/forum/forum_posts.asp?TID=11299

>>14909590

>resized 16x16

For all I know it might be, I made this in Piskel. Is there a better program I should be using? And I'll certainly read through that thread later.

>>14909611

Piskel's alright, but I would recommend Aseprite or GraphicsGale personally.

>>14909617

If possible I'd want it to be as close to what the dimensions for Zelda 1 sprites are as possible, but 16x16 seems as close as I'll get for now. One thing I could do would be to have the air be longer in the side and back sprites.

>>14909630

I'll check those out later, thanks. So for drawing out the sprites on paper I should be using graph paper right?

>>14909655

Just use the graph paper to draw a square, and then draw freely inside of it. Don't worry about lining anything up to the grid; that's what the pixel art phase is for.

>>14909667

Ah, okay. I'd thought I'd try having each graph square be a pixel.

>>14909673

I'm being 'strict' with colors in the sense that each of the game's main areas primarily uses only 1 color for its graphics like in Pokemon Yellow, but when it comes to the characters I wasn't really planning to limit anything. I'd though about having each character use the same pallete as the area they come from but that might lead to issues with them blending in too much. The battle screen will likely lead to more problems but for now I'm focusing on overworld stuff. I'll check out the BS Zelda sprites, they might end up being more like what I'd want than Zelda 1.

>>14905979

>older project

nigga why did you abandon it

>120 posts in 2 days

agdg is kill

>>14911558

Pay my bills for me, and I'll make at least twice as many progress posts as I do now.

>>14911584

>twice as many

>(1)

>>14911591

I promised double progress posts, not double posts. Donate to my patreon.

>>14911558

Just think of all the progress people are making while not posting here!

>>14911668

This honestly. I've been for the last few days working on making a female character base, something which I haven't ever done before. Results are good but I will wait until it's done before I post it.

>>14911558

>(1)

>doesn't post progress

Faggot, you want me to post fucking pics of my scrumboard or something? I've just been fixing bugs, sprint planning and translating, nothing fun for people to look at.

>>14912275

Good luck with the job quest.

>>14912286

Do you need help with anything ?

>>14912286

No, I think I'll do alright, I get past my autism when I talk about programming. I do when I be honest and talk about programming instead of the usual "i'm really trying to get a job" bullshitting. I'm at the last stage of the process so I must have done something right.

sorry for the off-topic posts

>>14912312

Ok, don't forget to come here and tell us if you got the job.

>>14912312

I've been stuck at Walmart for almost 3 years. FT positions have come and gone, but I still have PT work. We're about 17 people short storewide, and there are job postings, but management complains and they refuse to give us more hours. Why are you not willing to give more work to people who are here while we wait for applications? Morons.

Anyways, as stressed as everyone is, years of masochism on /v/ and imageboards in general have galvanized me to where I don't give a fuck and just plug away at my work, and I end up coming off extremely professional for it

>>14914282

What do you plan to do with that song ?

Also, did you use some Hitler speech for it ?

>>14914900

If Die Totenmaske's developers are involved, expect unbelievable jank and games that almost make sense.

>>14915491

Maybe we can help you if you give us some more info. What is the final end purpose for this? Is it going to be some kind of destructible terrain or are you just making a basic shader to give to your artists to paint meshes?

>>14915235

The topology on that back leg is a fucking mess

Does topology matter for static level geometry?

>>14915491

>A face only saves which vertices it is made of, not which edges it is made of.

Maybe you should. Otherwise I guess you can annotate the vertices with the edges they've been used to create, then for each face go back over the edges and remove the ones that do not have at least one vertex that's part of only two edges of that face.

That would solve the case in your pic, anyway, but it might fail for one with concave areas with more than one vertex.

Why do you have concave faces in the first place?

>>14916958

I might be retarded, but triangles on professional models seem to be the result of subdividing while testing animations. If you feel that subdividing a more important area for animation, but not like redoing the topo, or subdividing unnecessary areas, then pushing the created triangles to an area that receives less movement sounds like a good tactic (see the extra edges added to nathans eye that terminate as triangles at the forehead). If you're not directly altering the topo for animation right this moment, it's probably best to try and stick with quads as best as you can. Although I could be talking out of my ass, feel free to ignore.

