Image prompt
Creates a tight close-up crop from a two-panel image-quality comparison while preserving the original labels and anonymized subject.
Copy directly into the target model and adjust for your input assets.
Using REFERENCE_0 as the source image, create an extreme zoom-in crop for a detail-comparison view. Keep the same two stacked comparison sections and preserve the existing anonymizing face blocks, but crop tightly around the subject’s face, hair, jawline, and neck so the facial area fills most of the frame. Leave only partial edge context from the original forest background and the left-side white labels visible. Preserve exactly 2 visible section labels: “GPT IMAGE 2 NO UPSCALE” in the upper section and “TOPAZ PHOTO” in the lower section. Maintain the dark cinematic lighting, original colors, and comparison-divider structure; do not add new objects, new text, or change the subject.
Original source details are kept for attribution and review.
GitHub source：YouMind-OpenLab/gpt-image-2-prompts-search
This page is adapted from YouMind OpenLab GitHub prompt libraries. License: MIT。