Place the ears before you describe the room
The same restaurant scene needs different sound depending on whether we hear the whole room or one person at the end of a table. In an objective wide view, plates, chairs, and several conversations can sit in layers at comparable distances. Once the scene enters one character’s perspective, that order should break. A fingertip rubbing the rim of a glass, a breath before someone speaks, or the movement of a collar comes forward because it is close to that person. The room is not being rebuilt. The mix is deciding what the listener is no longer allowed to miss.
Translating “subjective” immediately into a muffled filter or a tinnitus tone can make the scene feel like an effect. It is better to choose one listening anchor first. It might be the character’s breath, the consonants of the person across the table, or the repeated wheel of a cart at the end of the corridor. Once that anchor is legible, the amount by which everything else should recede becomes easier to judge. Perspective is revealed less by what was added than by what stopped masking the chosen sound.
One EQ curve cannot move a point of view
Rolling off the highs of a distant sound and increasing its reverb can help, but those moves alone do not place the listener beside a character. Distance also lives in attack definition, the ratio of direct sound to reflections, and the degree of masking from the room. A nearby fork does not have to be loud. If its short metal attack briefly clears the restaurant murmur, it is close enough. A loud voice can still remain at the back of the scene when its first consonant is soft and the wall reflection seems to arrive first.
In practice, divide the scene into three zones. Preserve transient detail and small movement in one or two sounds attached to the character. Keep middle-distance dialogue only as intelligible as the story requires. Hold the background at the minimum texture needed to keep the location continuous. Do not erase that last layer. When room tone and the breathing of a crowd disappear entirely, the ear does not move closer to the character; it teleports outside the set. Keep evidence of the same room while distributing a different resolution of information to each distance.
Move the focus half a beat before the cut
If the soundtrack narrows only after the picture cuts to a close-up, the handle for the transition arrives late. A more natural move is to soften the consonants of surrounding conversation just before the cut and let the next character’s anchor sound appear a fraction early. When the ear has already found a direction, the new shot confirms listening that has begun instead of throwing in unrelated information. In a scored scene, folding the rhythmic definition or stereo width before dropping the music level can move the perspective without overstating the dialogue.
The return journey matters too. When leaving a character’s interior and opening back to the room, do not restore every background layer at once. Release the close breath, then return events at different distances: a chair dragging, a laugh from the next table. That brief sequence puts air back into the scene. A good perspective change does not leave the impression that a filter was switched on. It leaves only the feeling that a world which had gone unheard has quietly widened again.