Showing a generated image as a concept sets an expectation you may not be able to keep

A rough concept no longer looks rough

Visual exploration used to arrive in a form that announced its status. A sketch, a mood board, a grey layout with boxes: everyone could see it was a proposal. Generated concepts do not look like that. They look like the finished thing, and a viewer with no stake in the process has no way to read them as anything else.

What follows is familiar to anyone who has shown work too early. The direction stops being a direction and becomes the deliverable, and the conversation you wanted about approach turns into a conversation about details of a picture you had not committed to.

The pictures did their job as thinking. The trouble is what happened when they left the room.

Approval creates obligations the image cannot meet

Once a specific image is approved, you inherit a set of requirements it may not be able to satisfy. It has to exist at print resolution. It has to have a companion in a different orientation, and another one for next month with the same subject. Somebody has to be able to say where it came from and what rights attach to it.

A concept generated to answer a question about direction was not built with any of that in mind, and the gap between the approved picture and a usable asset is often larger than the work that produced it.

This is the practical reason for care rather than a matter of etiquette. The cost lands later, on somebody, and it is much cheaper to prevent than to explain.

There is a second cost that is harder to see. An approved concept ends the exploration, and it ends it at whatever the tool produced quickly rather than at the best available direction, which is not usually the same thing after twenty minutes of work.

Show the direction, not the picture

The straightforward mitigations are the ones used with any early work. Present two or three options rather than one, because a single image reads as a recommendation. Say in plain words what each is exploring and what is not decided. Keep them at a size and crop that supports discussion rather than admiration.

Where it fits, degrade them deliberately: grey them, crop them, place them small within a layout, or annotate them with what is being asked. Anything that restores the visual signal that this is a proposal will do more than a verbal caveat, which is forgotten within a minute of being said.

And separate the questions being asked. If the point is the composition, show the composition. If the point is the palette, do not attach it to a specific subject that will become the thing everyone discusses.

Agree what happens if the concept is chosen

The conversation worth having before showing anything is about the route from concept to final. Will the chosen direction be photographed, illustrated, licensed or generated at production quality? Who owns the result and on what terms? Is there a constraint — an internal policy, a client requirement, a regulated context — that rules out one of those routes?

Having that settled changes what you show. If the final images will be photographed, generated concepts are a briefing tool and should be presented as reference for the shoot, which is a role they perform well.

If the final images will themselves be generated, then resolution, consistency across the set, and any depiction of people are constraints that belong in the concept stage rather than after approval.

Where showing them at all is the wrong move

For some audiences and some subjects the risk outweighs the speed. Anything depicting a real product that does not exist yet, a building that has not been designed, a person who has not agreed, or a claim about how something will look and perform: in those, a convincing image creates a commitment that the project may not be able to honour.

The same applies where a viewer will use the image to make a decision with money attached. A picture is persuasive out of proportion to its status, and an audience deciding on the basis of a concept is deciding on the basis of something you invented to think with.

In those cases the older instruments are better precisely because they look provisional. A sketch, a wireframe, a written description or a reference photograph with a note about what will differ. Slower, less impressive in the room, and considerably less likely to be approved as though it were the work.