>>14916958

That's because professionals know exactly how to use triangles along the topology. If you take a closer look, the triangles are on areas where not much deformation ocurrs. This is because triangles are awful for deformations. The reason everything should be quads is because they work best with subdivision and because they can flow with direction which is what you want when you are deforming meshes. That model would be better indeed without those triangles. But they do seem to be placed at pretty good locations to reduce triangle stretching.

>>14916981

>>14916992

Thanks, that's very useful advice. I'll try to use quads as much as possible. I could try to put the triangles more towards the areas which are sure to not deform much.

The face itself seems to be mostly alright but I'll probably change the position and size of the yes slightly.

>>14917339

Nice Squilliam.

>>14909823

>>14914282

breddy good, my dude. So this is the combat stuff? How does it mesh with the gameplay?

I think the muddled sound would fit the grimdark setting.

>>14918233

What do you mean batch draw? Drawing a single vertex array is a single draw call, there's no batching involved. There's something else wrong for sure.

>>14918275

You did.

>>14918265

If i have a large vertexarray and it gets drawn with one call then why does the performance tank so hard?

FPS is already capped at 60.

>>14918360

How large is the array? Maybe it's actually a fill rate problem? I don't know, it's hard to judge without seeing code or knowing what your system specs are.

>>14916488

It is meant to replace making the map in blender, albeit the features that would make it worthwhile are not implemented yet (which is automatically placing props like rocks and trees, grass, etc)

>>14916801

>then for each face go back over the edges and remove the ones that do not have at least one vertex that's part of only two edges of that face

I think that it would solve the issue, I just have to run multiple times, until no edges are removed.

the only time it would not work is if ALL edges have the problem in the pic

>>14918384

Around a million entries, no changes to the array after initial filling

>>14918594

>a million verts

>not polygons, vertices

>not alot

Ok famalam.

>>14918594

It is for your toaster, buddy.

>>14918594

You're still sending them to VRAM each time you draw them, and for that a million is a lot yeah. Sounds like you're trying to draw a tilemap or something, just send only the portion you can actually see.

>>14918608

>>14918614

Alright, how do you draw lots of shit in a performant way agdg wizards?

>>14918628

How about you put some effort into actually explaining what the hell you're trying to do? Supply some fucking code, use cases, specs, performance targets etc? There's about a million ways to optimize your rendering and none are applicable to your case before you actually explain your shit.

>>14918628

I'd write out a post but I agree with >>14918633. Stop being a fucking faggot and explain what you even are trying to do.

>>14918633

>>14918645

nvm, i fixed it :)

>>14918633

Obviously no him, but I'm still struggling trying to figure out a good way to do sprite batching and good culling techniques on sprites

>>14919532

For culling, use quad trees and frustum culling, should be more than enough. For batching I prefer this: http://realtimecollisiondetection.net/blog/?p=86

Oh hey also, this isn't strictly gamedev, but is there anyone around who knows how to deal with 36-character codes?

Eg, my pepsi case has a promotion to send in codes, and I assume it uses 0-9 and A-Z to get a code. Eg a valid unredeemed code is: axy79z3xka. I assume probably two digits are checksums, but I have no idea

>>14904575

>houses carved from the inside of a tree

That's a terrible architectural idea.

>>14920815

cool magical treehouse > gay realism

>>14921925

I have a friend who makes models,i'll give him your info.

>>14918418

>It is meant to replace making the map in blender

Why? Is there something wrong with making terrain in Blender? Why can't you just model terrain in Blender with the better tools and then instead write the a tool that places rocks and whatnot? Also what kind of strange terrain are you going to be making that can have weird polygon fills in the middle and 3 planes be intersecting?

>>14922390

>Blender is kind of dogshit for modelling levels for games in.

