Jump to content

  • Log In with Google      Sign In   
  • Create Account

new here. working on a game...

Old topic!

Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

  • You cannot reply to this topic
8 replies to this topic

#1   Members   


Posted 02 February 2013 - 04:36 AM

well, anyways, I am new here, and was mostly looking for places where I could write about stuff (I have been developing my project mostly in isolation over a long period of time).
actually, the "game" part of the project is more recent, most of the rest was written over a much longer period of time.
some videos of it are available on my YouTube channel (among other things):
here is a video showing stuff going on:


ADD: it may also be downloaded here (if/when server is up):

so, general information:
ground-up engine, mostly written in C.
takes some technical inspiration from both Quake and Minecraft (and, to some extent, games like Doom 3);
gameplay tries to follow in a similar style to a mix of Quake and Minecraft;
renderer basically works vaguely similarly (in concept) to the Doom3 renderer (it uses dynamic lighting and depth-pass stencil shadows).
most of the high-level entities work similarly to those in Quake (most have similar or matching classnames, and typically use the same basic key/value pairs).
however, currently, I am using a voxel-based world-structure, more like Minecraft (errm, and many similar blocks with similar names and appearacnes exist), though there are some more block-related features, and some differences. granted, the number of block-related features and interactions isn't nearly as extensive as in Minecraft at present.
it also supports Quake-like brush-based worlds as well, and on/off I had worked to better unify voxel-based and brush-based world-structure (the idea being that each "region" could have its own local BSP for the brush-geometry, rather than a single unified BSP). note that like with voxels (and unlike Quake and friends) the BSP is rebuilt dynamically. technically, bezier-patches and mesh-objects are also supported. the main drawback is that these can't really be as effectively integrated into gameplay nearly as well as breaking and placing blocks can.
it uses a custom scripting language (BGBScript), which is roughly in a similar style to JavaScript and ActionScript3 (and by uses similar syntax, but can also use more traditional C/Java/C# style decalration syntax), but incorporates elements of Java, C#, C, and C++. it supports both static (manifest) typing and dynamic typing, as well as both class/instance and prototype OO styles.
dynamic features, like ability to load code directly from source code, compile/execute code via "eval()", ... are also available.
design is mostly just being a "grab bag" of whatever features I liked or found useful from the other languages, so it also has things like structs and (bounded) pointers (they resemble C-style pointers, but are boxed, and often use bounds-checking). many other design choices were made on a "most common option" basis (where the option with the more languages that use it wins).
the VM is also smart enough (and capable of parsing C headers), such that it automatically generates any glue needed to call between the scripting language and C land (though a special tool is still needed to allow C->script calls absent using API calls or function-pointers or similar). also, script code and VM metadata can be (and often is) packaged up inside of DLL or SO files (both PE/COFF and ELF are supported), ...
there are a lot of special features as well, such as vector and quaternion types being promoted to built-in types, ...
gameplay concepts:
player runs around and shoots stuff (optional, there is a "peaceful" skill level where only passive AI-entities spawn);
enemies may drop stuff when killed (weapons and items);
player can also break and place blocks and build stuff;
the available weapons (and weapon numberings) are partly inspired by those in Doom 1/2;
currently, no higher-level goals or obectives exist.
general game concept:
on a distant colony in space, mysterious aliens have invaded with armies of mass-produced bio-mechanical robots.
the player is (or at least thinks he is) a human bio-mechanically augmented cyborg (the augments are both biological and mechanical).
mostly, the human colonists were turning themselves into cyborgs to better combat the invading alien robots.
well, there is more plot and backstory than this, but I may need some sort of mechanism to reveal plot and backstory in game.
could always make plot-fragments be rarely dropped items, but would probably need a UI for them or something.
or such...

Edited by cr88192, 17 February 2013 - 05:23 PM.

#2   Members   


Posted 02 February 2013 - 09:57 AM

Looks cool! Sounds like you've been busy. You definitely added a lot of features. Keep up the good work!

#3   Members   


Posted 02 February 2013 - 10:13 AM

If you have more to say about your project, you might look into starting a journal here at gamedev.net.

