Four people sit around a meeting table, but the empty chair nearest the door attracts the most attention. A notebook lies open in front of it. At the far end, someone is pointing towards a wall of coloured notes. For a proposed leadership workshop, I would use this deliberately ordinary scene to ask a difficult question: what do we actually know about the meeting, and what are we merely assuming?
I would want a still illustration that supports several plausible interpretations. It should not announce that one person is a bad manager or that another is disengaged. The exercise would begin with observation, then move into the questions a leader might ask before acting. A useful image would leave enough uncertainty for participants to disagree without turning the discussion into a guessing game about facial expressions.
For this concept, nano banana 3 would serve as an AI image generation and editing model. I would start with a written description of the fictional room and then consider edits to the resulting image. The intended deliverable is a static workshop illustration, accompanied by questions I write separately. It is not a recording of a real team or evidence about anyone’s behaviour.
Build the room around a question
Before describing the people, I would decide what the session needs to explore. My choice would be the difference between interpreting silence and inviting participation. The image would show one person taking notes, another looking towards the door, a third presenting, and a fourth sitting with a closed notebook. None of those details proves what a person thinks. Together, however, they offer concrete material for a conversation.
The room would be modest: an oval table, a window along one side, a whiteboard at the back, and enough open floor to see the empty chair clearly. I would avoid a grand boardroom, expensive suits, or a commanding figure at the head of the table. Those visual shortcuts could steer the exercise towards status before anyone has discussed the actual meeting. Everyday clothing and comparable seating positions would suit the intended ambiguity better.
Describe visible details with nano banana 3
My opening prompt for nano banana 3 would read: “Create an editorial illustration of a fictional workplace meeting in a small, naturally lit room. Four adults sit around an oval table with one empty chair near the door. One person presents beside a whiteboard with abstract coloured notes, one writes in a notebook, one looks towards the doorway, and one has a closed notebook. Neutral expressions, restrained colours, eye-level viewpoint, no readable writing or logos.”
The distinction between “looks towards the doorway” and “wants to escape” matters here. The first describes something the illustration can show. The second supplies an interpretation that I want learners to examine rather than inherit. Likewise, I would avoid asking for an arrogant presenter or an anxious employee. A picture designed around those labels could make the intended discussion feel predetermined.
I would choose an illustration style with recognisable posture and simplified faces. Photorealism is unnecessary for this exercise and might encourage participants to treat tiny facial details as meaningful evidence. A subdued blue wall, warm wood table, and mustard notes would give the scene a coherent palette without turning any person into the obvious centre of attention.
Revise the detail that gives away the answer
Suppose the first proposed image showed the presenter towering over everyone while the seated people appeared visibly unhappy. I would regard that as a mismatch with the brief. Rather than adding more explanation below the picture, I would try an image edit in nano banana 3 that asks for a more balanced composition: the presenter seated at the same level, relaxed shoulders, neutral expressions, and an unobstructed view of the empty chair.
That edit would still need inspection. I would compare the revised image against the original and look for changes that were not requested, such as an extra participant, a different notebook position, or a missing doorway. If the room became cluttered, I would simplify the prompt and work towards one clear scene. The value lies in the final illustration’s usefulness, not in preserving a particular generation at all costs.
A second possible problem would be the empty chair occupying half the foreground. It would then look like an obvious clue rather than an ordinary detail. My revision would make it similar in size to the other chairs while retaining its position near the door. Participants should notice it through observation, not because the image practically points an arrow at it.
Give the facilitator two images with distinct jobs
Once the main composition was suitable, I would consider using nano banana 3 to develop a companion version of the same concept. In that version, a hand would rest beside an open notebook where the earlier notebook had been closed. I would treat this as a separate illustrative alternative and check it carefully, rather than assuming every unrelated feature would remain identical.
The first image would open the discussion. The second would arrive only after the group had named its assumptions. I would ask whether that small change altered anyone’s interpretation, and what additional information they would still need. The purpose would be to make the relationship between visible cues and confident conclusions easier to discuss. I would not claim that either image measures empathy, leadership ability, or psychological safety.
Keep the finished slide easy to read
I would place the chosen illustration on a plain slide, with the question “What can you observe?” written beside it in the presentation editor. A later slide would ask, “What would you ask before deciding?” Keeping these words outside the generated image would make spelling, size, and contrast easier to control. A text description of the room would also help participants who cannot comfortably inspect the picture.
The final nano banana 3 illustration would therefore occupy a specific, limited role in the workshop: a shared fictional scene for practising careful observation. My delivery package would contain the selected still image, its alternative if useful, and a facilitator note explaining the proposed discussion. Before using it, I would check that no visual detail unfairly assigns competence or authority to a demographic group. The strongest ending would be a practical question from a participant, not a verdict about the people in the room.
