Your Image Says 72 DPI, and That Number Cannot Hurt You
DPI is a note stored beside the pixels rather than a property of them, which is why one file reports three different numbers in three programs. What actually decides print quality, why typing 300 into a box changes nothing, and where the real ceiling sits.
By Novus Stream Solutions Editorial Team · Part of the Image processing and on-device AI topic cluster
A print shop looks at your file and says it is 72 DPI and they need 300. You open it somewhere else and it says 96. You open it in a third program and there is no number at all. Meanwhile the image is the same image, pixel for pixel, in every one of them.
This confuses an enormous number of people, and the confusion is entirely reasonable, because software presents DPI as though it were a property of the image. It is not. It is a note attached to the image, and the note can be wrong, absent, or ignored without anything happening to a single pixel.
What the number actually is
An image file contains pixels. A PNG can also contain a small optional chunk, and a JPEG a small optional header field, that says something like "if you print this without being told otherwise, render it at this many pixels per inch."
That is the whole content of the claim. It is a default suggestion for a printing decision, stored alongside the picture. It does not describe the picture, it does not change when the picture changes, and nothing is obliged to honour it.
If the field is absent, which is very common, each program falls back to its own assumption. Some assume 72, some assume 96, some show nothing. That is why the same file reports three different numbers in three different places: you are reading three different defaults, not three different measurements.
The number that does describe the image
Pixel dimensions. That is it. A file is 2400 by 1800 pixels, and that fact is not open to interpretation.
DPI only becomes meaningful when you combine it with a physical size, and then it is arithmetic:
pixels = print size in inches x dots per inch
A photograph printed 8 inches wide at 300 dots per inch needs 2400 pixels of width. The same file printed 4 inches wide is 600 dots per inch and looks superb. The same file printed 24 inches wide is 100 dots per inch and looks soft. One file, three answers, and the tag in the header had nothing to do with any of them.
This is why "is my image 300 DPI" is not a question that can be answered. "Is my image big enough to print at this size" can be.
Why setting DPI to 300 in an editor does nothing
Open an image editor, find the resolution field, type 300, save. The file now says 300. The print will look exactly as good or as bad as it did before.
You changed the note. There is a checkbox in that dialogue, usually called resample, and it is the one that matters. With resampling off, you are editing the note and nothing else. With resampling on, the editor will generate new pixels to hit the number, which is upscaling, and it is a real operation with real consequences that have nothing to do with the tag.
If a print shop's checker is looking at the tag rather than at the dimensions, then typing 300 into that box will indeed make their checker happy, and the print will be exactly as soft as it was going to be. That is a bad checker, not a fixed file.
What these tools do about it
They work in pixels and they say so.
The Image Resizer asks for width and height in pixels, shows you what you will get, and names the output file after its actual dimensions so that the number travels with the file. The Canvas Extender adds pixels of padding. The AI Image Upscaler gives you 2x or 4x the pixels you had. Nothing anywhere asks you for a DPI.
Nothing here writes a resolution tag into an export either. That is worth saying plainly: your file will arrive somewhere and be reported as whatever that program's default is, because the field is not there to read. If a workflow of yours genuinely requires the tag to be present and set, you will need a step outside these tools to set it, and that step changes only the note.
One related detail, since people ask. Stripping metadata will not disturb this. The Metadata Remover targets EXIF, XMP, ICC profiles, C2PA provenance, IPTC blocks, and text chunks. A JPEG's structural header, which is where a JPEG's density fields live, is never dropped, because dropping it would produce a broken file.
Sizes worth remembering
At 300 dots per inch, which is the usual assumption for print you hold in your hand:
- A 4 by 6 inch photo print needs 1200 by 1800 pixels.
- An 8 by 10 needs 2400 by 3000.
- A4 needs roughly 2480 by 3508.
- A 12 inch wide t-shirt graphic needs 3600 pixels of width.
Large-format work such as banners and billboards is viewed from a distance and is routinely printed at far lower densities, which is why a billboard does not need a hundred-megapixel source.
When you genuinely do not have enough pixels
Upscaling is the honest answer, with an honest caveat.
The upscaler will give you the pixel count. It routes to different approaches depending on the source, using a restoration-oriented model for compressed photographs, edge-preserving scaling for logos and flat artwork, and a different model for clean photographs, and the results on a modest enlargement are genuinely good.
What it cannot do is recover detail that was never captured. A 600-pixel web thumbnail enlarged to 3600 pixels has 3600 pixels of width and the information content of a 600-pixel image. For a t-shirt graphic where the art is flat shapes, that can be fine. For a photograph printed at size, it will look like what it is.
The reliable answer remains the boring one: find the original file.
The ceiling that applies here
Images are capped when they are decoded, at 4096 pixels on the long side on a desktop machine and lower on phones and low-memory devices. At 300 dots per inch, 4096 pixels is about 13.6 inches.
That is an honest boundary rather than a hidden one. These tools are built to do real work inside a browser tab without crashing it, and holding a 100-megapixel print file plus its working copies in a tab is how a tab dies. For posters and packaging beyond that size, this is not the pipeline, and it would be dishonest to pretend otherwise.
What to tell a printer
Give them the pixel dimensions and the size you want it printed. "It is 3600 by 3600 pixels and I want it 12 inches square" is a complete, unambiguous statement that lets anyone compute the density.
"It is 300 DPI" is not a complete statement, which is why the conversation so often goes in circles.
Applies to NSS Background Remover v2.6.2, the release that was current when this article was published.
Was this article helpful?
Your answer is saved in this browser only. This control makes no network request and does not send the answer to an NSS feedback endpoint.