Thanksgiving Table Portrait
A candlelit turkey dinner, everyone in a row facing the camera, hands free of forks and glasses — the Thanksgiving photo, made from a group shot.
Turn this group photo into a warm Thanksgiving dinner portrait. Keep every person in this photo and do not add or remove anyone. Seat them in the same left-to-right order their faces appear in the photo: whoever is furthest left in the photo sits furthest left at the table. Keep everyone's exact facial features unchanged; every person must stay recognisably themselves. Everyone sits in one straight row behind a long wooden dinner table, facing the camera from across the table, all at the same distance from the camera so every face is the same size. They wear coordinated rust, cream and olive knitwear. On the table in front of them: a roast turkey, a pumpkin pie, small pumpkins, autumn leaves and lit candles, all kept below their shoulders. Hands rest on the table or in laps; no one is holding cutlery, a glass or food. Everyone looking at the camera, faces fully visible, no one turned away. Warm candlelight with late-afternoon window light, natural film colours, fine grain. No text anywhere in the image.
How to use it
Only edit photos you have permission to use, including permission from the people pictured.
- Use a source where every person is visible and the faces are large enough to check separately.
- Attach the source in an AI image editor compatible with this prompt (Gemini · ChatGPT (any AI image editor)), then paste the prompt. Adjust the shared setting and clothing while keeping the instructions to preserve each person. Avoid adding extra people to the scene.
- Compare each face with the original, not just the nearest one. Inspect overlapping arms, hands, and the places where people touch.
- Save the accepted image separately from the original and label it as AI-edited when sharing. Use the saved edit as the source if you choose to animate it next.
What happened when we tested it
We ran this twice from the same five-person group hug. Both rounds seated everyone with every face toward the camera, and nobody held a fork or a glass, so the hands line did its job. Both rounds also got the seating order wrong. Round one swapped the woman in the middle with the man beside her. Round two, the published version, kept her in the centre but swapped the two men end for end. Round one also lined everyone up down the length of the table, so the far face came out small; 'facing the camera from across the table, all at the same distance' fixed that. The knitwear colours only half held: both men's dark source shirts came back as blue and navy sweaters. If the order matters, add a line naming who sits where, and check it before you share.
