Is the conversion truly lossless?
Yes — the JPG → PNG step is lossless. PNG uses DEFLATE (zlib) compression, the same algorithm as ZIP, which preserves every byte. However, the original JPEG was already lossy — the JPEG encoder discarded high-frequency detail when the photo was first saved. Converting to PNG cannot recover that detail; it only ensures no further loss happens. The PNG is identical to what your browser sees when it decodes the JPG.
Why is the PNG much larger than the JPG?
JPEG is designed for photographs and uses lossy DCT compression that achieves 5–10× smaller files than lossless formats. PNG must store every pixel exactly, with no perceptual shortcuts, so a typical photo PNG ends up 3–8× larger than the source JPG. This is normal. If size matters more than losslessness for further saves, keep the JPG. If you need a format that supports transparency or further editing without compounding compression artefacts, use the PNG.
Is my JPG uploaded to your servers?
No. Conversion runs entirely in your browser using HTML5 Canvas and the native toBlob API. The JPG is read with FileReader, decoded by the browser, drawn to a canvas, and re-encoded as PNG. Open DevTools → Network and click Convert: zero requests are made. Family photos, ID scans, and confidential documents stay on your device.
Why convert JPG to PNG at all?
Common reasons: (1) you need transparency support for further editing — PNG has alpha, JPG does not; (2) you plan to edit and re-save multiple times — every JPG re-save compounds quality loss (generation loss); (3) the destination platform requires PNG (some print providers, archival systems, scientific workflows); (4) you need pixel-exact comparison — PNG hashes match if and only if pixels match; (5) you are converting for OCR or vision pipelines where DCT artefacts hurt text recognition.
Can the PNG be smaller than the JPG?
Rarely, and only for sparse images — large solid-colour regions, line art saved as JPEG, or screenshots that JPEG handles inefficiently. For photographs the answer is almost always no. If the JPG was saved at very low quality (heavy DCT compression), the PNG of that already-degraded image will still be larger. Run-length and DEFLATE compression simply cannot match perceptual lossy schemes for natural photos.
Will EXIF metadata be preserved?
No. Canvas-based conversion strips all metadata — EXIF (camera make, exposure, GPS coordinates), IPTC (captions), XMP (Adobe metadata), ICC colour profiles, and ID3 chunks. The output PNG contains only pixel data plus the standard PNG IHDR/IDAT/IEND chunks. This is often desirable for privacy: stripping EXIF removes location data and camera identity. If you need to preserve metadata, use a server-side tool like ImageMagick with -strip omitted, or a dedicated metadata-aware converter.
How big a JPG can this tool handle?
Browser memory and the maximum canvas dimensions are the constraints. Most browsers support canvases up to 16,384 × 16,384 px (Chrome) or 11,180 × 11,180 px (Safari). A 6000 × 4000 (24 MP) photo decodes and re-encodes in 2–4 seconds on a modern laptop. Be aware: a 24 MP RGBA buffer is 96 MB in RAM, and the resulting PNG can be 50–100 MB on disk. There is no server-imposed file size limit because no upload happens.