scissorman posted:

I'm planning on using block-based terrain, with each block having individual properties, so I don't have any empty/homogenous space for quadtrees to take advantage of.

Orzo posted:

And your blocks are all the same size and don't move? If so, you already have an implicit structure within the blocks themselves. You can just store them as a 2D array (or array-of-arrays) when you load them, and then you can index into that array to figure out which blocks are visible based on your camera.

scissorman posted:

Do you mean that I should assemble the vertices etc. every frame based on what's visible?
Wouldn't it be faster to have prepared vertex buffers for each chunk and just throw those at the graphics card and let it cull those that aren't on screen?

scissorman posted:

Do you mean that I should assemble the vertices etc. every frame based on what's visible?
Wouldn't it be faster to have prepared vertex buffers for each chunk and just throw those at the graphics card and let it cull those that aren't on screen?

Orzo posted:

Oh, I missed the part about how you're using chunks. Yeah your approach will probably work out nicely. By 'prepared' vertex buffers, that's the same as static vertex buffers right (sorry, I only know DirectX, and that's the term they use for vertex buffers you create not-every-frame)?

Suspicious Dish posted:

BGM1-004.mp3 is your favorite song too? I can't get it out of my head!

Shalinor posted:

There's actually a trick you can pull, if you want to stream the terrain in one chunk at a time but keep it in a single vert buf. Google "toroidal arrays," or just read up on Geometry Clip Maps.

The actual purpose of the algorithm is for huge view distances, and variable vertex density with smooth transitions between. If you ignore the layers beyond layer 0 and don't do the blending, though, shazam - it becomes a pretty fantastic algorithm for smoothly scrolling a window across a gigantic terrain.

The gist is that you pan the logical zero point of the array around as you move through the world. By doing that, it makes it simple and logical to use the space left over from unloaded off-screen terrain as a spot to dump in new terrain data and stitch it in. You just set your VB to write-only.

scissorman posted:

I'm sorry but could you elaborate on that?
From what I read, you basically use a number of static buffers for the local (x,y) coordinates and get your global z from the elevation map.
So what do you mean by dumping new terrain in the VB?
Wouldn't you have to manipulate the elevation map instead?

Suspicious Dish posted:

https://www.shadertoy.com/view/4sf3Dr

Something stupid like this? Most of the work in that shader is because I can only use one fragment shader. If you were doing this for real, you'd do this as several post-processing effects by rendering each step to an FBO if possible.

The effect is also exaggerated. You'd want to tone it down for the final product.

DeathBySpoon posted:

Apparently he took that screen while looking at my post in the Making Games Megathread:



Has anyone here written a CRT / Scanline HLSL shader? I'm fairly new to shaders and I'm not 100% sure how to go about it, so I'm looking for resources on writing one. My game runs at a fairly low resolution so I figure a neat CRT effect would be a good way to upscale it (I'd leave nearest neighbor as an option for those who prefer it, though). Also any suggestions for retro / pixel art friendly scaling filters, really. I'd like to avoid stuff like 2xSai though, I think that'd make it feel too much like a poorly emulated game. Basically I just want to find a way to make my game look good on different screens while keeping it at the current resolution internally, and a scaling shader seems like a good option. I'm working in XNA with C#, for what it's worth.

Suspicious Dish posted:

Why do you need substr? indexOf gets you the next space, then you use charAt to read the character after the space, and then do whatever you want with the data in between that and the next index. Perhaps you want to use a binary file format that you parse with ArrayBuffer? I don't know.

scissorman posted:

What's the advantage of using voxels?
Do you know how fast a javascript application like this runs vs traditional programming languages?


I thought about using Minecraft as a base but I see two problems with that:
a) The simulation would primarily use a top-down, RTS-style camera, which is probably not what the engine is optimised for.
b) I'm mostly interested in the simulation and thus want to keep the rendering simple. All the additional stuff in Minecraft would just take away from that.

