Conspiracy - Beyond Expectations

The final executable for a 64k intro is nothing more than the engine packaged together with a datafile and its loader. The loader parses the datafile and builds up the structures and resources required by the engine to run the release the same way it does in the tool. Separating the code and data this way is **one of the key insights** in creating a modern 64k intro - the compressibility gains are insane. This article is a summary of our experiences with minimal file formats as we've been through a couple iterations. ## Problem Definition Like any data driven piece of code, a 64k engine works by having certain data structures and resources in memory that it evaluates during execution. While working in the tool these structures usually have some additional metadata that helps the editing process, but isn't actually needed for demo playback (think of things like names, graph node location, organizational folders, etc.).

So you decided to create a 64k intro. You fire up Visual Studio 2022, create a new C++ project, set it to 32 bit release mode, type `int main(){}`, hit compile... and the executable is 9 kilobytes. This used to be a lot worse, but this is surely not a great start.

I first experimented with texture generation in 2000. For a while I didn't really get far because there weren't too many resources on the topic - until I found an article written by Ile/Aardbei, which I still highly recommend to anyone new to the subject. The article you're reading is meant to be a follow up to Ile's - we now have graphical hardware that is basically built to do this sort of stuff, but using only the GPU to implement a texture generator can be tricky due to the nature of pixel shaders.

Part 4: Let's talk about synthesizers.

Part 3: The basic system.

Part 2: Why are SMF files smaller?

Part 1 : The concept

This is a story of a 64k intro modelling tool and explanation of the ideas behind the file format.

(Republished with the kind permission of Ile - thanks!)