If you use the script to make a single file, such as your complete project as a PDF or an EPUB, then you can use the -o switch to specify this file, otherwise the file will land in the current working directory.
If your output consists of many files, such as all the HTML for your complete project, then you can specify a directory with the -d switch. Again, the default is the current working directory.
If you specify one of the components in List 6.6.2 and you are using a publication file to specify directories that are managed (Section 6.6) then the multiple files for that component will automatically be placed into necessary directories by default (rather than in the current working directory). Of course, you can override this behavior by specifying a directory with -d. For example, for many operating systems, by using -d . you can have the results land in the current working directory. (Note there is a necessary period there.)
In the early stages of a project, you might rebuild your images regularly. But you may not always want those results landing within directories under revision control. Later in a project, your images may be relatively stable, and you want to distribute them with your source, perhaps so others can re-purpose them in handouts or other materials. To accomodate this, make two publication files, and in one make a relative path for the generated components that is outside of the main directory that is under revision control. Likely you will need some file path syntax for a parent directory, such as ../ on Unix-like systems. Then you can switch from testing to distribution, and back, easily.