How to Remove Image Backgrounds Online (2026)

How Color-Based Background Removal Works

The simplest approach is flood fill removal: you click on the background color, and the algorithm removes all connected pixels of that color. This works beautifully for solid-color backgrounds — product photos on white, green screen footage, or graphics on solid backgrounds.

The algorithm starts at your click point and expands outward, checking each neighboring pixel. If a pixel is close enough in color to the target (within the tolerance threshold), it becomes transparent. The expansion continues until it hits pixels that differ too much from the target color.

This is fundamentally different from AI-based background removal, which uses neural networks to identify the subject. Color-based removal is simpler, faster, more predictable, and works entirely in your browser with zero server dependency.

Understanding Tolerance and Edge Settings

Tolerance controls how similar a pixel must be to the target color to be removed. At 0% tolerance, only exact color matches are removed. At 100%, everything is removed regardless of color. The sweet spot is typically 15-35%.

Low tolerance (5-15%) works for perfectly uniform backgrounds — digitally generated solid colors. Higher tolerance (25-45%) handles real-world backgrounds with subtle color variations from lighting, shadows, and camera noise.

Edge softness (feathering) blends the transition between the subject and the transparent background. Without feathering, cutouts look harsh and artificial. A subtle 1-2px feather creates a natural edge that composites smoothly onto new backgrounds.

Handling Difficult Cases

Hair and fine detail: color-based removal struggles with hair, fur, and wispy edges because these areas blend gradually with the background. For these cases, AI-powered tools generally produce better results.

Shadows: subject shadows on the background create a gradual color transition. Low tolerance preserves shadows (which you may want for natural compositing). High tolerance removes them (cleaner cutout but less realistic).

Multi-colored backgrounds: gradient backgrounds, textured backgrounds, and environments with varied colors require multiple passes at different click points and tolerance settings, or a completely different approach.

Saving Your Cutout

Always save background-removed images as PNG. JPG does not support transparency — saving as JPG replaces transparent areas with white, defeating the purpose of the background removal.

PNG files with transparency are larger than equivalent JPGs. A cutout that would be 200KB as JPG might be 800KB as PNG. This is the cost of preserving the alpha channel.

If you need a white background (not transparent), the tool can flatten the transparency onto white before export. This produces a smaller file that works anywhere, though you lose the ability to place the image on other backgrounds later.