Removing one object from a crowded shot: make the prompt specific

I'm on the remover.work team. If several objects share a frame, a removal prompt needs to identify which one you mean.

Start with a short clip you own. Describe the target with a distinctive detail and its position, such as 'remove the red backpack behind the cyclist.' This is more precise than 'remove the object.' Our prompt tool uses that description to select what to remove and reconstruct the area behind it.

Preview the result through the whole movement, especially when the target passes behind something or the camera moves. Check nearby objects and background texture as well as the selected target; reconstruction can still need another pass.

The attached image shows the tool's upload page on our English website. It is an interface screenshot, not a before-and-after result.

Tool: https://remover.work/tools/remove-anything-from-video-by-prompt

When describing a target in a crowded shot, do you usually start with its color, position, or movement?