Memory Fragment Portrait

Design & animation notes

At a glance the dominant picture is ONE coherent recognizable human face, divided into aligned rectangular photo tiles by narrow off-white gutters. The facial tiles are crops of one continuous portrait, not different portraits or randomly rearranged features. Replace a small minority of face-grid cells with discrete photographs of shared everyday memories. These memory photographs occupy whole cells inside the facial mosaic, never as a separate moodboard beside the face and never as translucent double exposure. Leave selected grid cells completely empty as warm off-white paper. These absences interrupt the portrait meaningfully while enough eye, nose and mouth information remains to recognize a face. Variables to swap: aspect_ratio, style_fidelity_anchors, source_content_to_avoid, subject (+10 more). Aspect ratios: 9:16 / 16:9.

How to use this prompt

1. Click Copy prompt (no account needed).
2. Paste it into ChatGPT, Claude, Nano Banana Pro.
3. Replace the variables in the style JSON, keep the visual system, generate.