nzgda.com

marksibly · Joined: 23 Oct 2003 Posts: 22 Location: New Zealand

Hi,

Just to curious to know if anyone out there is working on an original, from scratch, 3D engine?

I realise that in these market driven times its not cool to 'reinvent the wheel' from scratch etc, but I would still like to think the expertise to do so exists here!

Semi-related: Am I correct in thinking that the Auran Jet engine was originally funded by the OZ govt? And that several NZ game developers are either using it or considering doing so?

Bye!
Mark

grm · Posted: Mon Apr 12, 2004 7:52 pm Post subject:

We have been tinkering around with a directx based "engine".
I guess it all depends what your targets are. Being able to create something that was API independant is what these major game engines have going for them. Being able to cross platforms is an absolute must if you want to maximise your art and mesh assets. Does tend to lead to "ported" looking games though.
Personally I love the flexiabilty and features I get form Directx (8.1), being able to program shaders and know the workings of our engine means we are constantly adding to it, and trying to get the best from this particular platform.
We are about 75% of the way through it, I'm currently trying to deal with switchs (both light and animation) to turn things on and off and then sorting out camera control. Then it basically onto sound which will be a new experience.

Some of the features already implemented are:

* Bump Mapping
* Stencil Shadows
* Keyframe animation engine
* "Movers" animation (basic matrix animations)
* Cube mapping
* Level editor
* Sky Domes

Everything is done via Vertex/ Pixel Shaders, with no support for leagacy cards.
(It's way easier to pick on a none moving target).

The main reasons we are "reinventing the wheel" really comes down to 2 things.

1) I couldn't write a word of C++ when we started. So I figured it was one way of learning. Having done Z80 Asm, Pascal , Logo and Basic over 10 years ago I figured it was time I caught up.

2) Money.

Mario · Joined: 03 Oct 2003 Posts: 241 Location: Wellington

Simon · Joined: 23 Oct 2003 Posts: 42 Location: Dunedin,NZ

We had a look at Auran Jet for our games course but the cost was too high as it had more features than we needed.

In terms of 3D game engine development. All of our third year graphics students write a 3D raytracer. Although this is not realtime, given the advancement in graphics cards and CPU, I expect to see the first raytraced realtime games coming out in the next 18 months-2years. It may be a brave call but the new gen of cards from NVidia and ATI are pushing the boundary for realtime raytracing.

As far as our games course goes we do not ask the students to reimplement a 3D engine. They either write a 2D engine from scratch or use a provided 3D engine. We don't quite have the time to get them to create a full engine.

One company that has written a fairly complete 3D engine is Pandromeda here in Dunedin. Have a look at www.pandromeda.com. They are showing off a game they have been working on for 18months at E3. So yes the talent is there and some are doing the work you just have to dig a little Wink

Simon
_________________
Simon McCallum
Computer Science Department
University of Otago
simon@cs.otago.ac.nz, (03) 479 5740

darkwing · Joined: 14 Jan 2004 Posts: 9

I've kinda been working on one in conjunction with my MSc thesis. It's hosted by Bitkeeper http://edge.bkbits.net. I guess the main point of difference is that it combines a physics engine with a renderer(OpenGL) and scene graph system. There aren't many opensource engines around that do this. I've also tried to apply modern C++ design practices. For example I've used various packages from Boost (ublas, shared_ptr, unit test library and various other things) and applied techniques from Alexandrescu's book. The overall architecture isn't as good as I'd like it to be - the holy grail for me will be something akin to the STL.

Scene graph/renderer features at the moment are basic:
Single textures
materials
A mesh object

When I get the architecture right, more features will follow. (It's easier to add features, harder to change the architecture).

Physics Features:
Particle based constraint system (based on Witkin and Baraffs siggraph notes - handles fixed distance, vector, angle, position contraints.)
Mass spring system
Ideal Gas System (I combined this with the mass spring system to make a squishy body)

Still to do:
Rigid Body dynamics

Side note - did you know that Renderware which bought Mathengine was originally worked on extensively by the ODE http://ode.org/ guy who did his Phd at Auckland uni? So the talent exists and is around except that it gets shipped overseas.

I've spent some time tracking over the net ex kiwis who are in game dev and graphics:
Russel Smith, PhD (ODE, Mathengine - don't know where he works)
Chris Butcher, MSc (Bungie)
Andrew Willmott, PhD (Maxis)

Don't know of any others but it would be interesting to know more names. They happen to have advanced degrees but I don't think it's necessary for game development, given that you can buy something like Renderware or Havok and leverage that. (With a degree it's the training that's important, not whether you're definately going to apply everything what you learn at work - anyway that's what I tell myself.)

darkwing · Joined: 14 Jan 2004 Posts: 9

darkwing · Joined: 14 Jan 2004 Posts: 9

kaffiene · Joined: 11 Nov 2003 Posts: 1 Location: Dunedin

Zaita · Joined: 22 Sep 2004 Posts: 8

Yea,

Im working on my own original Engine/Framework from Scratch. Its being designed not solely for games but is also being designed to provide some physics engine and graphics modelling capabilities.

Im doing this for the learning experience and to see how good a job i can do creating an engine which is designed to be scalable, good looking while ultimately low-cost on a system and stable..... something like the recent UT2004 engine... and not something liek the recent Doom3 and FarCry engines.

As nice as those engines where, they were very flawed which is sad considering where they came from, ups to UT developers thou.
_________________
---------------------------------------------
Zaita
www.zaita.com
* Home of the 3D Zaita Framework *
---------------------------------------------

Guest · Posted: Sat Nov 06, 2004 2:32 am Post subject:

I am currently starting to look into this. Not with any specific desire to end up using it, but just to brush up on my minimal 3d knowledge. If I was looking at creating a game now, I would more than likely go with an engine that's already done. Probably something free like Ogre3d, or cheap like Torque at least to get to prototype stage. Then if funding was established, maybe look at purchasing a real cross platform (Consoles and PC) engine and reimplementing the graphical interface between it and the client.

I tend to feel I don't have the luxury of time like I did all those years ago - I would see wasting a year or two on a good engine to be a bad use of my resources unless I was trying to go cutting edge graphically.

Aaron

jack.19987 · Joined: 28 Jul 2010 Posts: 3

In terms of 3D game engine development.As far as our games course goes we do not ask the students to reimplement a 3D engine.

Some of the features already implemented are:

Bump Mapping
Cube mapping
Level editor
Sky Domes
_________________
16gb cf card