Image DPI Changer

DPI is not part of the picture. It is a number stored in the file's metadata that says how many image pixels the author intends to fit into one inch when the image is printed or placed in a page layout. A 3000-pixel-wide photo is always 3000 pixels wide; at 300 DPI it is meant to print 10 inches wide, and at 150 DPI it is meant to print 20 inches wide. Changing the number does not add or remove a single pixel.

That is exactly what this tool changes. It opens the file, finds the density record, rewrites it, and copies every other byte through untouched — the compressed image data is never decoded, re-encoded or re-compressed, so there is no quality loss of any kind. In a PNG the number lives in a `pHYs` chunk, stored as pixels per metre (the value is multiplied by 0.0254 to get DPI, because one inch is exactly 0.0254 m). In a JPG it lives in the JFIF APP0 segment, stored as a unit code plus an X and a Y density; a dots-per-centimetre value is multiplied by 2.54 to get DPI.

Because only metadata changes, the pixel data stays identical and the file size barely moves: a file that already records a density keeps its exact length, and a file that records none grows by 18 bytes (JPG) or 21 bytes (PNG) for the new record. Note that DPI cannot make a picture sharper. Asking a 500-pixel-wide image to be 300 DPI only tells a printer to reproduce it 1.67 inches wide — the detail is still 500 pixels. If you need the picture to be physically larger at the same DPI, you need more pixels, which is a resize (interpolation), not a DPI change.

Whether the new number is honoured depends entirely on the program reading the file. Browsers, social networks and most screen viewers ignore DPI completely — they draw by pixel dimensions. Print drivers, Adobe Photoshop and InDesign, GIMP, Microsoft Word and similar layout and print tools do read it, and that is where it matters: a photo for a 4 × 6 inch print, an image dropped into a report where the pixel dimensions would otherwise make it the wrong physical size, or a printer that refuses anything under 300 DPI. The tool writes the same number into both the X and Y density, so the aspect ratio is never distorted.

Two honest limits. First, this works on PNG and JPG only: WebP, GIF, HEIC, AVIF and TIFF put their density somewhere else, or not at all, and are not touched by this tool. Second, a JPG can also carry resolution tags inside a separate EXIF block; this tool rewrites the JFIF density, which is the block every reader understands, but it leaves any EXIF resolution tags as they were — a few programs read those first, so if your DPI appears unchanged in one specific application, that is the likely reason.

Everything happens in your browser: the file is read locally, changed in memory and handed back as a download. Nothing is uploaded, and no request carrying your image leaves the page.

1. The file's size is checked (50 MB) before it is read, and its leading bytes confirm it is really a PNG (89 50 4E 47 0D 0A 1A 0A) or a JPG (FF D8 FF). The name and MIME type are not trusted. 2. The existing density is read back. PNG: the `pHYs` chunk gives pixels per metre (unit specifier 1 = metre), so DPI = round(pixels-per-metre × 0.0254). JPG: the APP0 JFIF segment gives a unit code and an X density, so DPI = the value for unit 1 (dots per inch) or round(value × 2.54) for unit 2 (dots per centimetre). 3. The new value is written. PNG: the pixel count to store is round(dpi ÷ 0.0254) — 300 DPI is 11,811 pixels per metre — written into both the X and Y slots of a `pHYs` chunk, whose CRC32 is recomputed over its type and 9 data bytes. An existing `pHYs` chunk is overwritten in place (it is always 21 bytes on the wire); otherwise the new chunk is inserted directly after IHDR, where the spec places it. 4. JPG: an existing JFIF APP0 segment has its units byte set to 1 and its Xdensity and Ydensity fields set to the requested DPI; a file with no JFIF segment gets a fresh one (length 16, "JFIF\0", version 1.1, units 1, thumbnails 0) inserted directly after SOI, where JFIF requires it. 5. Every other byte — the entire compressed image, colour profiles, other metadata — is copied through unchanged, so the saved file's pixel data is byte-for-byte identical to the original's. The output is named like photo-300dpi.png or scan-300dpi.jpg.