It depends what kind of level. If you are making crowded busy tall building cities then yeah it's kind of a hassle but small towns it's not so bad. But the question was terrain. You can make good terrain with Blender's sculpt mode and dyntopo. Then tweak it afterwards.

>>14922454

Sorry, I misread slightly.

It handled terrain decently.

>>14920658

does this mean Banterlord is kill again?

>>14922390

>We're still figuring out what to do with the combat honestly.

so will it be turn based like DTM1 or something more actiony?

>>14921901

fairies be wookies now

If you show me your progress and I'll show you my progress.

Maybe. If I get around to it..

>>14923674

I'll think about writing down some sweet ideas, and reading a book about pure C++ code.

>>14923680

Ideas are like art. Not progress.

>>14915235

>>14915021

What the fuck is this monstrosity. Delete your work and start again.

>>14923420

>Banterlord is kill

i just got it to run cutscenes from timeline 2-3 days ago, but i'm waiting on my artist before doing anything

anyways, gonna start trying to make spaceships for this shit

>>14923674

>progress

Kirby now starts falling when he walks off a platform

Next, i'll implement the "Skater" and "Quick Stopper" customization effects.

Then i'll start working on horizontal aerial movement.

>>14923938

I said show me your progress, not tell me about it.

>>14924228

i do not deem my fan game's current development progress sufficient yet to warrant screenshots and videos.

>>14924547

>most of them seem to be one touch too convoluted

well, simplify them, dont just copy 1:1.

>>14924737

movement in camera is really stiff and janky but the game looks nice

>>14924737

Nice penguin

should i learn opengl or skip to the vulkan meme?

>>14926794

Consider DirectX. It's just much nicer to work with and you get to write HLSL instead of the nightmare that is GLSL. I can't speak about Vulkan though.

>>14926817

What is nicer about HLSL?

>>14924737

>bls give me more attention

>>14926847

Just about everything

>.f notation for implicit floats, difference between halfs, floats and doubles

>min-precision data types like min16float

>smarter casting from float3s to floats and vice versa

>much more builtins and nice-to-haves, like any(float3), min(float3, float3), saturate

>no inconsistencies and weird shit like

<% does not support floats, use mod() instead

>vertex and pixel (fragment) shader in the same file

>multiple shader entry points

>more preprocessor features like [unroll] instead of having to use semi-portable shit like

<#pragma optionNV(unroll all)

There's a reason every large project uses some sort of abstraction language or custom precompiler / transpiler for GLSL.

>>14926794

Learning both is a good idea. Use OpenGL 2.x if you need DX9-level hardware support, GL 3.x if you need DX10-level stuff, and Vulkan for both DX11/DX12-tier stuff and kicking the driver's "cleverness" to the curb if you're feeling really confident.

>>14926817

>>14926926

If he uses Vulkan, I'm pretty sure he can compile HLSL to SPIR-V.

>>14926794

>>14926962

What you should know though is that while Vulkan was originally planned to succeed OpenGL, it does not. It's just a different, lower level API, OpenGL is still in active development. I never worked with Vulkan but it requires much more groundwork for tasks that are simpler in OpenGL and DirectX. It's also not necessarily faster, a naive port from OpenGL to Vulkan would make your application much slower, only if you actually exploit multithreaded access to VAOs and so forth will Vulkan shine. I wouldn't recommend something like it as the introduction to 3D graphics.

>>14926903

Yeah, they're probably being processed as AABBs or points in the broad phase, using an AABB hierarchy, some form of grid, an R-tree, a quadtrees or whatever else to rapidly cut down collision candidates. Once that's done, there's a much smaller set of potential collisions for the more demanding final testing, possibly using GJK, SAT, fiddling around with Minkowski differences or something of the sorts.

As for what data structure or method they're actually using for the broad phase, I have no idea. There are far too many options to list, and some are more suitable for certain workloads than others. Uniform sizes, varying sizes, varied aspect ratios, lots of static things, lots of dynamic things, very sparse, very densely packed, and the amount of entities as well.

