Sorry your browser is not supported!

I've been enjoying game dev recently, so have been starting a few threads. My next line of curiosity is, what is the main technical challenge you have faced recently in your latest game? What has got you thinking a lot about how to achieve it?

Mine is spaceship to spaceship collisions. In my space battleship game I am making, the ships are fully destructible. This is fine with bullets and rockets whizzing in and damaging the hull, but what happens when I crash ships together? At the moment I'm trying to get 100+ wall sections to collide with 100+ other wall sections and destroy each other, plus all the men, guns, boxes etc. that get crushed in the process. Meanwhile, I want correctly handle the speed/momentum involved, and the ship in a convincing damaged state afterwards. It has me scratching my head, and working really hard to make the collisions CPU efficient.

So what is currently testing your grey matter?

Recently, it was trying to work out the orientation of a video camera from the image it returned (in realtime). It is complicated on so many levels, from accurately recognising the green points on the image, despite the varying light levels, to using these points to generate the camera view matrix.

That sounds interesting. That's not related to a discussion a while ago about 3D software for compositing video, is it? It sounds like a similar technical challenge. If you make some DBPro software for adding 3D into a video stream, I will give you 8 million pounds.

Some people might just call you nosy Fallout

My biggest problem with development at the moment is in UnrealScript trying to do the simple task of creating a completely new camera class based off the spectator camera class that can support camera animations whilst still having limited control over the camera's rotation itself. Causing me some problems at the moment though. Don't know whether its the pneumonia or my stupidity thats throwing me down the wrong path every time but i'll get it eventually.

Recently i'm playing with my hypersphere project I first posted in code snippets ages ago. Past few days i've been trying to implement a plains (planes) system where the planes remain facing the camera, but do not rotate along the line of sight from observer to object (in the world frame). In other words you can roll when looking at the plains, and they wont roll with you.

Playing Half Life 2 on the Xbox, i noticed in the level where a giant rocket launches, and you look up at the trail it's left, the particles spin with you, which looks totally rubbish.

Anyway, it sounds simple, and it sort of is, but for the past 3 days, feeling that it should pop out easily with a bit of trial and error, and having a "feeling" for what form the solution should take, kept trying things that didn't work at all. So today I actually sat down and thought about it and got it done. I feel happy but also stupid.

The reason i'm doing this current project is mainly because of the challenge. It's pretty pointless otherwise. The version I posted in code snippets uses 4x4 rotation matrices, and works fine. I figured i'd make it work with (double) quaternions. The reason : because it's more elegant, and i was worried about my so(4) matrices going all wonky. Unless you play the game for a week though, they won't. I still did it because it's cool. There are clues on the internet about how to do it, but compared to the standard 3d case, it's pretty poorly documented. So i feel more smug to have got it working.

I find the hardest part of coding is doing the intellectually unstimulating but necessary parts of a project. The parts where you know what has to be done.