It all comes down to being able to control your audio source signal. In most cases when you record audio you really don’t need it to be very loud. Think about some movies that you might have seen that contained explosions. Did it sound real? Did it feel real? Probably not, but it was close enough to the actual transients created by a concussive explosion that you really didn’t think about it.
It would be almost impossible to record a real explosion clearly with a simple recording device and a microphone. This is why we need so many volume/level knobs and sliders. Each one allows us to mix and control the level of our source program or sound. The reason a real explosion would be difficult to capture is because of the concussive shock that is related to an explosion. Ask any buddy of yours that has served in the military and they will tell you that the movies don’t come close to the real thing. The reason why is because the audio program in the movie will be missing an important component; the blast shock wave itself.
So instead of reproducing the actual concussive shock wave and audio engineer will use the bass frequencies to give the sound some “meat” and in if played in a movie theater that has a professional audio system you can “feel” the bass. The problem is that the audio engineer doesn’t want to cause the movie audience any kind of discomfort. After all, you are seeing a movie to have fun, right? Here is where we have the beginning of why we use so many level adjustments.
You source audio can be any vocal, instrument or special effects sound you are working with. You capture it as clean and as high of a volume possible without clipping. For a quick refresher see this article about input levels. For more about output levels see this.
Your input is your initial signal before it is processed including EQ and audio effect settings. If it is a guitar amp you are capturing then you will need to use lower settings in order to tame the mix later. If it is simply a vocal capture then the levels will need to be more dynamic and probably boosted in some places. Just make sure that your noise floor is well below your constant actual signal flow.
Once you have your input level set for each track and the sum of all audio signal from all tracks is not clipping the master output you are ready to start mixing and mastering your work. Yes, even explosions are mastered before they are released in a movie.