Some of these might see massive improvement when tweaking some parameters or adding custom implementation details.

Why do you ask? Checking everything against everything getting a bit slow?

>>14927128

Oh no, just something to mull over until I invariably try to make my own shitty clone and give up after ~ 2 weeks

>>14928062

Cool. How much did it cost?

>>14928084

150 bucks.

>>14928090

What model is it, my dude?

>>14928098

Alesis v49

>>14928110

Not sure why they would build it with backlight but that's pretty cool. I didn't know they made keyboard+drum boxes.

>set up comfy Void Linux dev environment

>focus on terminal applications over graphical crap

>GTK+ and Qt applications have overly large fonts and buttons

>search around and can't find a solution

>realize I somehow overlooked the font size field in lxappearance

I may be retarded but at least my dev environment is comfy and mostly free of Poetteringware.

Finally figured out how to do spherecasts in Godot, so now I have non-fucked camera collision. It's rather convoluted, so I feel compelled to share the code for any other Godot devs.

onready var _cam = $Camera
onready var _player = get_tree().get_nodes_in_group("Player")[0]

const BASE_DIST = 12

var cam_shape = ResourceLoader.load("res://Player/Cam.shape") # a SphereShape with radius 1

func _physics_process(dt):
	new_dist = BASE_DIST
	new_dist = clamp(BASE_DIST + _player.velocity.length()/2, BASE_DIST, BASE_DIST*3)
	_cam.translation.z = lerp(_cam.translation.z, new_dist, dt)

	# check for cam collision
	var space_state = get_world().direct_space_state
	var query = PhysicsShapeQueryParameters.new()
	query.set_transform(self.transform)
	query.set_shape(cam_shape)
	query.exclude = [_player]
	var result = space_state.cast_motion(query, self.transform.basis.z * (new_dist))
	
	if result:
		_cam.translation.z = result[1] * new_dist

>>14928280

How lightweight is Void? I've been considering installing Linux on a laptop I have which is currently running W10, and I want something that's light in both performance and disk space. It's an "ultrabook" which means it has a decent CPU but a tiny 128 GB SSD for storage.

>>14928755

Fairly light: the default live image with no X11 or anything requires 350MB of disk space and 96MB of RAM, while DistroWatch says the MATE live image is 2GB when installed and takes up 240MB of RAM. Thing is, Void Linux is kind of like a saner Arch Linux so it's not a great choice for newer Linuxfags, but if you know your way around the command line, check the manpages regularly, and know what programs/libraries/daemons you want it's pretty fucking comfy.

If the hardware supports it and you can live without shit like Steam or most proprietary vidya, I'd probably install OpenBSD on a laptop.

>>14929799

use assimp and never look back

literally every model format is completely different and there is no widely adopted standard because everybody puts custom shit into their model format

>>14930060

>3rd party library

okay… if you say so, boss… :(

💖

>>14930104

The problem is that you speak of a "common 3d object file", but no such thing exists.

As such, using an importer library is the closest thing you'll get.

But ASSIMP does allow you to just read the vectors you want. You'll probably have to apply a few transformations to them to get them in the space you want though. http://ogldev.atspace.co.uk/www/tutorial22/tutorial22.html

>>14930122

do the renders come out fine?

>>14930164

nevermind, fixed it

and i already forgot which random setting it was, i think i changed the texture channel to combined textures

>>14930450

Nice, but I'd add some footstep sound effects for better immersion.

>>14930570

of course, i'm just getting the basics in

>>14931294

Not bad. How far are you taking this? Will you add obstructions to the map?

>>14931363

not gonna touch the gameplay anymore right now since that's all the design document required

+ i can't actually make a game with star trek ships and not get my ass lawyered

just want to finish the models and i'm gonna send it

>>14931365

I'm sorry I keep forgetting to spoiler this shit

>>14931365

your boob topology doesn't seem right if the nipple is misplacing itself

>>14931378

I didn't make the model, it's downloaded. I'm trying to make the boobs bigger by using blender sculpting.

