DPI is a print-density instruction stored in a file's metadata, not a measure of image detail, and Microsoft Paint cannot change it without resampling. Every time Paint's Resize dialog touches the DPI value, it also recalculates the pixel count and saves a recompressed copy — a 3000 by 2400 pixel photo becomes a smaller or larger file depending on the percentage or absolute size you enter, and the original pixels are gone. That default behavior is the reason most readers searching for a Paint-based DPI workflow end up with blurry, stretched, or upscaled files that no longer match the source. The clean fix is to leave Paint out of the metadata step and use a purpose-built local tool. The Image DPI Changer writes a new print-density value into a JPEG or PNG and leaves the pixel dimensions and image bytes untouched, so the file you download has the same 3000 by 2400 pixels whether you set 72, 150, 300, or 600 DPI.
Why Microsoft Paint Can't Change DPI Without Resampling
Microsoft Paint has never exposed a standalone "DPI" field. The only place DPI appears is inside the Resize and Skew dialog (Home → Resize, then untick "Resize image dimensions" in older builds). When that checkbox is off, the dialog lets you change the horizontal and vertical DPI values; the moment you click OK, Paint recomputes the pixel canvas based on the new DPI and the physical size you typed. There is no "edit metadata only" option, no JFIF inspector, and no PNG pHYs field to write to. The result is always a re-rendered canvas.
This matters because most readers open Paint expecting to do what they can do in a text editor — change a label without disturbing the content. Image metadata does not behave that way in Paint. If you type 300 DPI for a 3000 by 2400 pixel image at a 10 inch target width, Paint multiplies 300 × 10 to get 3000 pixels. If you then change the DPI to 150 at the same 10 inch width, Paint recomputes the canvas to 150 × 10 = 1500 pixels wide. The pixel count literally follows the DPI, which is the opposite of a metadata-only edit.
Two side effects follow. First, the original pixels are discarded and replaced with a nearest-neighbor resample, which is why large DPI reductions look blocky and large DPI increases look soft. Second, Paint then re-encodes the canvas as a JPEG or PNG through its own save path, which adds another round of compression artifacts. Readers who want a true no-resample workflow should treat Paint as a viewer rather than an editor for this task and reach for a browser tool that edits the file header directly. The companion guide resizing an image in Paint? Try your browser instead walks through the same pivot in more detail.
What "Changing the DPI" Actually Does to a File
A JPEG stores its print-density instruction in a JFIF APP0 marker, specifically in three small fields: a units byte, an Xdensity word, and a Ydensity word. When the units byte is 1 (dots per inch), printers and layout programs divide the pixel dimensions by the density fields to compute a physical size in inches. When the units byte is 0 (no unit) or 2 (dots per centimetre), the math shifts but the purpose is the same — a label, not an image transformation. The JFIF format is fully documented in the W3C JPEG File Interchange Format 1.02 specification.
A PNG stores the same information in a pHYs chunk. The pHYs chunk holds a single unsigned 32-bit integer: pixels per metre. Programs that understand the chunk convert pixels per metre into DPI by dividing by 39.3700787, which is the exact 0.0254 metre-per-inch relation. Because the field is an integer, a value such as 300 DPI becomes 11811 ppm and a value such as 150 DPI becomes 5905 ppm — neither round-trips exactly. That is a storage precision limit of the PNG container, not a change in the file's actual pixels. The pHYs chunk is documented in the W3C PNG Specification, Third Edition.
In both formats the metadata is portable, documented, and read by every mainstream print and layout program. Web browsers usually ignore it because CSS dimensions take precedence, but Photoshop, InDesign, Affinity Publisher, GIMP, and most RIP software will read JFIF or pHYs to populate the "Document Size" or "Print Size" panel. Updating the metadata is what tells those programs how many pixels per inch to assume when you place the image at a target physical size.
Set a New DPI With the Image DPI Changer
- Open the Image DPI Changer in your browser and click the file input. Pick a single JPEG or PNG up to 25 MiB; the tool validates the file signature and shows any print-density metadata it detected on the way in.
- Read the detected DPI value in the result panel. If you see "no usable JFIF header" or "no pHYs chunk", the original file had no print-density metadata and the output will add one from scratch.
