/agdg/ - Amateur Game Development General

no-one else bothered edition

> Play Demo Day 14
itch.io/jam/agdg-demo-day-14

> Current Monster Jam
itch.io/jam/agdg-monster-jam

> Helpful links
Website: tools.aggydaggy.com
New Threads: Archive: boards.fireden.net/vg/search/subject/agdg
AGDG Logo: pastebin.com/iafqz627

Previous Thread: Previous Demo Days: pastebin.com/rmiZV5yX
Previous Jams: pastebin.com/LKEdLxdG

> Engines
Construct 2: scirra.com/construct2
GameMaker: yoyogames.com/gamemaker
Godot: godotengine.org
LÖVE: love2d.org
UE4: unrealengine.com
Unity: unity3d.com

> Models/art/textures/sprites
opengameart.org
blender-models.com

> Free audio
freesound.org/browse
freemusicarchive.org
incompetech.com/music
fantasymusica.org

Other urls found in this thread:

youtube.com/watch?v=nhhmFltg82g
i.4cdn.org/wsg/1496599261478.webm
youtube.com/watch?v=ovaGOvdYCIk
strawpoll.me/13112703
youtube.com/watch?v=w3P2QURwZx8
twitter.com/NSFWRedditGif

godot

Friendly reminder your game will never be this good

youtube.com/watch?v=nhhmFltg82g

2nd for being a functional human being

Second for Rust

Nice try

How do we fix VR?

Is the problem that most of the developers who could buy a set are rich incompetent San-Fran types?

...

We make a game and it is a _________good___________.

define _________good___________

What's a good algorithm for pathfinding in empty 3D space, like a swarm of drones around a carrier ship or something, such that they wouldn't collide but without a bajillion linecasts?

It's fun

trying something simple for easy animating

...

Monster jam is almost over
itch.io/jam/agdg-monster-jam

>making a game
>not shitty tutorials
how do you feel when your game will never make 20k$ while tutorials about making games did?

where's the meme tabletop rpg one

CUTE
U
T
E

just saw an unity live course in a near city, I think I'm gonna apply to be a teacher

Two (2) games

hahahahahaha

2017 YEAR OF THE GODOT

soon to be three

Explain

gameplay loop is done - the boss can be defeated, there's an ending "cutscene", and you can die.

Spicing up gameplay and adding more enemy/room variety now. Made it so the shield bumps most enemies and there are hazards that enemies won't walk into, but you can shove them into

w/ audio: i.4cdn.org/wsg/1496599261478.webm

godot is winning, and it's not even the "waiting for 3.0" gm should die at this point being non free and offering less

good progress, can you make pickups sprites be on top of everything? being behind spikes is weird even if i know it's correct in perspective

the missing outline pixels look sloppy, almost like scaling issues tbqh

First for Vulkan.

>tfw you spend a week building a framework to determine attachments used by shaders to build a renderpass to build pipelines to render a triangle.

>tfw you haven't yet rendered a triangle.

Control is great and all, but a lot of this crap is something the driver can do by itself.

are you enginedevving

I'm an absolute beginner in modeling, and following up on my few hours of Udemy-induced Blender experience from last year, I created the layout in pic related.

It made sense doing so, and if I'm not mistaken, the instructed suggested the same thing, to logically group things under empty objects (mostly using plain axes).

I found it quite bothersome and to add and apply mirror modifiers and materials to each mesh. Also, it's a low poly model with only 3 colors (one for the body, armor and beard), so I hate the idea of duplicating materials for each piece of the armor.

I mean I can join the armor so it uses a single material, but I don't want to do something I can't reverse easily.

So, is this the correct way of arranging my objects, or do I just create a flat layout and group things together?

Ye.

Sick to shit of braindead OpenGL backends (hurr durr one device allocation per buffer).

If I'm going to manage my own sub allocations, I'm also ditching as much shitty driver code as possible.