#4   Members   


Posted 02 February 2013 - 05:49 PM



a downside of this project is that thus far it has had a longish development time (the game part of the project starting in 2010, many other parts being older).



I was also working some on a Native Client port of the engine as well, but there is still a bit more work here, most of the work being in moving a lot of the renderer over to wrappers such as to allow glossing over the issues of full OpenGL vs OpenGL ES, and probably also getting around to all the NaCl specific stuff (pepper API, ...).


as-is, it is desktop only, and primarily Windows (it is built and used on Linux, at least occasionally).



for 32-bit x86, there is currently also a naive JIT for the script language (written fairly recently).

there is also a plain C interpreter, but it is a bit slower.

performance also depends somewhat on the code (statically typed code is faster, ...).


currently, script code is the 2nd place language within the engine proper (slightly beating out C++ in terms of line-count), but still a small minority of the project (behind C).

#5   Members   


Posted 09 February 2013 - 02:01 AM

got distracted from Native Client port for now...
(may get back to this...).

ended up then doing this:
wrote some DXTn encoders, mostly "quick and dirty";
was looking at making video-mapping more efficient, but as-is, I don't actually have enough video-maps in the correct format in order to be able to really bog down the engine (basically, using "Motion-JPEG" / "Motion-BTJ" codec, where "BTJ" is a custom extended form of JPEG).

here is a video where I am using video mapping:

present video streams (currently 7): "slipgate", "fire_anim", "ponyboost", "redstone_torch", "greenstone_torch", "bluestone_torch", "torch".

"ponyboost" is the highest resolution, at 512x512, but the others are lower, like 256x256 or 128x128.

note that, unlike normal animated textures, video-maps don't necessarily have a defined maximum length, and each frame is loaded into the texture as the video plays.

granted, this could be better demonstrated with longer video clips, but off-hand I lack a tool to convert arbitrary videos into the correct format (power-of-2 sized Motion-JPEG AVIs or similar), which as-is, is a requirement for the engine (hence, I am using converted-over animated textures... which doesn't really look like much...).

#6   Members   


Posted 09 February 2013 - 04:02 AM

Awesome! Bonus points for coding in C, and for including ponies biggrin.png


Looks like you're ready to get into full content production mode. Are you doing all the design/art/music yourself, or rounding up a team? I don't think you'd have much trouble finding people. Great work, and best of luck :)

#7   Members   


Posted 09 February 2013 - 08:42 AM

Awesome! Bonus points for coding in C, and for including ponies biggrin.png
Looks like you're ready to get into full content production mode. Are you doing all the design/art/music yourself, or rounding up a team? I don't think you'd have much trouble finding people. Great work, and best of luck smile.png

I am currently the sole developer (including art and similar as well), and am operating on pretty much no budget.
so, yeah, I am partly limited by the quality and amount of artwork I can make.

#8   Members   


Posted 12 February 2013 - 10:10 PM

status update (effort over past few days):
direct DXT5 video transcoding (MJPEG transcoded directly to DXT5), which slightly improves performance.

experimentally was also writing custom DXT encoders, and putting a few up on pastebin:

DXT encoder (early, intended mostly for speed):

DXT encoder for video (YUV 4:2:0, also intended somewhat for speed):

DXT5 encoder (intended for slightly higher quality):

(this was an experiment in sending YUV data directly in textures and converting it to RGB on the GPU, which could theoretically pull off better image quality).

#9   Members   


Posted 17 February 2013 - 02:24 AM

for anyone interested, some of my stuff can be downloaded from here:


(granted, assuming of course that my server is working...).
I just recently (2013-02-17) put a new version of it online.

the project mini-dump comes with the project source, but this doesn't necessarily mean free rights to do whatever with it (it isn't all FOSS, though certain parts are under the MIT license).
if anyone wants to use the source in creating derived works or similar, they can ask me about it.
people are free to experiment with it or use it for their own personal uses, or distribute mods or similar.

currently, it doesn't cost anything, but donations are appreciated...
(like, with enough donations, I can better justify continued investment of time and effort into all this...).

Edited by cr88192, 17 February 2013 - 02:28 AM.

Old topic!

Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.