EditorNodesPricingBlog

BACKGROUND REMOVER

Cut the subject out, cleanly

Edge detection aimed at the boundaries that usually fail, meaning hair, fur and semi transparent material. Works on files you upload and on images you generated in the same workspace.

Remove a background

Where background removal is hard

Cutting out a solid object against a plain wall is straightforward, and almost any tool manages it. The cases that separate good removal from bad are hair, fur, and anything partially transparent, because the correct answer along those edges is a partial value rather than a decision to keep or discard.

BEN2 handles the removal, with edge detection built for exactly those boundaries. What still governs the result is contrast. A subject that separates clearly from what sits behind it cuts cleanly, and matching tones across the edge is what produces ragged output.

That is worth knowing before you generate the frame rather than after. When you know a subject will need isolating, prompt for a background that contrasts with it, and the removal becomes easy instead of marginal.

Background removal in the Image Enhance form

Two inputs, since the model handles the rest without parameters to tune.

Image source section for supplying the picture to have its background removed

The source section. Upload a file or reuse a previous generation.

1

Supply the image

The source is either an upload or one of your previous completed tasks. Reusing a generated frame directly is common, since a subject you just created often needs isolating before it goes anywhere else.

Resolution of the source matters here. Fine edge detail such as hair is read from the pixels available, so a small source gives the model less to work with along exactly the edges that are hardest.

2

Choose background removal

Background removal sits under Image Enhance alongside upscaling, so you select the generation type first and then the operation. BEN2 handles the removal itself.

The two operations are separate runs. When a shot needs both, remove the background first and upscale afterwards, so the enlargement is not spent on pixels you are about to discard.

Operation type selector with background removal chosen in the Image Enhance form

Background removal is one of the two Image Enhance operations.

Settings reference

The background removal form is deliberately short. These are the inputs it takes.

Source image

The picture to cut out. Comes from an upload or any of your previous completed tasks in the workspace.

Upload or previous result

Operation type

Selects background removal rather than upscaling. Both operations share the Image Enhance form.

Background removal or upscaling

Model

BEN2 performs the removal, using edge detection tuned for the boundaries that usually fail, such as hair and semi transparent material.

BEN2

Where it fits in a workflow

Background removal is a compositing step, so it belongs after the content is settled and before delivery. The usual sequence generates or edits the frame first, isolates the subject, and only then upscales what remains.

Running it in the other order wastes the upscale. Enlarging a full frame and then cutting most of it away means the credits went into a background nobody sees.

Source resolution is the one thing worth protecting. Edge quality is read from the pixels available along the boundary, so an isolated subject from a small source will show its limits at the edges regardless of what happens downstream.

Related tutorials

Deeper coverage of matting and removal on the blog.

BEN2 background removal

Edge quality on hair and transparency, and where the model reaches its limits.

Video matting with MatAnyone 2

The same problem across a moving sequence rather than a single frame.

Background removal questions

How well does it handle hair and fur?

Hair is the hardest case for any background removal, and BEN2 uses edge detection aimed at exactly that boundary. Results depend heavily on contrast between the subject and what is behind it. Dark hair against a dark background is difficult no matter which tool is doing the work.

Partially transparent regions are the other hard case, because the correct answer is a partial value rather than keep or discard. Expect these areas to need review, and where the shot allows it, generate the subject against a contrasting background in the first place.

Before. Removing first means the upscale is applied only to the pixels you are keeping, which is both cheaper and avoids spending resolution on a background that is about to be discarded.

Yes. Edge detail is read from the pixels present, so a low resolution source gives the model less information exactly where precision matters most. If you control the source, start larger.

Yes. The source picker accepts any of your previous completed tasks, so a frame from Text to Image or Character Generation can be isolated without downloading and re-uploading it.

Contrast at the boundary. A subject that separates clearly from what is behind it, in brightness or colour or both, cuts cleanly. Matching tones across the edge is what produces the ragged results, which is worth planning for when you generate the frame.

Other ways to generate images

Background removal is one of seven generation types in the image workspace.

Text to Image

Write a prompt and generate images from scratch. Pick the model, aspect ratio and output format, and add a LoRA style if you have one.

Image to Image

Start from an image you already have and describe the change. Useful for style transfer, edits and mood changes on an existing frame.

Character Generation

Upload one to three reference photos and keep the same character consistent across many generations.

Image Upscaling

Increase resolution on an existing image while keeping identity and detail intact, with a selectable upscale factor.

Model Training

Train a custom FLUX LoRA on your own images so you can generate in your own style or with a specific subject.

Draw to Edit

Sketch an idea on the built in canvas, add a prompt, and turn the drawing into a finished image.

    © 2026 - AI FILMS LLC