The single cheapest way to improve a generation

Background Removal

A cluttered background is the most common cause of a bad reconstruction. Cutting the subject out before it is submitted costs almost nothing and changes the result more than any other setting.

What it does

Isolates the subject

A transparent cut-out, so the model reconstructs what you photographed rather than what happened to be behind it.

Runs where it matters

Available on its own, and on by default in AI Texturing — where a backdrop’s colours would otherwise bleed into the inferred material.

Fixes the classic failure

The floor rendered as part of the object, a shadow becoming geometry, a wall fused to the subject’s back. All the same cause.

How it works

  1. Start with your photographWhatever you were going to submit.
  2. Remove the backgroundA transparent PNG comes back.
  3. Check the edgesHair, foliage and glass are where cut-outs fail, and they are worth a glance before you spend a generation on them.
  4. GenerateThe same prompt and the same model, on an image with nothing to misread.

What people use it for

  • Preparing a phone photograph for image-to-3D
  • Cleaning a set of reference views so they agree with each other
  • Preparing a reference image for retexturing
  • Producing product cut-outs from studio shots

What it cannot do

  • Fine detail at the edge — hair, fur, foliage — is where any cut-out struggles
  • Transparent and reflective subjects have no clear boundary to find
  • It removes the background; it does not remove the lighting the background caused

The details

Output
Transparent PNG
In texturing
On by default
Best for
Photographs with a single clear subject