How Transparent Backgrounds Work

The alpha channel

An ordinary RGB pixel stores red, green, and blue values. A transparent PNG can add alpha: a value describing how visible that pixel is. Fully transparent pixels reveal the background; fully opaque pixels cover it; intermediate values blend the two.

This is why soft hair, fabric, and antialiased curves do not need a staircase-like hard edge. Their boundary can include partial coverage.

Why editors show a checkerboard

The checkerboard is a viewing aid that makes invisible areas visible to the editor. It is not stored as the new background unless someone deliberately flattens the image against that pattern.

Different applications use different checker colors, and simple image viewers may show transparency as white, black, or the window color.

What happens at the boundary

A partially transparent edge pixel contains both a color and an alpha value. If its color was influenced by the old background, placing it over a very different color can reveal a fringe.

Background removal therefore involves more than marking pixels visible or invisible. Good edge behavior also depends on preserving genuine soft detail while limiting old-background contamination.

Transparency in common workflows

Use transparent PNGs when a subject needs to appear over changing colors, web layouts, slides, or design compositions. Keep an opaque version when the destination does not support alpha or requires a specific background.

Avoid JPEG for a final transparent asset because standard JPEG does not store an alpha channel. Converting the finished PNG to JPEG will flatten or replace the transparent area.

Conclusion

Transparency is an alpha value, not a visible color. Understanding partial edge pixels explains both the flexibility of transparent PNGs and the halos that can appear when a cutout moves to a new background.