Project Focus

As I said as few posts back I was unsure of exactly where I'm heading with this project. Well I've spent sometime thinking about what I want to do and why.

The why overall is pretty straight forward, experience. Also I want to know how things work, how it fits together. The what, is a bit more complex. The plan of attack is to create a Real-Time 3D Graphics Engine. I use Graphics Engine instead of Game Engine as, I hope, it to be flexible enough to work in many situation, including Games and Level Editors. Once the base engine is set up then I might branch it to different genres.

Here are the main features:

• Graphics Interface independent of API
• Graphical User Interface Framework
• Scene/Object Management
• Collision Detection System
• Rigid Body Dynamic System
• Scripting Language

I know the above is still not that specific, but it a start and a direction. The more I work at it the clearing it should become.

As I said previously, I'm starting with a toolbox, a number of common used classes and function that can be used throughout this project. I know I could create these utilities as I went along, but I want a good firm oundation to build upon. Thats why I'm starting with the maths Section.

thing2k

Toolbox Design

I'm spent the last few days deciding on where to start, I said the toolbox would be first but the question remaining, was which part. The maths system seem the obvious choice, as a number of other systems will rely on them.

The vector class, now this should be relatively straight forward, a class that holds 4 floats, x, y, z and w. Throw in a couple of functions and operators and your done.

Now wait right there, as this class will be used often through out this project, such a gun'ho attitude just won't cut it. A small mistake here would lead to countless error further down the line. So as I plan to treat this project professionally, I'll be designing each class, in reasonable detail, before any coding. Actually I plan to design each system before I start coding.

What this means is sitting down with pen and paper, quite a bit of paper, and thinking about how the system is going to work. The main parts to consider are:

• The classes and the what they will represent
• The public, private and protected properties
• Testing Methods
• How they interact
• Just about anything else

When finished, it time to type it up nice and neat, clear and concise.

For the maths section there will be, initially, 3 objects, a vector, a matrix and a quaternion, and these will be represented by the classes cVector, cMatrix and cQuat. Also there will be a number of helper functions, for things like creating a rotation matrix.

Also I say initially as the project progresses there may be a need to come back and add more to this system. Maybe some more helper functions, some slight changes to the classes or even add a new class that I didn't consider before.

For the most part this advise applies for any system, and I'll go through the same routine for each system of the toolbox, and beyond.


thing2k

A New Project

I'm just starting a new, large and probably over ambitious, project. This project, will be a collections of smaller relatively independent projects that will come together in a final application. For the moment I haven't quite decided what this application will be, just that it will be based around a Real-Time 3D graphics Engine.

I plan to try to design the each part of the project in detail, and to try and keep this blog up-to-date on progress. The starting point will be the Toolbox. The idea behind the Toolbox is for a collection of classes and functions that could be used in most graphical programs. This would include common maths classes for things like vectors and matrices, also useful graphics classes such as Mesh and Texture wrappers and maybe File Handling functions for loading images or model files.

Next in line is a Rendering Engine, I hope to make a interface that will support different rendering APIs, like OpenGL or DirectX. Initially it will have an OpenGL implementation, as my experience in greater and more recent the DX. Also I'm going to consider creating a Software Rastering system from near scratch, this is more a 'just to see if I can' than a feature of the project. From there I'm not sure where I'll go next, maybe a GUI, but as that could be a number of months away, I should have an idea by then.

To finish, if I discover something I think might be of interest or I get a related request, I'll write an article or tutorial about it here.

thing2k