Compressing images for print vs web

Web images are constrained by bandwidth and by a viewer looking at a backlit screen from arm's length, usually while scrolling. The goal is the fewest bytes that still look right in that context, and there is a great deal of headroom — the eye is forgiving of detail it never dwells on.

Print is constrained by ink on paper, examined at whatever distance the piece invites, permanently. There is no download to optimise and no second chance. The goal is to give the press as much correct information as possible and let it decide what to do with it.

Once you see it that way, every reversal below follows.

For the web you work backwards from the layout — the widest the image can be displayed, doubled for high-density screens. Beyond that, extra pixels cost bandwidth and show nobody anything.

For print you work forwards from the physical dimensions: roughly 300 pixels for every inch of final size. A 6 × 4 inch photo needs about 1800 × 1200; an A4 page about 2480 × 3508.

Large-format work is the exception — a poster viewed from three metres, or a billboard from fifty, needs far less than 300 PPI because the viewing distance does the work. Ask the printer rather than assuming.

Browsers work in sRGB and handle it universally. Anything wider is a gamble across the range of devices real visitors use, so the web answer is: convert to sRGB, embed it, done.

Print does not use RGB at all. Presses mix cyan, magenta, yellow and black subtractively, and the range of colours that produces overlaps with but differs from any screen's. Some vivid screen colours simply cannot be printed, and some printed colours cannot be shown on screen.

So print work is usually supplied in Adobe RGB — wider than sRGB, converted to CMYK by the printer — or in CMYK directly if they ask for it. Converting a print master to sRGB throws away gamut you were meant to use.

Quality 80 is right for the web because artefacts at that level are invisible on a screen at normal viewing size. On paper, examined closely, the same artefacts can be visible — and a press amplifies rather than hides them, particularly in smooth areas and skin tones.

For print: TIFF if the printer accepts it, or JPEG at quality 92 or above. The file will be large. That is correct, and it is not a problem to solve.

Never send a print shop an image you have optimised for the web. It has been downscaled, converted to sRGB, and had detail discarded that the press would have used.

The workflow that avoids every problem here is simple: keep one high-resolution master in a wide colour space, and generate everything else from it.

Web copies are downscaled, converted to sRGB, stripped of metadata and compressed. Print copies are supplied at full resolution with the colour profile intact. Neither is derived from the other, because each conversion is lossy in a direction the other one needs.

The failure mode to avoid is round-tripping: compressing for the web, then later trying to prepare that file for print. The pixels and the gamut are already gone.