It just kind of happens, I can't seem to fix it

>>14931668

can't say I like the screech

I imagined it would be more bones clacking on stone with the wheelbarrow creaking and rattling

>>14931735

Yeah I know it's goofy, I just overlaid the daggerfall skeleton sound for kicks.

How can you determine a client computer's maximum texture resolution?

Again, using Factorio as an example, the HD mode breaks all the high res art into several sprite atlases. It ends up creating approximately 8-9 16384x16384 atlases, for example

>>14932111

Have you tried querying the graphics card like a normal person? https://www.khronos.org/registry/OpenGL-Refpages/gl4/html/glGet.xhtml

aaand done

download link for anyone who wants to try

https://doctard.itch.io/laser-chess

time to abandon it

>>14932111

GLint size = 0;

glGetIntegerv( GL_MAX_TEXTURE_SIZE, &size );

>>14932123

No, because I went from XNA to SFML and haven't touched OpenGL or the underlying stuff at all

>>14934411

sin (alpha/2) = (a/2)/h

area = (1/2) * b * h

If you divide your triangle in two down the middle, you can use the sin() of the resulting angle to figure out the base of your half triangle. Then you just double that and you have the base of the original triangle. Then when you have the base, it's just area = b * h * (1/2)

>>14935060

Fuck me, >>14934870 is right, it's the tangent, not the sin.

>>14935089

yes its tan not sin but

>>14934870

is wrong in that it isnt tan(α/2)=a/h, its =(a/2h) like you have

>>14935999

Why do I keep getting 403: Forbidden with webms?

>>14935999

I haven't tried multi-threaded streaming assets yet. I was considering sending messages to a single thread to load assets. When an asset finishes it would message the main loop and mark the asset as ready. Just have the disk access thread work through it's task list until finished. Are you are creating a thread for each asset? I think all 3 major OSes have significant overhead creating threads, and you might want to use just one for disk access reasons anyway.

I haven't noticed a performance issue with the Nvidia driver's glsl compiler. I could see needing an optimizer for each piece of hardware though as some of AMD drivers are still pretty bad with glsl.

>>14937058

>you might want to use just one for disk access reasons anyway.

Very good idea, yes, right now i have one thread per asset.

>haven't noticed a performance issue

I think it's not so much about performance issues, it's just that you can squeeze out a little more with certain optimizations.

>>14936656

Only for you anon:

https://my.mixtape.moe/yynhyr.webm

>>14928839

got any idea on how the glider/skateboard should look?

>>14937676

The skateboard should be a gun

>>14937676

The glider is going to be a big hat like the man in the yellow hat wears in curious george. The board I didn't have a specific idea for, just that its a primitive piece of technology, like a wooden snowboard.

>>14901542

Every other language in the C tree is easier so just jump in to Java/C#/whatever.

>>14937420

Thx. Don't know much about async loading but it looks like you've nailed it.

>>14939878

Pretty laggy if you ask me, but thanks. I implemented the single worker thread that >>14937058 suggested now, it already runs much nicer, also meshes now stream in just like materials. Next steps would be a lock free queue for the worker thread, im just using a std::queue with a mutex right now, and to benchmark if using more than one worker threads improves performance.

https://godotengine.org/article/how-actually-make-your-dream-game

based.

>>14937420

>I think it's not so much about performance issues, it's just that you can squeeze out a little more with certain optimizations.

I need to look into this. My parallax bump shader has poor performance and it ends up on a large % of the screen pixels. I'll take whatever gains I can get with it at this point. I'm about at the stage of cutting out features and moving towards a more "faked" effect.

>>14940041

>also meshes now stream in just like materials. Next steps would be a lock free queue for the worker thread

That's really cool. I'm glad your threaded streaming is working out. your lock free queue system seems like the ideal setup.

>>14940393

related post

https://kircode.com/post/how-i-wrote-my-own-3d-game-engine-and-shipped-a-game-with-it-in-20-months

>>14941136

nvm, i forgot to glFinish() the streaming before swapping out placeholders

>>14941156

>I need to look into this.

