> 1. I want to spider based on my given levels.

Internal level ? You can set the depth to a defined level -- but generally you
don't need one (scan rules are sufficient).

> 2. I want to ignore external url's.

External URL's are ignored unless you ask not to ignore them. Just erase all
default filters (+*.jpg and so on..)

> 3. I don't want to download mentioned images.

Use scan rules too.

An example:

+www.example.com/*
-www.example.com/images/*
-*/bar.gif

Will mirror www.example.com, except the /images/ folder, and all images named
"bar.gif"