On paper, every angle costs a drawing, so you choose carefully. In a 3D viewport an angle costs nothing, and that is the trap. Scenes get shot from wherever the camera happened to be, geography falls apart, and the viewer quietly stops knowing where anyone is standing. Blocking is the habit that prevents this, and it works the same for two fictional adult women in a bedroom as for a car chase.

Fix the room before the camera

Start by deciding what the space is. Where is the door, where is the window, which way does the bed face, how far is it from the wall. Write it down or sketch a plan view. This sounds tedious and it saves you the specific misery of discovering in shot nineteen that the window has moved to the other side of the room since shot four.

A plan view also tells you what is possible. If the room is small, long-lens shots from across it do not exist, so stop looking for them. Constraints of that kind are useful. They rule out three quarters of the choices immediately and leave you arguing about the ones that actually differ from each other in meaning rather than in degree.

The line of action still applies

Draw an imaginary line between the two characters. Keep the camera on one side of it. Do that and the woman on the left stays on the left through every cut, and the audience never has to re-orient. Cross it without warning and the two of them appear to swap places, which reads as a mistake even to viewers who have never heard the rule.

You can cross the line deliberately: with a moving camera that carries the audience over it, or on a cut to a neutral head-on angle first. Both are old tricks and both still work. What does not work is crossing it by accident because the viewport was already facing that way when you set a key.

Block in grey first

Turn off textures. Turn off the good lighting. Work with untextured geometry and a flat light, because at that stage you are judging silhouette, spacing, and screen composition, and nothing else. Any problem you can see in grey is a real problem. Anything you cannot see in grey will not be rescued by a nicer shader later on.

Grey blocking is also where you find intersections. Two bodies close together will pass through each other constantly during a first pass, and it is far cheaper to notice it now than after a lighting artist has spent two days on the shot. Fix contact and clearance while the scene is still cheap to change.

Keeping two bodies readable

Two figures at similar distance from the camera tend to merge into one shape. Separate them: change the distance, change the height, change the direction one of them faces. Contrast in silhouette is what lets a viewer parse the frame in the quarter second before they consciously look. It is the same principle behind giving each woman her own posture and rhythm.

Overlap helps too, as long as it is deliberate. A near figure crossing in front of a far one creates depth and tells you the order of things. Random overlap creates confusion. The test is simple: freeze a frame, squint, and see whether you can still count the people and tell which is which.

What blocking in a viewport buys you

The payoff is that later stages get cheap. Lighting knows where the camera is. Animation knows how long a shot runs. Nobody rebuilds a set because a wall was never really there. A scene that was blocked properly can survive having its lighting thrown away and redone, which is not true of a scene assembled by wandering the viewport.

It also makes the work honest. When the camera has a reason to be where it is, the audience feels a point of view rather than a floating eye. Weight problems in the animation are a separate matter, covered in the piece on timing and settle, and the studio route for building an original scene lives on its own page. Generation stays on pose creator where the safety controls sit.