PixKit

Home / All Tools /Utility

Replace Color

Pick any colour in an image and swap the whole region for a new one, with tolerance, edge softening and a live pixel count.

✓ Local processing · No uploads · No signup

Drop an image, or click to browse

Click the image to pick the colour to replace; export recomputes at full resolution

Replace

—

With

#FFFFFF

Real-world uses

  • A supplier logo PNG comes on white and needs to sit on a dark page: click the white, set #0F172A with tolerance around 12 so JPEG noise goes too, keep edge softening on to avoid jaggies, then export PNG.
  • A product set shares one off-white background that drifts a few levels per shot: raise the tolerance to about 20 and they all become pure white at once — far faster than opening each in an editor. Zoom in afterwards to check for a leftover halo.

Limits: Matching is by colour value only, not by shape — a wide tolerance also recolours every similar shade in the image, including the parts you wanted kept. JPEG noise means a "solid" fill is never truly solid, so a tolerance of 0 often matches nothing; 8–15 is the practical floor. Anti-aliased edges are blends between the old and new colour and sit outside the tolerance, which leaves a fringe; softening or a wider tolerance helps. The result is opaque — use the background remover to get transparency. Output is untagged sRGB.

Related tools:Confirm the exact value firstRemove the background entirely

How it works

  1. Drop an image and click the colour to replace on the canvas — sampling always reads the original pixels, not the preview
  2. Choose the replacement colour, then drag the tolerance from an exact match up to swallowing nearby shades
  3. Check the pixel count, pick a format and download; export recomputes at the original resolution

FAQ

Why does tolerance 0 replace nothing?

Zero tolerance means an exact match: one level off in the red channel and the pixel is skipped. JPEG is lossy, so neighbouring pixels of the "same" white usually differ by 2–5 levels and almost nothing matches exactly. A tolerance of 8–15 covers that noise — which is also why sampling the middle of a large flat area is the safest bet.

Can I replace only a selected area?

No — matching runs across the whole image. For a local change, crop that region out first and process it on its own, or use a selection with fill in a design tool. Keep the tolerance low and step it up: that is far less work than widening it once and hunting for areas that got changed by mistake.

Why is there a coloured fringe after replacing?

Those are anti-aliasing pixels sitting between the old and new colour: far enough from the target to fall outside the tolerance, so they survive. Turn on edge softening to blend the boundary, or widen the tolerance to absorb the transition band. Exporting as JPG re-introduces noise and makes the fringe worse, so PNG or WebP is the better output here.

Can the background become transparent?

No, the replacement colour is always opaque. The background remover is the right tool for transparency: it works from the subject outline instead of colour values, so it will not punch holes through same-coloured areas inside the subject, which colour matching does.

Related tools