pouët.net

for the past years, i've watched every win32 release from a certain party at least twice.

for the rest of the year, i watch the compos of parties i'm at, and maybe a dozen or so other releases, and that's it.

Now, after half a year with a combination of fulltime job + in-house relationship I've noted that my old time-consuming demomaking concepts don't really work anymore.

Here's my 5-part plan:

1. Whenever the IRC session is lame, just put the terminal away and do something more constructive with the computer.
2. Do something creative every day just to keep the spirit up, don't care so much about quality.
3. Recycle old shit instead of constant perfectionist rewriting.
4. Avoid finetuning with edit-test-edit-test cycles, write suitable realtime editing tools instead. Also generally lower the step for making and using new tools.
5. Start daily "fastcoding" exercises to improve the speed for making ad-hoc tools and proofs-of-concept.

to be productive again i have to throw out my ps2 and cancel my ISP subscription. and oh yeah, stop working.. that's the best one.

6. Entertain yourself only for improving yourself!

7. Profit

and let's not forget:

6.5. ???

Oh, damn. I forgot. :(

1. create demo
2. ???
3. profit!

How about when you want to pick up code: do you find it easier to start writing an effect from scratch after it has been thought up in your head, and then integrate it into your framework?

What would be the best order of events to do that creative code?

in my experience, the hardest part in doing stuff is getting started. it's so much easier to continue than to click "new project", start including pieces of framework, etc... at least i hate it when you keep coding for minutes just to see something.

i'd say the opposite

One of the points in my "fastcode exercise" idea was to minimize the start-up effort by getting used to quickly doing something from scratch.

The way I see it there are three good ways to structure your work if you want to be efficient with very limited time and stay motivated too.

#include <stdio.h>

void main()
{
printf("Fast demo!\n");
}

Gargaj's words.

Coding in JS, I can either copy the few files making my "framework" or hit a keystroke in my editor and start from a "single-file FX template" I wrote. It saves time and allow me to focus on the juicy part.

Gargaj: yeah um... that is _SO_ not my problem.

Viznut: Hmm. Yes. Hmmm. [adopt plan]

Gargaj; you could have the DXM sdk it has a template for 64k's in Visual Studio, it even has a few options to get the fractal clouds you need from the start!

oh, and it comes with a selfinstalling installer.

I make demos at work and at home, usually between 11.00-1.00 am. Weekends also, favourite time is Saturday afternoon and early Sunday.

But that happens every day. Give it a miss a few days and the schedule will slip.

I also work without access to internet. I play backgammon and watch TV while compiling/rendering.

i work only hard before deadlines. so i must relearn everything for each demo

late nights...

You can keep yourself motivated with these incredibly tough coding challenges. But be careful, if you're not a real 1337 haxx0r your head may explode.

It used to be easy when I had a boring 30 hour/week job. Then even with wife/child (but no games/tv!) there's quite a lot of time left.

Now that I have a really interesting job I don't code demos anymore! Ha! Is that the reason why we didn't release anything since 2005? :)

having girlfriend and a complete loft to repair myself, and not having a demo-related job, the only way for me to do anything useful during the day (not necesarily scene related) is from 11pm-2am (no TV at home, no games installed but windows defaults). I wish I was student again!

In a 4k, once you have an initial idea you really just need to bash the thing out in your code and explore creative avenues that open up while you're implementing, since you've got so little space to work with, and it's tough to merge things.

In larger stuff - it's indispensable to have a framework you can easily recompile and test, feed interesting data into, and run on different machines. I play with things in my personal D3D WMP vis framework. a 1024-tap FFT can makes for nice data, if you listen to weird music.

If you can handle the multi-pass "technique" structure and don't mind being windows-only, then .fx files are good, and ATI's Rendermonkey is king. This limits you to what a card can do in GPU, but that's not much of a limit these days. In any case, if you implemented ATI's scene graph in your engine you'd be flying. That's a HEUGE first step though.

Anyway, the best effects I've done have been dreamed up on paper, on the bus or in airports. :D There's nothing like a complete design to make implementation easy.