🔒 Locked

Project Ideas

Started by Kuladus Jul 20, 2004 at 2:55 AM 10 replies 2k views

Original Post

Jul 20, 2004 at 2:55 AM

Hi, I'm doing an advanced project subject for my CS degree. The project lasts 13 weeks, is not marked, and can be pretty much whatever we want. The thing is, I can think of a few things, but nothing really is standing out at the moment. Does anyone have any suggestions, it does not have to game related, and probably is better off not being a game but rather something else. The language doesn't matter. Some ideas... A single or multi-user (over the net) drawing/cad/modelling type application. An opengl demo. A chat server and/or client. (possibly an irc client) A console emulator. (8-bit nintendo etc.) A raytracer or software renderer. Some sort of plugin for existing software (ie a gimp plugin, or a wings3d plugin).. I have a week to think about it, so please let me hear your ideas. Thnx. Edit: Oh yer a couple more things of note: this project is marked as a pass or fail and completion of the project isn't necessarily required. Also I may have to work with another person on it. [Edited by - Kuladus on July 20, 2004 3:33:48 AM]

DigitalDelusion

Jul 20, 2004 at 3:04 AM

* Implement some compressions algorithms like Huffman and LZW, profs just love the smell of stuff like that.

* A high performance math library using expression templates and/or SIMD. Fun stuff to work with just don't aim it at being "complete".

Toolmaker

Jul 20, 2004 at 3:09 AM

Write your own WinZip clone with support for 2 or 3 compression algo's. Make the thing puggeable so you can easily create new DLLs for different compression algo's and GO!

Maybe even with a cool auto update notification:
"The ZIP format has been updated. Do you want to download and install the new update?" [Yes] [No]

Toolmaker

risingdragon3

Jul 20, 2004 at 3:34 AM

Meh. Do something interesting. All of the above is boring. Do something with machine learning. Runtime optimization. Code analysis and automatic code generation. Aim to do something small but new in some way - don't aim for something big but only adding very little "new" stuff.

I would suggest writing a plugin or using preexisting opensource code to use as a framework. Use those 13 weeks to do something substantive and interesting, not writing base code.

joanusdmentia

Jul 20, 2004 at 3:50 AM

Quote:
Original quote by risingdragon3
Meh. Do something interesting. All of the above is boring. Do something with machine learning. Runtime optimization. Code analysis and automatic code generation. Aim to do something small but new in some way - don't aim for something big but only adding very little "new" stuff.

I would suggest writing a plugin or using preexisting opensource code to use as a framework. Use those 13 weeks to do something substantive and interesting, not writing base code.

Seconded.

Real-time model deformation, soft-body physics, articulated body responses to physical stimuli (hit someone with a wooden plank and watch them try to keep their balance), just to name a few....

superpig

Jul 20, 2004 at 4:20 AM

Quote:
Original post by joanusdmentia
Quote:
Original quote by risingdragon3
Meh. Do something interesting. All of the above is boring. Do something with machine learning. Runtime optimization. Code analysis and automatic code generation. Aim to do something small but new in some way - don't aim for something big but only adding very little "new" stuff.

I would suggest writing a plugin or using preexisting opensource code to use as a framework. Use those 13 weeks to do something substantive and interesting, not writing base code.

Seconded.

Real-time model deformation, soft-body physics, articulated body responses to physical stimuli (hit someone with a wooden plank and watch them try to keep their balance), just to name a few....

Thirded.

Dynamic recompilation (self-modifying code) would be my choice.

risingdragon3

Jul 20, 2004 at 4:34 AM

Quote:
Original post by superpig
...

Dynamic recompilation (self-modifying code) would be my choice.

A lot of that is going on in JITs. I hear they are doing interesting stuff with runtime optimization / dynamic recompilation - getting speeds faster then a static exectuable even with the interpretation penalty. However it's hard to find heursitics to spend the very small amount of time on recompilation / optimization that you have (you have to run the app too) :(

I wonder if somehow you could have a compile phase where you pinpoint target areas for JITs: very tight inner loops, lots of virtual calls in loops, stuff like that. It's probably been done before.

Toolmaker

Jul 20, 2004 at 4:45 AM

Or implement ragdoll physics?

Rob Loach

Jul 20, 2004 at 7:21 AM

Quote:
Original post by Toolmaker
Or implement ragdoll physics?

... with LZW compression [wink].

Extrarius

Jul 20, 2004 at 7:28 AM

It might be fun (and fairly simple) to make a text-based diablo-like game. When I say text based, I don't mean mud-style "You are in the room. WDYD?" but 'graphical text' using various characters to mark different interesting things, with randomly generated levels, items (diablo-style with various attributes etc), monsters, quests, etc.

Max_Payne

Jul 20, 2004 at 7:51 AM

Making a raytracer is always fun. It involves a medium degree of mathematical complexity (at least linear algebra). You can have fun by creating your own effects, and work on possible optimizations. You can even raytrace some 3D animations.

Kuladus

Jul 23, 2004 at 1:13 AM

*bmp*

I'm considering doing a rigid body simulation.

More suggestions welcome!

Topic Locked

This topic has been locked by a moderator. New replies are not allowed.