Convert PNG to SVG

Only up to a point, and it is worth setting expectations first. Tracing turns pixel regions into paths, which works well for logos, line art and flat-colour graphics with clear boundaries. A photograph has no such boundaries, so the trace yields either a crude posterised approximation or an SVG with tens of thousands of paths that is larger and slower to render than the original. SVG cannot represent a photograph — tracing one produces either a crude approximation or a file larger than the original raster.

In an SVG, text is real text — selectable, searchable and re-styleable, and it needs the font to be available or embedded to render as intended. Converting away from SVG turns that text into pixels or into document text depending on the target; converting towards SVG only produces real text if the source had a text layer to begin with.

No pixel data, no. Both formats store the image without lossy compression, so the picture that comes out is identical to the one that went in and the conversion can be repeated indefinitely. What changes is file size, and which features travel — colour depth, embedded profiles, layers and animation are all supported unevenly between lossless formats. PNG uses DEFLATE compression over filtered scanlines, so every pixel survives exactly, plus a full 8-bit alpha channel. An SVG stores shapes as mathematical paths and text as text, not as pixels, so it is a document rather than a raster grid.

Transparency survives when going to PNG, WebP, TIFF, GIF, or SVG. Converting to JPG flattens the alpha channel against a white background — if you need transparency, target a transparency-aware format instead of SVG.

Embedded ICC color profiles are read from the source PNG and re-attached to the SVG output where the format supports it (JPG, PNG, TIFF, WebP). Formats without profile support fall back to sRGB.

Camera EXIF (ISO, shutter, lens, GPS) is preserved by default during PNG → SVG conversion when both formats support metadata. Use the privacy option to strip metadata before download if you want to share images without geolocation.

Yes — drag multiple PNG files into the upload zone and we queue them in parallel. Free users get 100MB per file; Premium has no per-file cap and runs more parallel workers, so a 200-image batch typically finishes in under two minutes.

For straightforward format conversion we run the same libpng / libjpeg-turbo / libwebp / ImageMagick pipelines a desktop editor uses, with the same quality. The difference is that desktop tools open one file at a time; we accept a batch and run them in parallel.

Default behaviour is 1:1 — your SVG output has the same pixel dimensions as the source PNG. Use a separate /resize-image/ step after conversion if you need to downscale or upscale as part of the workflow.

For JPG / WebP, quality 75-85 typically gets the file 60-80% smaller than the source PNG with no visible difference at normal viewing distance. For lossless targets (PNG, TIFF), expect smaller savings — usually 5-30% via better deflate / LZW compression.

Yes — uploaded PNG files are processed in isolated workers and deleted within minutes. We never read, store, or share the pixel data. The full privacy policy and retention window are documented at /privacy/.