Using libgdx/java.
What's the most efficient way to create a factory that spawns a random entity with a given characteristic? Like "spawn a random level 5 enemy" or "spawn a flying enemy".

post mesh, why do you need to group objects?

THIS is a Godot engine game. Why are you fucking around with Unity and GayMaker again?
youtube.com/watch?v=ovaGOvdYCIk

>another kirby clone puzzler

Isn't OKAM literally the Godot devs?

Nice """proof of concept"""

I want to make an isometric game on Game Maker. How much lube should I buy in advance?

I'm learning some OpenGL. Is my rendering performance good for the amount of objects I'm drawing? Any tips on how to increase performance if not?

20k triangles all drawing the same thing at 25 FPS seems really, really low

I'm no expert but that seems pretty slow.

The second protagonist (the marshal) is basically furry Raiden and the color grading/cybernetics in the art are all really designed to be kind of homages to the MGS style. It's less "Oh I am just knocking this thing off" and more "the entire reason I got into doing art was the Ashley Wood comics." Pretty flattered though thank u
If you can A: pay me or B: guarantee me that the project will likely see release and I can put it on my resume then sure. I honestly don't care too much about money versus having a shipped game with my name on it.

I haven't really refined the design for the second protagonist and I gotta redraw him more in the same style as the detective.

kys

Working on something.

This is why we advise against making your own engine

Very low FPS for a few hundred quads.

Though, it depends on how you are measuring FPS - is it based on time taken to build rendering commands (glBindBuffer etc...), or is it based on time spend waiting on glFlush or similar?

On Windows you will have a min. 8ms scheduler delay for SwapBuffers, and possibly a few hundred ms waiting for glFlush or glFinish.

Try instanced rendering if you aren't already.

If you are using instancing, then you are probably rendering on a smartwatch.

LIGHTNING NEEDS TO BE REBUILT

welcome to ue4
although holy shit 3k objects is a lot

k

strawpoll.me/13112703

>strawpoll.me/13112703
havoc

My draw method is just, for each object on screen:
>bind shader
>bind texture
>Apply transformations
>pass transformations to shader
>Bind VAO
>DrawElements

Binding the same shader and texture multiple times might be wasteful, but I don't know yet how I'd do a "bind if not already bound" sort of check.

>is it based on time taken to build rendering commands (glBindBuffer etc...), or is it based on time spend waiting on glFlush or similar?
I just take the time elapsed, increase a counter on each loop, and when 1 second has passed, I update the window title, reset the counter and repeat the process.

>tfw I can't answer this without googling both

never ever too late for more stuff.

bullet is way more stable

physx is unstable garbage and I'm not sure why Unity and Unreal use it

because it's fast

>having the choice between letting the driver do everything by itself using OpenGL and doing everything by oneself with Vulkan
>choose Vulkan
>complain on AGDG about having do to everything by oneself

havoc costs money, bullet and physx don't

>unironically enginedevving
Now here's a meme I will never understand.

only set the shader and texture when they change, never ever in any other situation

>physx is unstable
>can properly simulate projectiles moving at near light speed

lmao ok

Play KSP and you will understand why PhysX sucks.

I think the only excuse is if someone want to practice is programming. That's it, if you really want to make game you can just skip to the good part right away, there's enough work already especially for a one-man-team.

I can understand if its not a one-man-team and is something that is worked on in full time, but then its not "AGDG" since its no longer amateur.

I started integrating bullet in my engine, and it works. The problem is it's very poorly documentated and I have to look into the implementation nearly every time I need to do something new (the implementation is pretty dirty, lots of uncommented code, not implemented or TODO). Physx has very good documentation (like you would expect from nVidia), but looks way more intimidating to implement. Still can't decide if I wanna switch.

youtube.com/watch?v=w3P2QURwZx8

I took the binding code out of my sprite class and called it once before starting the game loop. VAO binding remained in it. It's a 10 fps increase, but still slow I assume, so I think I have to move the VAO out as well?

