Remove object from photo and review what replaces it

"Remove the cup" works only when there is one obvious cup. If the picture contains several, add a position, color or relationship: the small blue cup on the right, beside the plate. Then name the nearby objects that should stay. Those details give the request a clear target and make the result easier to judge.

Use Remove Object from Photo for bounded objects such as litter, table clutter, cables or a small sign stand. People and lettering have their own pages because their failure checks are different. If you want the cup to become a vase, use replacement instead. A removal brief should end with an appropriate surface, not another focal object.

The visible object may be only part of the distraction. A tall bottle can cast a long shadow, and a cup can appear again in a reflective table. Removing the body while leaving those traces can make the edit look unfinished. Before submitting, look around the object and decide which associated shadow or reflection should disappear.

Be precise about shared shadows. If a cup and a plate overlap, asking to remove all shadows could flatten the plate as well. Explain that the cup's shadow should go while the plate and its lighting remain. This is one reason a Remove Object from Photo request benefits from a brief that describes relationships rather than a single delete command.

If an unwanted object is partly outside the frame, inspect the edge of the exported picture too. A small remaining fragment can look like an intentional border detail until the image is placed against a contrasting page.

Upload the clearest available JPG, PNG, WebP or HEIC, up to 20MB. A source with visible detail around the object gives you more evidence for judging the fill.

1. 1

   Select the source version

   Use the original or an accepted previous edit. Confirm which photograph is active when the project contains similar versions.

2. 2

   Describe the removal and fill

   Identify the object and its relevant shadow, then name the underlying surface: pavement, grass, wall, sky or tabletop. Preserve neighboring objects explicitly when the area is crowded.

3. 3

   Review the complete photograph

   Check that the correct object disappeared, then inspect the old footprint at full size. Compare distant parts of the scene too, so an unwanted change elsewhere does not escape attention.

Illustrative prompt, not a tested output: "Remove the cup at the far right and its small shadow. Continue the wooden tabletop through that area, following the visible grain. Keep the plate, cutlery and front edge of the table unchanged." The table edge and grain give the generated patch two concrete constraints.

In this example, Remove Object from Photo succeeds only if the cup is gone and the remaining surface reads as one table. A vaguely wood-colored patch is not enough if the grain turns abruptly. If the removed cup covered the table edge, trace that edge through the repaired region and check whether its thickness and perspective still agree.

The more structured the hidden background, the more specific your review should be. A patch of grass can look plausible in many arrangements; a railing has a narrow range of believable continuations.

First follow the former outline. Look for a thin rim, a small fragment of a handle or a patch that looks blurred compared with its surroundings. Then inspect the shadow footprint and any reflection you identified. A successful-looking central fill can still leave the object recognizable through those peripheral traces.

Next look for new objects. A removal may replace clutter with a plausible but unwanted feature, such as an extra leaf or a new joint in a tabletop. Remove Object from Photo is a generative edit, so ask whether the new detail serves the intended image. It may look coherent while changing what the photograph documents.

Keep an untouched version for records, property details or anything where the original scene matters. Cleaning up a personal composition is different from presenting the edited picture as an exact account of a place or item. The missing surface is inferred even when no visible artifact remains.

A generated removal returns one result and costs 10 credits at 1K, 20 at 2K or 30 at 4K. Auto normally starts at 1K. Internal retries do not add another charge. If you decide to request a different fill or remove another item afterward, that new submission is a separate image task.

Keep the original and completed versions in view. Returning to an earlier result is free, and branching from it lets you test a revised request without treating an unsuccessful patch as the new source. Before another Remove Object from Photo attempt, write down what was wrong: leftover shadow, broken line or unintended neighbor change. That observation is more useful than asking for a generally better result.

Review the photograph at the intended display size after the close inspection. A tiny remaining artifact may be invisible in a small card, while an empty repaired area can still feel awkward in the overall composition. A modest crop may improve balance after removal. Cropping and resizing are free basic operations and do not need another generated fill.

Export PNG, JPG or WebP once the composition works. Keep the original filename distinguishable from the edited version outside the app as well. For a crowded scene involving people, continue with the dedicated person-removal guidance. For a damaged surface that needs repair without deleting a distinct object, the inpainting page explains the boundary checks in more detail.