I got a question that has been bothering me for a long time...

So if a company like EA games creates a game, they often make a type of file, for combat arms, its .rez, in gta san andreas, model files are .dff, mw2 is .iwd...

But how do you reverse engineers find out how those files can be handled/extracted ? Is it like: you look at it and you're like : "oh like that"...

Or do you really look at the code of the game and see how the game handles it ?

Most developers are lazy and use a common format with a different extension.

iwds are zip files, ff are just zlib compressed memory dumps. These were discovered by just knowing that zip files start with "PK" and zlib files start with the hex 78 DA. Knowing what kind of data is being stored also helps a lot (models store verts/faces and scripts store raw text). You can even ignore a lot of it because most files include either an index or some kind of system that indicates how large blocks of data are that you can skip over and come back to later. It just takes long hours of staring at a hex editor.

Most developers are lazy and use a common format with a different extension.

iwds are zip files, ff are just zlib compressed memory dumps. These were discovered by just knowing that zip files start with "PK" and zlib files start with the hex 78 DA. Knowing what kind of data is being stored also helps a lot (models store verts/faces and scripts store raw text). You can even ignore a lot of it because most files include either an index or some kind of system that indicates how large blocks of data are that you can skip over and come back to later. It just takes long hours of staring at a hex editor.

So the developer that discovered it knew every file format ?

Battlefield 3 stores data in .cas files, and since they don't provide any SDK's, I still didn't find any solution to extract the sounds/models... Thanks for your answer !

Here is another question, some guy made a savegame editor for GTA SA, how did he find that out ?(He provided the source)