A composition with two figures is a relationship and the eye handles it automatically. Add a third and something changes: the image becomes a crowd, and crowds are read by grouping rather than by individual attention. Most multi-figure compositions fail because they were built as several two-figure compositions sharing a frame.

The tools that fix this are compositional rather than anatomical, and they are the same tools used for crowds in painting, film and comics generally. None of them require drawing anything better; they require deciding what the picture is about and then arranging everything to support that decision.

Readers count to about three

There is a well-observed limit to how many separate elements a viewer tracks before switching to estimation. Below roughly four, each figure is registered individually. Above it, the eye starts grouping, and a picture with six figures is read as two or three masses rather than as six people.

This is useful rather than unfortunate, because it means you can control the reading by controlling the grouping. Arrange figures into two or three clear masses and the composition becomes legible immediately, even if the number of people has not changed at all.

Decide who the picture is about

Every multi-figure image needs a primary subject, and the others exist in relation to her. Without that decision the composition has no hierarchy, which produces the specific restlessness described in the piece on symmetry, where the eye has no reason to prefer any candidate.

Once decided, everything else follows: the primary figure gets the strongest value contrast, the clearest silhouette, and the most detail, while the others are grouped, overlapped and quieted. That is not a reduction in their importance to the story; it is how a still image directs attention.

Overlap creates depth and order

Figures placed side by side at the same distance produce a flat row and no reading order. Overlapping them establishes which is nearer, which creates depth and simultaneously creates sequence, since the eye reads front to back. Overlap is the cheapest structural tool available in multi-figure work.

Amount matters. Very slight overlap reads as an accident and produces tangents, where two edges touch confusingly. Substantial overlap reads as deliberate. The rule of thumb is to overlap by a lot or not at all, and never by a few pixels.

Value grouping does the rest

Once figures are grouped physically, group them tonally. Two figures sharing a value range read as one mass; a third in a contrasting range separates. This lets you build a composition of two or three tonal shapes regardless of how many people are in it, which is exactly what the eye wants.

Squinting is the check, as always. Squint at the image and count the masses. If the answer is more than three, the composition is still a crowd rather than an arrangement, and further rendering will not help. The related value discipline is covered in the palette notes.

Legibility and what it protects

In intimate multi-figure work legibility does more than aesthetic duty. A reader who cannot tell who is doing what, or who cannot locate every participant's face and attention, cannot read the mutual participation that makes the scene coherent. That is the choreography requirement described in the piece on visible consent, and it gets harder with every figure added.

The practical consequence is that multi-figure scenes need fewer participants than artists usually attempt. Three well-composed figures communicate more than six poorly arranged ones, and everybody stays legible, which is both the better picture and the more responsible one.

Arranging a crowd so it still reads

Working order: decide the subject, group the rest into two or three masses, overlap decisively, separate the masses by value, then squint and count. If the count is three or fewer, the composition will read. If it is not, regroup before doing anything else.

Characters here are fictional and unmistakably adult, and every participant in a scene must remain individually legible as such, which is a standard set out in the reader agreement.