This is the GLSL optimizer Unity uses, might wanna have a look:

https://github.com/aras-p/glsl-optimizer

>My parallax bump shader has poor performance

Care to post the code? Does it contain loops or multiple texture sampling?

>>14941251

why wouldn't you make an engine for a generic 3D unity asset flip?

>>14941251

>A timestep system ensures that an engine displays frames at specific intervals. Unlike most current game engines, YUME does not have a framerate cap, i.e. it's not tied to 30 or 60 frames per second. In fact, the framerate can be anything from 1 to 250 frames per second, and the game will still appear to be running at the same speed no matter what the framerate is (only the screen will be updated at different intervals). This is achieved by separating game logic from rendering. Rendering a single frame can take different amount of time on different computers, so the render framerate depends on the program's performance. The logic cycle, on the other hand, is tied to a 64 tick rate - meaning the game logic is updated 64 times per second, no matter what the render framerate is. There are a few edge cases that the timestep system must account for, but the principle remains the same. As a result, the framerate decreases on weaker computers without ruining the game flow.

Speebotdevs work is very impressive, but this is paragraph is pretty jarring. What current game engine has a locked framerate, and game logic that is tied to framerate? If there even is one, it's probably some obscure japanese legacy engine used exclusively for PS Vita JRPGs. Certainly not a good reason to develop your own engine.

>>14941359

most of his work seems like very generic unity platformer.

>>14941401

He's a very rare yesdev, even if its the game is not technically impressive, the fact that he went through with it from concept to release on steam, over 20 consecutive months, and with a custom engine, is impressive.

>>14941299

>>14941359

TBH I originally started making the engine as a learning experience. It was my first encounter with OpenGL, 3D math, shaders and all that stuff. I wanted to learn and understand the underlying basics to later use that knowledge with Unity. Turns out that the complexity of writing a custom engine is over-exaggerated, so by the time I had a functional renderer I just decided to roll with it. This also let me shape my own workflow, which is important to me.

Speebot was an experiment to complete a proper polished game mostly from scratch. My current project is more ambitious, and is using the same engine.

If you're starting out and a platformer is your end-game then you should probably use an existing engine.

>>14942031

I'm consistently impressed with how good you are at animation and programming, you're a waifu one man army.

>>14942031

i wonder what the creator of Shinobu thinks of your work

I saw a roll image awhile back with a bunch of game ideas with post numbers tied to them. Would someone mind posting it for me? I don't know what to do with my life right now so I want dumb luck to decide for me.

>>14942031

Those animations are incredibly adorable.

>>14941838

>i'll definitely abandon it this time

That's too bad. It looks like it could actually be a decent mobileshit game worth downloading. I'd play that on my way to work.

>>14942609

>1: Character action game in the Taimanin Asagi setting.

>2: Overhead puzzle game with block grabbing mechanics

>3: Spiritual successor to Space Harrier/Panzer Dragoon

>4: First person dungeon crawl

>5: Isometric strategy RPG with a focus on exploration instead of battles

>6: Vertical scrolling bullet hell

>7: Dating sim (SLG, not VN)

>8: Spiritual successor to Mechwarrior Mercenaries/Armored Core

>9: Crime game set in a comic book capeshit world

>0: I Wanna Be The Guy fangame

>Dubs: HEAVEN OR HELL: Pick the one matching the number and one other and combine them

>>14942031

I was just driving by the thread, and was surprised to see the animation that well done.

What exactly are you doing that makes it so much better than most projects I see?

Is it the animation speed?

>>14942695

The whole body's involved and coordinated, and there's subtle anticipatory motion and follow-through to most movements. A bit of bounce after overshooting the final destination and then coming back. Those are what sell it. It's not perfect, but considering how many artists fail miserably at it, it looks damned good.

Often you see something like parts of the body remaining oddly static or acting completely independently, or you have that particularly disconcerting situation where actions that should come after one another happen at the same time instead. Like a lazily thrown fist not putting an opposite force on whatever it's connected to during acceleration, or not tugging the body along after the arm joints are at their straightest, but instead all involved pieces simultaneously moving toward their final destinations and reaching there at roughly the same time.

