Background Removal and Transparency

Automatic Background Removal

The Remove BG tool analyzes image colors and attempts to separate the foreground from the background.

It uses a traditional image-processing algorithm rather than a machine-learning segmentation model.

How to Use

1. Open Studio.
2. Upload an image.
3. Locate Remove BG.
4. Adjust the Feather setting if desired.
5. Click Remove BG.
6. Inspect the resulting transparency.
7. Download the result as PNG.

How Background Removal Works

The implementation uses several steps:

This approach works particularly well when the background is relatively uniform and distinguishable from the subject.

Feather

Feather controls the softness of the generated transparency boundary.

A low value produces a sharper boundary.

A higher value can create a softer transition, although excessive feathering may soften subject details.

Limitations

Background removal may be less accurate when:

• The background contains multiple complex textures
• Foreground and background colors are similar
• The subject extends into the image edges
• The image contains fine hair or transparent materials
• Shadows are similar in color to the subject

Because the algorithm is color-based, it does not understand objects semantically.

Color to Transparency

The Studio also provides a Color → α tool.

This operation makes pixels transparent when their RGB color is sufficiently close to a selected color.

How to Use

1. Upload an image in Studio.
2. Enter a HEX color in the Color → α field.
3. Set the Tolerance.
4. Click Make Transparent.
5. Review the image.
6. Download as PNG.

Example

To remove a white background:

Target color: #FFFFFF
Tolerance: 20

Pixels sufficiently close to white become transparent.

Color Input

Supported HEX examples:

#FFFFFF
#FFF
#000000
#3B82F6

Three-digit HEX values are expanded to six-digit RGB values.

Tolerance

The tool calculates the Euclidean distance between the pixel's RGB values and the selected target color.

A pixel becomes transparent when its distance is within the configured tolerance.

Low Tolerance

Removes only colors very close to the selected value.

High Tolerance

Removes a wider range of similar colors.

Increasing tolerance can accidentally remove portions of the subject.

Example: Removing White

Consider a pixel:

RGB(255, 255, 255)

With target:

#FFFFFF

The pixel's color distance is zero, so it matches at any nonnegative tolerance.

A slightly gray pixel might also match if the tolerance is sufficiently high.

Preserving Transparency

Use PNG when exporting images with transparent backgrounds.

JPEG does not preserve an alpha channel.

Choosing a Method

Important Implementation Detail

The automatic Remove BG operation uses the originally loaded image as its input.

If you've already rotated, resized, or modified the current image, automatic removal may regenerate its result from the original upload rather than the most recent edited version.

The Color → α operation instead processes the current image data.

Best Practices

• Start with a clear image.
• Use a low feather value when sharp edges matter.
• Use small tolerance adjustments.
• Inspect the entire subject before exporting.
• Save transparent results as PNG.

Note: These tools provide automated color-based transparency editing, not guaranteed perfect subject extraction.