I bet sorting objects by shader/texture/VAO would speed things up, but that sounds like a major pain in the ass, too.

didn't physx start as a fast nvidia physics engine for a billion particles
why the fuck are people using it for game logic

Monster Jam ended, and it was a big failure proving that Loli Jam didn't fail because of its theme
I couldn't enter it at all, but I remember that like 25 people joined it. So how come there are only 2 games?
If you're one of those 23, what happened? Why did you decide to give up on it?

because the physics engine (no matter which one) is also the collision world in most cases.it would be pretty nonsense duplicating every object in your own system and the physics engine

obviously his code related since entering and exiting a vehicle changes the physics behaviour, most likely his character's values aren't properly set until after he leaves a vehicle

yeah which means this is a TOTALLY WHITE game made by 100% PURE SNOW WHITE TRUE TO LIFE SILVER HAIRED ARGENTINIANS.

Reposting progress just so that I have a place to host the .webm

I'll spend today juicing this up and making it pretty

oh gosh hello there, I loved your game. Glad to see it's still coming along

Oh shit you're alive?

oh shit, I loved your game. City of Pitch or something, right?

Oh oh oh!

>tfw the argie devs will make mad dosh with godot games and will be able to buy the malvinas back

Pitch is the name of the game, but the 'full" title is Tales of the Ecclesiarchy: Pitch

Thank you! I'm alive and trying to work on it more, since I have a full time job that eats all of my free time. Between four moves over the course of a year, several different jobs, and relationship stuff, finding dev time is difficult but I'm finally able to get back into it.

May gamedev be the compass that leads you through troubled waters

Does the absence of answer mean that I will be ok?

Simple-minded argentfags cannot into taking advantage. The fools are doing it for FREE, and will end up dissolute and forgotten, while Godot reigns as the superior engine.

I thought that eyeless waifu meme game was supposed to be a mosnter jam submission

the only difference between isometric and other views is how you move the objects, the rendering is the same: back to front, left to right

>you were not part of the team that worked on bravely default/second

I wanted to use Game Maker's drag and drop for simplicity and progress and now I can't find how to display a simple value in a list.

...

Don't use game maker's drag and drop

It's not good for learning, just write the code yourself. GML is already one of the easiest programming languages to ever exist, despite how robust it is.

>progress
That's not something DnD provides.
Just forget it exists, it's completely useless.

The game is finally starting to feel cohesive.

Got the enemies fov done, depending from what side they approach you, depends if they get bonus in combat.

Working on the menus for when out of combat, you can cast spells and use items now, will add changing equipment and overview of your party next.

I'm done with """"learning"""", I want progress.

Learning begets progress

You must learn before you can do, there is no easy way around it.

If you think DnD is worth it then you learned absolutely nothing.

Hahahahahahahaha
Good luck with that """"progress"""" when you are having trouble displaying a value in a list

But good on you for not being utterly ashamed at asking this

—said no dev ever

>team
Thank god.

>GML
>robust

shader in the outermost loop - changing it implies a pipeline recompilation.

textures & uniforms can probably be interchanged, as those correspond to descriptor updates.

OpenGL usually just has one descriptor set - changing one uniform means changing one value in an "all uniforms" descriptor set, the entirety of which - including values that didn't change - is then sent to the device.

Ass-pains are a fundamental part of graphics. The trade-offs are marginal improvements in speed for every 10ksloc of BS you add to deal with scheduling and resource management.

I can guarantee you will need a at least 30ksloc for a minimally functional OpenGL-based renderer.

GMS vs your own engine - simplicity
GM DnD vs code - just no

Looks like to draw 10k objects you make 10k glDrawArrays calls. That's what a crappy renderer (like sdl2) would do, but you can do better by making one glDrawArrays to draw all sprites that share a texture (sprite batching).

jam spam made the jams uninteresting