Removal is a description, not a selection
No lasso, no mask. Write the object out of the scene - remove the trash bin on the right -. The model erases it while reconstructing what was behind: pavement, wall, foliage, sky.
What removes cleanly
Distinct objects on textured or repeating backgrounds vanish best. People in streets, cars on roads, wires against sky, logos on walls. Objects that touch your main subject need a precise sentence about what must remain.
Finish the shot
After removal, most photos deserve one more pass: auto-enhance to lift the exposure, or 4x upscale before print. Both are one click from the result.
FAQ
Can it remove a person?
Yes - describe the person by position (the person on the left) rather than by identity. State what should stay.
Will the fill look fake?
On natural backgrounds the reconstruction is usually seamless. Complex man-made patterns may need a second, more specific pass - that second run is a fresh generation.
How much does it cost?
From 14 credits per photo with GPT Image 2. The exact quote is shown before you run it.