Ideal inputs
Upload one image with a visible subject and describe the replacement location, surface, depth, light direction, and amount of negative space.
Replace the setting behind an uploaded subject with a described scene while keeping the foreground direction clear.
Upload one image with a visible subject and describe the replacement location, surface, depth, light direction, and amount of negative space.
Choose a source with a clean silhouette and note any edges, hair, or transparent parts that need care.
Name the place, surface, horizon, time of day, palette, perspective, and space needed around the subject.
Check edge halos, scale, contact shadows, reflected light, and whether the subject still belongs in the scene.
GPT Image 2 generates a new background concept around the uploaded foreground subject.
This is not a pixel-locked background replacement. Hair, reflections, transparent material, and subject lighting may change during the edit.
Both use an uploaded image and a described setting; this page is framed around replacing the existing background, while the generator page emphasizes building a new scene around the subject.
No. The edit asks for preservation, but generated edges and lighting can vary. Inspect the subject before use.