The general speed, pacing and easing are good, too.

Being cute and expressive in the reactions helps, too.

At least that's what I see. It's not that much different from the qualities you'd want in traditional animation. Maybe he can give some better insights himself.

>>14942926

Progress is for the weak. A real aggy daggy would dev every day, never making an inch of progress towards their dream game.

>>14942926

>Fireworks

Holy shit that takes me back. I used to do all my artfag shit in Fireworks before migrating to Photoshop ~12 years ago. What happens to the item you craft based on the ranges in the bar? Do you get a shittier version? No progress as I'm getting distracted by side projects and spring cleaning.

>>14943070

I think it's more about making progress that constantly pushes you to work in circles.

>>14943092

Fireworks is #1 pixeldev software. The bar will scale based on your smithing skill and crafting difficulty. The bar's speed/scale will scale with that meaning that it's harder to create good weapons/armor if you're not skilled. You can repair shitty weapons later at the cost of more materials.

>>14943070

Gotta optimize this implementation of linked lists for the credits screen.

>>14944233

>guts man's ass

now that is a vintage meme

>>14944233

Fugg yeah turrican.

sometimes I think that with all the shit that the industry is making, all the bad practices, all the generic gameplay, I think any fucking indie game that is barelly functional, have interesting new gameplay, don't look like crap is like a masterpiece for games.

>>14944783

Damn that filter really puts a tax on compression, huh? Gameplay idea seems solid but it definitely needs sounds.

>>14945169

yea I should of turned it off

>>14942695

It depends on which animation you are talking about. Some are mostly procedural and others are mostly regular static animations. When she reaches out with her arms to tap your hand, that animation is done procedurally with some realtime IK. I animated a little "bump" on her hand as it reaches the end. Also the whole body moves forward with a blend tree. Then it has a layer of "look at" logic with random blinks generated for her eyes, spine, and head which is sometimes turned off depending on the animation. The look at target changes with every animation as well.

For the static animations I try to animate all parts of the body. There usually isn't a single part of the body that doesn't move for anything. Also I don't sync up bones. Syncing makes stuff look robotic and inorganic.

>>14941359

>What current game engine has a locked framerate, and game logic that is tied to framerate?

You'd be surprised. Dark Souls 1 engine had the network packet loop tied to the framerate.

>>14941359

>>14945893

Didn't Fallout 4 have actual loading times tied to the framerate, at least for a while? Might have been some weird shader issue.

>>14942031

You are the once in a decade nodev that could actually land a decent job in the industry if he wanted to and spoke japanese

>>14945893

>Dark Souls 1 engine had the network packet loop tied to the framerate.

That's because From is one of those obscure japanese developers who don't know their shit. The actual engine, PhyreEngine, supports everything he talked about. It's a ubiquitous and trivially implementable feature

>>14949771

He apparently got a job as a technical artist by showing off his marriage.

>>14949771

>You are the once in a decade nodev that could actually land a decent job in the industry if he wanted to and spoke japanese

He got a job as lead technical artist you dumdum

>>14949975

>>14950007

Oh boy. is his current progress part of the new job?

>>14950226

No, I saw a webm of it last thread, it's some other game.

>>14949771

>That's because From is one of those obscure japanese developers who don't know their shit.

Have you played their PS1 and PS2 games? It's a study in just doing shit and putting it out. I celebrate their early methods even though the results are technically unsound. I'd rather have interesting ideas in trash games than the empty, professional, polished and committee approved kike propoganda at $60 + DLC. Game logic and physics chained to framerate is far less of a creative sin.

Anyone got resources on writing damage equations for RPGs and the like? I'm not entirely sure if I'm making mine interesting enough and wanted to see how other RPGs did it.

>>14950420

I don't think he's saying they're a bad dev, the point he's making is that on the technical side they don't really know what they're doing.

