What a generated portrait is actually good at

The argument about generated portraits is usually framed as replacement: will these things put photographers out of work. It is the wrong question, and both answers to it are wrong.

Here is what the technology is genuinely good at, and what it is not.

Good at: the fortieth person

A photography session is an event. A workforce is a flow. Someone joins in March, and the March joiner does not match the January set. Consistent output across people is the thing that generated portraits do that scheduling cannot, and it is the strongest case in the whole category.

Same crop, same background, same colour temperature, on the fortieth face and the four hundredth. That is a genuinely different technical problem from making one good image.

Good at: volume with a deadline

Sports leagues are the clearest example. Four hundred players across twenty clubs, all needed before the season opens. The constraint is calendars, not cameras.

Good at: people who will not otherwise do it

Most people never book a session. Not for cost reasons — because it requires deciding to, and then a date, and then travel. The ordinary working portrait is a low-stakes need that loses to inertia, and removing the appointment removes the obstacle.

Not good at: telling you what is wrong

A photographer looks at you and says your chin is slightly too high, or that you look tense in the shoulders. That feedback loop is the actual product of a session, and no generation step reproduces it. Where the two sit relative to each other makes this point better than the vendor pages do.

Not good at: faces far from the average

Strong prescription glasses, tightly curled hair, darker skin tones, pronounced asymmetry. Failures cluster there, consistently, across products. Any comparison that does not include a difficult face has not tested anything.

What the process actually feels like

Shorter and less controllable than expected. The experience end to end is upload, wait, then choose from a set — and the choosing is where most of the quality difference lands, because a bad selection ruins a good batch.

If you just want one decent image

The plain version of the process is enough for that, and the one preparation step that matters is unglamorous: gather photographs from different days, in different clothes, in different light. That single input decision moves the result more than choosing between vendors.