One should be careful not to confuse perspective with volume.
In the medium long shot, Tough Guy speaks at minus 6 and Nerd at minus 10. The boom mic is maybe two and a half feet overhead due to the loose framing of the shot. When we push in to a single head close-up of Nerd, the boom mic is able to move in to a much closer position. It becomes relatively easy to record Nerd at minus 6 because the mic is so close. That would be an error!
In real life, when we talk with a person standing ten feet away from us, we tend to both see and hear more of the surroundings. But when we step in closer to the person, our mind tends to blank out some of the surroundings as we focus our eyes on the face in front of us. This is a perspective change. It is also a gradual and self-motivated change.
In cinema, changes in camera angle occur spontaneously and are motivated by the director/editor, not the viewer. The change may be a bit of a sensory shock.
Audiences tend to accept the visual change, since in real life our brain is constantly shifting focus and scope of what we see (a biological imitation of zooms and cuts, if you will). But it takes us longer to adapt to outwardly imposed changes on the audio, especially when it creates a discontinuity of levels (normal, loud, soft, loud, normal, soft, soft, loud, etc.) within the scene.
Getting back to our example scene above, when we move the mic closer to the Nerd for his close-up, the effect is to make his voice dominate over the surrounding background, which is in keeping with the natural change in perspective. But if we allow his voice level to rise above its established level range, then the audio becomes disjointed from the time line of the complete scene and will not smoothly intercut with the rest of the footage.
Therefore, when you move the microphone in for a close-up, re-adjust the volume so that the actor's voice level remains constant with the rest of the sequence. Characters' audio should be somewhat constant throughout the course of the scene, even as the shot changes from wide shots to mediums to singles to reverses to mediums, etc. If you were to close your eyes, the changes in audio from shot to shot should not be unnatural nor unexpected.
This is not to say that if an actor walks distantly away from camera that his voice level should not diminish. Of course it should, as it would in real life. But a variation in camera angle (as opposed to a change in actor location within the set, visual or implied) does not warrant a major change in audio levels. However, a major change in camera LOCATION may justify a change in relative audio, particularly the background.
Of course there are always going to be some changes in audio levels. This is an art form, not a controlled manufacturing process. The nature of production is such that we can't always control things as much as we'd like to, such as mic placement and background ambiance. The idea, though, is to at least try and keep these level changes as minimal and inconspicuous as we can when we record them; and then to fix them completely during post-production.
Not only does sound need to be consistent within a shot, and from shot to shot, since this footage may all be integrated during editing -- but sound must also match up when scenes butt up against other scenes.
Throughout the duration of the production, try to establish and then maintain relative audio levels for all of your characters. Change perspective (the blend of background to dialogue) as necessary, but try to keep your characters as constant as possible.
Equalization is another important aspect to consider, as well as straight audio level. Avoid using any more equalization than is absolutely necessary on the set. Traditionally, a mixer will roll off the excess bass frequencies to reduce wind noise and rumble, especially out of doors. Some mixers like to boost the mid-range frequencies just a smidge, in order to emphasize speech over ambiance. High frequencies are usually left alone.
If you choose to employ some equalization on a shoot, make certain than you apply the setting consistently from the first till the last day of production. For instance, many mixers have a set degree of bass rolloff that they will use outdoors and a lesser amount of rolloff for interiors. That is okay since people do sound different outside than inside. But do not vary the intensity of rolloff from day to day based on the local wind conditions. Otherwise, what sounds good Monday and Tuesday may not intercut well with material recorded the week before, or the month later!
Resist the temptation to sweeten the mix on location by playing with all those colored dials. Once you record something with EQ, it cannot be undone later on. Record your tracks as plain as possible, and save the special effects and final tweaking for post, where they have the liberty of working with edited sequences and of repeating their attempts until it all sounds right.
The only time that a mixer is justified to employ extraordinary EQ to improve a shot (that is, anything over and above your "permanent" bass rolloff and possible midrange bump) is when the alternative is to absolutely have to loop the scene unless correction is applied. In other words, you can play with the EQ only when you have absolutely nothing to loose and anything to gain. If in doubt, leave the EQ settings alone!
In conclusion, plan ahead!
See how your characters interplay, and then try to establish and maintain their relative audio levels and EQ regardless of close-up or wider shot. Changing perspective does not mean changing volume, only reducing background. Louder does not mean better.
Adjust your EQ "permanent" settings for interior or exterior, but do not mess around from shot to shot nor scene to scene.
Above all else, think like an editor. All this stuff has to intercut smoothly and seamlessly. From consistent work habits you will achieve consistent soundtracks.