Image resolution for print: the dpi that actually works

dpi, ppi, and lpi: three numbers people mix up

Before the rule makes sense, it helps to separate three measurements that get flattened into the single word "dpi."

• PPI (pixels per inch) is a property of your image file — how many pixels of real detail it holds per inch at a given size. This is the number you actually control when preparing artwork.
• DPI (dots per inch) is a property of the output device — how many individual ink dots a press or printer can place per inch. A platesetter might image at 2,400 dpi while your photo is 300 ppi; the two describe different things.
• LPI (lines per inch) is the frequency of the halftone screen — how many rows of halftone dots the press uses to fake continuous tone. Commercial offset often runs around 150 lpi; flexo and coarse substrates run lower.

In everyday practice, nearly everyone says "dpi" when they mean the image's ppi, and mostly that's harmless. The problem is that the shorthand hides the relationship that explains where "300" comes from in the first place.

The 300 dpi rule — and why it's that number

A press can't spray a smooth gradient of ink. Each spot on the plate either takes ink or it doesn't. To reproduce a photograph — a face, a product shot, a soft shadow — the image is broken into a fine grid of dots that vary in size: big dots read as dark, tiny dots read as light, and from reading distance your eye blends them into continuous tone. That process is halftone screening, and its fineness is the screen ruling, measured in lpi.

Here's the link most people never see: to feed a halftone screen cleanly, you want roughly 1.5 to 2 image pixels for every halftone dot. Multiply a typical 150 lpi screen by two and you get 300 — 300 ppi. That's the entire origin of the "300 dpi" rule. It isn't a magic constant; it's shorthand for "enough pixels to satisfy a mid-fine halftone screen." Coarser processes (some flexo, corrugated post-print) are happy with less, and a few fine jobs ask for more, which is why the honest version of the rule is about 300 dpi at final size, unless your converter tells you otherwise. The screen ruling itself is tied to your color build and press — a topic we cover in spot color vs CMYK.

Effective resolution: the scaling trap

The single most common way good images print badly: the file is 300 dpi, but only at its original size. Resolution is a fixed pool of pixels. Placing an image larger in your layout spreads that same pool over more inches, so the real, effective resolution drops in direct proportion. A 300 dpi image dragged to 200% isn't 300 dpi anymore — it's 150.

The math is simply: effective dpi = original dpi ÷ scale factor. Here's what that does to a file that starts at a genuine 300 dpi:

The reverse is the good news: scaling a high-res image down raises its effective resolution, so oversized source images are always safer than undersized ones. When in doubt, start bigger.

Raster vs vector: know which one you're holding

Everything above applies to raster images — photographs and anything made of a fixed grid of pixels. There is a second kind of artwork that ignores the whole problem.

• Raster (TIFF, PNG, JPG, the photo inside a PDF) is a grid of pixels. It holds a finite amount of detail, so it's resolution-dependent and pixelates when pushed past its size.
• Vector (AI, EPS, SVG, the paths in a PDF) is defined by mathematical curves and points. It has no resolution — it redraws perfectly sharp at any size, from a business card to a billboard.

The practical rule: anything that's line art should be vector. Logos, type, barcodes, and the dieline itself should never be flattened into pixels. That's why the wordmark on a package looks razor-sharp while a low-res photo beside it looks fuzzy — one is resolution-independent, the other ran out of pixels. Keep type live and logos vector, and the only thing you have to manage resolution for is your actual imagery.

File formats, and when each is safe

Format doesn't guarantee resolution, but it tells you what to check.

The trap in this table is the middle rows: a PNG or a JPG can be perfectly high-res or hopelessly low-res, and the file extension won't tell you which. A PDF is the same — it's only as good as the images placed inside it. That's exactly the gap a preflight check is built to close.

Upscaling — and the honest truth about AI

When an image is too small, the tempting fix is to enlarge it in software. It's worth being clear about what that can and can't do.

Traditional upscaling (bicubic interpolation and its relatives) invents new pixels by averaging the ones around them. It makes the file bigger, but it adds no real information — the result is a smoother, softer version of the same limited detail. You can't recover detail that was never captured in the first place.

AI upscaling (super-resolution) is more sophisticated and genuinely useful in some cases, but it's important to understand what it's doing: it hallucinates plausible detail based on patterns it learned from other images. On a landscape or a texture that can look convincingly sharp. On the things packaging cares about — small type, logos, barcodes, precise edges — it can invent detail that isn't faithful to the original, subtly reshaping letters or smearing a code. It interpolates intelligently; it does not reproduce the real thing. Treat it as a last resort, use it only on continuous-tone imagery, never on type or codes, and always review the result at full size against the original. There's no substitute for capturing the image at the resolution you need.

How to check resolution before you send

You don't need special tools — just the pixel dimensions of your image and the size you'll print it. Two pieces of quick math cover almost everything:

• Is this image big enough? Divide pixel width by the print width in inches. A 1,800-pixel image printed at 6 inches is 1,800 ÷ 6 = 300 dpi — safe. Stretch that same image to 12 inches and it's 150 dpi — soft.
• How big can this image go? Divide the pixel dimension by 300. A 3,000-pixel image can print up to 3,000 ÷ 300 = 10 inches wide at full quality.

A few habits catch most low-res disasters before they start: never pull images off a website (screen graphics are typically 72 dpi and, just as often, copyrighted), be suspicious of any image that arrived by text message or chat app (they're usually compressed on the way), and always check the effective resolution after you scale, not before. The number in the file's metadata is only true at 100%.

How PackOS flags low-res images

Judging resolution by eye on a monitor is unreliable — a screen shows only around 100 pixels per inch, so an image that looks fine at 100% zoom can still be far short of what the press needs. When you upload artwork to PackOS, preflight measures each placed image's effective resolution at its final size — the real number after any scaling — and flags anything that falls below the threshold for your print process, before it becomes a reprint. It's the same idea as the rest of our complete guide to packaging print quality: catch the problem in the file, not on the pressroom floor. You can see how the checks run on the technology page, or run your own file through Quick Quote.