>>14950420

Im not saying they are bad developers, they made some great games. For some reason it's prevalent in japan to tie everything to the framerate, it's not exactly a unique Fromsoft quirk.

>>14950743

Perhaps it's a relic from bygone times, back in the Win95 days where the OS was a little more obtrusive with inserting random delays to do its thing. Dojinsoft has almost always done that. They like their integers, too. Working on ticks and integers keeps everything more predictable as well. In some 2D games, I could see slowdown being preferable to dropped frames.

That's a habit that doesn't translate all that well to 3D, in my opinion.

>>14951079

Honestly as dumb as it is, i kind of like the fact that in DS1 whenever the game stutters for a bit, the logic never jumps ahead to catch up with the time passed, the next frame will be displayed exactly as if the game just paused

>>14951331

Yeah, I guess it could still work. I'm not entirely sure how DS1 handles inputs when it's stuttering. Does it just queue them up or drop them entirely if something was both pressed and released during the lag spike?

I was thinking the kind of hell it'd wreck on something like a first-person shooter if it ate mouse movements. You want to flick a rail at that guy on the right edge of the screen? Fuck you, you shoot the wall in front of your face instead. And because weapon switch and firing arrived on the same frame, shooting got given priority, you shot a rocket instead and blew yourself up. And even if you had shot in just the right place, he wouldn't have been there yet.

>>14951832

i remember quitting street fighter 4 because for some reason multiplayer slowed down time for both of us because i was running at around 40 fps, even though singleplayer ran at the same framerate but at normal speed

Replaying a stalker has given me an extremely strong urge to mod it instead of working on my own game.

>>14950743

OK, that's pretty much how I see them too. I have a little sympathy for From and the framerate-locking of game logic. I tried it both ways myself. Once I separated the game logic (and physics) from framerate I had to start interpolating object transforms between render frames. Also I had to run Bullet3D in multi-step mode with variable timesteps. Getting rid of the jittering and other issues this caused was extremely frustrating. I never got the Bullet3D syncing issues completely solved. Debugging was a nightmare - Bullet was also running in a separate thread. I absolutely understand the decision to lock to framerate. Then throw networking into the mix. The right way to do it is to unlock everything and interpolate but it does take a little more processing power and it's a real pain in the ass.

I'm trying to learn javascript and I have seen some shit. I don't know anything about this but it seems like everyone is fat, flat out retarded, and blatantly contradicting themselves. I hate everyone who uses JS.

>>14960639

>I'm trying to learn javascript

>I hate everyone who uses JS

Are you trying to tell us something?

>>14960639

learn C instead

>>14960846

I want to learn an actually useful language, though.

>>14960849

Which is why anon told you to learn C

>>14960849

C is the most useful language you idort. You have basically every API at your fingertips, you don't need tookits and you can just fucking interface with the operating system directly. C++ is the most widely used language in the industry and even professionals write it in C++ that is basically C. C is just C++ that's easier to use, and doesn't have C++'s flaws.

>>14959209

>Bullet in variable timestep

Pretty sure Bullet and most physics engines weren't meant to do this. Usually you're supposed to run physics at a fixed timestep and use interpolation to match the framerate.

>>14961581

>/strek/ memes

Best board.

You should highlight enemy firing patterns when you mouse over them.

So how exactly to people develop openGL?

It was a bitch setting up GLEW and GLFW with MinGW.

>>14961622

At least tell me you're using Mingw-w64.

>>14961635

of coursh.

>>14961622

I use OpenGL and GLU. For the windowing system I use the native one for that platform (win32 and xlib). I don't use GLEW or GLFW. Since all of the libraries that I was using seem to come with the compilers I was using, I didn't need to download any new libraries to do this.

test

>>14961597

i considered it but it wouldn't work on mobile since that's only one touch controls

i could make it work if you click on an enemy and it shows his possible attacks, but then it just gets confusing since it's identical to your attack previews

tbh i just want to not do work

>>14962945

>>14962945

>>14962945

>>14962945

>>14962945

>>14962945

new bread