Suspicious Dish posted:

That's usually handled with a trie,

DancingPenguin posted:

About the top-down camera, I do not believe this engine will have any problem with it.
What you should check for is if you can use skyboxes/planes/domes or such, as to get a nice background for it, otherwise just fill the entire screen with <insertyoursimulationhere>.
Well all this engine really uses is blocks, so "all the additional stuff" (By which I suppose you mean the textures), is just added by:

code:

  generateVoxelChunk: generator,
  texturePath: './textures/',
  materials: ['grass', 'obsidian', 'dirt', 'whitewool', 'crate', 'brick']

roomforthetuna posted:

What I'm doing is parsing a dictionary file (word dictionary not programming-concept dictionary) into an 'object' (because that's Javascript's hash) for nice quick random access - the one character is my concession to cutting down the file size; I've converted the dictionary file from an alphabetical list of words into a sequence where, eg. "BANANA BANANAS BANGLE" would get encoded as "BANANA [chop0]S [chop4]GLE", with the "chop" markers being encoded as single characters.

Turns out that encoding the dictionary this way turns it from over 3MB to just over 1MB, so quite the saving, even though it gets re-inflated in RAM.

scissorman posted:

With "additional stuff" I was mostly talking about Minecraft and its non-graphics related sub-systems.
I don't know how the game is structured but, since I wouldn't use all the gameplay code, it would just add unnecessary complexity.

HardDisk posted:

I'm thinking of developing a game with Unity, and my current laptop (a HP G60 from 2009 ) really isn't cutting it anymore.

So:
Is there any specific recommended laptop to develop with Unity? If not, what laptop should I go for? I don't plan on going crazy on graphical effects, and I can't really afford the more gamey ones.
Are there any laptops that comes with Windows 7 anymore? If not, how's Unity compatibility with Linux?
I feel kinda silly asking this, but how's Unity compatibility with PS3 controllers on Windows (the MotionInJoy driver)?

Azazel posted:

I wasn't familiar with the Jump Point Search. What are the drawbacks to it? Because it seems quite fast.

FuzzySlippers posted:

The Gripper posted:

HardDisk posted:

I also own a fairly beefy desktop, but I am barely home nowadays. I'm planning using the laptop to develop on the downtime I have while I wait for classes to start.


Macbooks are out of my range for now, and those new processors will probably be out of my range as well. I don't know poo poo about laptop graphics adapters, if you could point me to a more specific one, it would be great. Knowing that I should look for OpenGL support helps a lot already.

Are the Intel adapters still completely terrible?

Thanks!

DancingPenguin posted:

Check the "About" page, all of the extra stuff seems to be assets that are not there unless you download them.
I believe that the actual gameplay code has to be created manually.
Might I ask what kind of simulation we are talking about?

HardDisk posted:

I also own a fairly beefy desktop, but I am barely home nowadays. I'm planning using the laptop to develop on the downtime I have while I wait for classes to start.


Macbooks are out of my range for now, and those new processors will probably be out of my range as well. I don't know poo poo about laptop graphics adapters, if you could point me to a more specific one, it would be great. Knowing that I should look for OpenGL support helps a lot already.

Are the Intel adapters still completely terrible?

Thanks!

scissorman posted:

I'm writing an artificial life simulation, starting with a simple predator-prey sim and expanding it from there.
There are a couple things I want to try out, like adaption to a changing environment, maybe some learning and evolution.

seiken posted:

Surely there's a Javascript implementation of a real compression algorithm like gzip or whatever you can use instead of this hand-rolled thing?

HardDisk posted:

Is there any specific recommended laptop to develop with Unity? If not, what laptop should I go for? I don't plan on going crazy on graphical effects, and I can't really afford the more gamey ones.

Torch Dexter posted:

I don't think anyone's made PC graphics cards/chips that don't support both in the past 15 years.

quote:

From BMFont's Options menu save the configuration settings file (extension: .bmfc) by choosing Save configuration as... And then save the bitmapped font itself by choosing Save bitmap font as... If you ever need to tweak it again, load the .bmfc file to restore the same settings with which you've created the .fnt file. It's a good idea to name both files identically.

yaoi prophet posted:

Picked up working on that old Haskell game I had; right now you can shoot at the blue 'chaser' objects, but there's no collision detection or anything. Only 250 lines, though, and I'm using bare OpenGL/GLFW.