- Type a whole-number target DPI into the input box. Common values are 72, 150, 300, and 600; pick the one that matches your target physical size for the pixels you already have.
- Click Update image DPI. The tool writes a new JFIF units/Xdensity/Ydensity field for JPEG or a new pHYs pixels-per-metre value for PNG, copies the image bytes unchanged, and produces a local Blob download.
- Save the new file with a clearly different name (for example, "photo-300dpi.jpg") so you keep the original alongside the updated copy.
- Open the downloaded file in your layout program, EXIF viewer, or print software to confirm the new DPI is reported. For PNG, expect the readout to be the requested DPI rounded to the nearest representable pixels-per-metre — that is normal format behavior.
JPEG vs PNG: Where the DPI Lives
The two formats store the same logical information in different structures, and understanding the difference helps you predict what a downstream program will see.
| Property | JPEG (JFIF) | PNG (pHYs) |
|---|---|---|
| Metadata location | APP0 marker right after the SOI marker | pHYs chunk placed before the IDAT chunk |
| Value type | Three small integers: units byte, Xdensity, Ydensity | One unsigned 32-bit integer in pixels per metre |
| Unit declaration | Explicit (1 = DPI, 2 = DPC, 0 = aspect ratio) | None — always pixels per metre |
| DPI conversion | Direct when units = 1 | Divide by 39.3700787 (exact 0.0254 m/in) |
| Round-trip precision | Exact for whole-number DPI | Approximate; integer ppm can shift DPI by a very small rounding fraction |
| Behavior when missing | Tool adds a small JFIF header | Tool adds a pHYs chunk |
| Result on invalid input | Operation rejected; previous result cleared | Operation rejected; previous result cleared |
Pick the Right DPI for Your Print Job
The right DPI is a function of the pixels you have and the physical size you need, not a magic threshold. The formula is:
print size in inches = pixel dimension ÷ DPI
So a 3600 pixel wide image set to 300 DPI prints at 3600 ÷ 300 = 12 inches wide. Set the same image to 150 DPI and it prints at 24 inches wide — same pixels, larger physical page. The Image DPI Changer's only job is to put that "300" or "150" into the metadata; the math is yours to do beforehand. The companion guide how to convert DPI and pixel dimensions to print size covers the same relationship with more worked examples.
A practical starting point:
- 300 DPI is the conventional value for high-quality photo prints, magazine pages, and most print-shop submissions.
- 150 DPI is fine for large-format banners, draft proofs, and screen-on-paper workflows.
- 72 DPI matches the historical Macintosh screen convention; web pages rarely need this label because browsers ignore print metadata in most contexts.
- 600 DPI is overkill for photo printing but useful for line art, technical drawings, and embroidery digitizing.
If you already know the physical size your printer or layout program expects, work backwards: target DPI = pixel dimension ÷ inches. For a 3000 by 2400 pixel image intended for a 10 by 8 inch print, 3000 ÷ 10 = 300 DPI. Enter that value into the Image DPI Changer, download the file, and place it in your layout software at the intended physical size.
Limitations and When to Reach for a Different Tool
The Image DPI Changer edits one field per file — the JFIF units/density trio for JPEG or the pHYs chunk for PNG — and nothing else. If your JPEG also carries EXIF resolution tags, XMP, IPTC, ICC, GPS, camera, or vendor extension blocks, the tool dynamically reads the standard EXIF resolution fields when you choose a file but does not attempt to merge, rewrite, or reconcile every metadata family downstream programs may consult. The W3C JFIF specification covers the JPEG output and the W3C PNG Specification covers the PNG output; both are documented, portable, and widely supported. Other metadata blocks remain as the source file contained them.
The tool is also not a pixel editor. It cannot add detail, sharpen a blurry scan, enlarge a small image, or convert a JPEG to a PNG. For those jobs the right tool is an image resizer, a JPEG/PNG converter, or an image compressor. A related workflow is to convert between DPI and pixel dimensions without writing a file at all — the DPI Converter lets you plan the print size first, then come back to the Image DPI Changer with a confirmed value.
Finally, the destination matters. If you place a 300 DPI JPEG into a layout program that prefers EXIF resolution, the program may still report the old value. Always verify the printed size in the application that will actually render or print the image. The metadata is a hint, not a command.