Choose between lossy WebP, lossless WebP, and optimized PNG
A PNG may be a photograph, a screenshot, a logo, or an interface asset. Its content and transparency—not its extension alone—should decide the output mode.
Identify what the PNG is storing
Photographic PNGs are often large because lossless compression preserves every decoded pixel. Lossy WebP can be a strong delivery option for those images. Logos, diagrams, and screenshots with small text have hard edges that expose lossy artifacts more readily.
If exact decoded pixels are important, compare lossless WebP with a losslessly optimized PNG. If the graphic uses a limited set of colors, PNG quantization may achieve a better balance while retaining the PNG workflow.
Test alpha on the real backgrounds
WebP can preserve transparency, but lossy color changes near semi-transparent pixels can create visible fringes. Review logos, soft shadows, and anti-aliased edges over both light and dark backgrounds used by the destination.
Do not flatten the image unless it has one guaranteed background. Flattening removes flexibility and can permanently bake the wrong edge color into a delivery asset.
Measure the result instead of assuming conversion wins
A simple indexed PNG may already be smaller than a lossless WebP output. Conversely, a full-color screenshot or transparent photograph may improve substantially. Compare the downloaded bytes and inspect the output before changing a production asset path.
Keep the source PNG or editable design master. Ziply downloads a separate WebP and does not overwrite the selected file, making it possible to reject an output that does not earn its place.