Someone asks for a print-ready file at 300 DPI. The request sounds complete, it is made in good faith, and it can be satisfied by a file that will print badly or refused by a file that would have printed well. The problem is not the number.
DPI is dots per inch. It is a ratio, and a ratio needs both of its terms.
DPI is a ratio and the file holds one of its terms
A digital image contains a pixel count and nothing else about physical size. A file is 3,000 pixels wide. That is a fact about the file, permanent until someone edits it.
Inches are not in the file. Inches arrive when a person decides how wide the thing will be printed, and that decision usually happens later, elsewhere, by someone else. Until it is made, the file has no DPI at all — it has a numerator waiting for a denominator.
Work through it once and the whole article follows. A file 3,000 pixels wide, printed 10 inches wide, is 300 DPI. The same file, printed 20 inches wide, is 150 DPI. Nothing was edited between those two sentences. One file, two resolutions, because resolution was never a property of the file in the first place.
So “300 DPI” names half of a requirement, and the missing half is the one that determines whether the file works.
What the request is trying to say
The shorthand did not come from nowhere. It comes from an era of print production where the output width was effectively fixed by the process and the viewing distance, so naming the ratio was a reliable way to name the pixel count — everyone in the conversation already knew the denominator.
Apparel decor broke that condition. A file can go to a chest print, a sleeve hit, a full-front placement, an all-over panel, or a hangtag, and these do not share a width. The habit of naming the ratio survived into a context where the shared denominator no longer exists, which is why the request still feels precise and no longer is.
The shorthand is sticky for a reason beyond habit. It is portable, it sounds like a quality bar rather than a production detail, and it lets both sides of the conversation avoid naming a decision that has not been made yet. Nobody has to commit to how wide the graphic will sit on the garment, which is often genuinely undecided at the point the file is requested. The ratio absorbs that uncertainty and hides it, and the file gets made against an assumption that was never stated aloud by anyone.
Treated as a question rather than a spec, it is a good question. It just has to be finished: 300 DPI at what width?
The number in the DPI field does not do anything
Image editors show a resolution field, and files carry that value as metadata. This causes more confusion than any other part of the subject, because the field looks like a setting you can turn up.
It is a note the file carries about a printed width it would prefer. Two files with identical pixels and different values in that field are the same image and will print identically at the same width. Typing a larger number in there does not add detail, and some print workflows ignore the field entirely in favor of the width specified at output.
Generated and exported files make this worse by arriving with whatever value the exporting tool wrote by default, a number chosen by software that had no idea what the image was for. A file can therefore fail a review for declaring a low value while holding plenty of pixels, or pass one for declaring a high value while holding too few. Both mistakes come from reading the field as a measurement of the image instead of a note about intended output, and the second is the expensive one because nothing downstream will contradict it.
The practical consequence is that a file can be inspected, found to say 300, and approved — while the only question that mattered, how many pixels across the print will be, was never asked. An approval based on that field is an approval of a preference, not of a capability.
Resize and resample wear the same word
Two different operations both get called resizing, and the difference is the whole of the confusion.
• Changing the stated print size without touching the pixels moves the ratio and leaves the image untouched, so a file can be relabeled from one DPI to another in a second with no consequence whatsoever.
• Changing the pixel count actually rewrites the image, and when the count goes up the new pixels have to come from somewhere other than the original scene.
The first operation is arithmetic. The second is interpretation. Interfaces frequently put them behind the same dialog with a checkbox between them, and the checkbox is the most consequential control in the window.
Apparel has no single print width
Here is the part that is specific to garments, and it is the reason this cannot be solved by picking a number once.
A size run prints the same artwork on bodies of different sizes, and there are two common ways to handle that. Some brands hold the print width constant across all sizes, so the graphic looks progressively smaller on larger bodies. Others scale the print with the garment, so it keeps its visual proportion. Both are defensible and both are in use, and the second one means the same file is printed at several different widths within one order.
The two halves of the ratio are also held by different people at different moments. Whoever generates or draws the artwork fixes the pixel count, often before the product line is settled. Whoever sets the placement fixes the printed width, often weeks later and without seeing the file’s dimensions. Neither person is withholding anything; the decisions simply do not meet, and the ratio that results is nobody’s decision at all. This is why the requirement has to travel as written numbers rather than as a ratio — a ratio assumes both halves were decided in the same room.
Under the scaling approach, the file’s effective resolution is different on every size, and it is lowest on the largest garment. A spec written against the middle of the run is therefore correct for the middle and wrong at the top, and the failure is quiet — nobody rejects a print for being slightly soft, and the softest garments are a minority of the order. Whichever approach a brand uses, the number that governs is the widest width the file will ever be printed at, and that width is set where the graphic is positioned and sized on the garment: applying a print to a garment.
Where the ceiling actually is
Pixel count stops mattering at some point, and that point is a property of the surface rather than the file. A woven surface can only resolve marks down to its own grain, and beyond that, additional pixels are describing detail the cloth cannot hold.
This is why “as high as possible” is not a spec either. Past the surface’s limit, more pixels cost storage and processing time and buy nothing visible. The limit differs by fabric and decoration process, so the useful move is to get it from the supplier who will be printing rather than from a general rule. The broader version of this — each substrate taking something different away from the same artwork — is its own subject: one artwork across three substrates.
How to write the spec so it cannot be misread
A spec that survives being read by someone who was not in the conversation states pixels, and states the width they were calculated for.
Name the pixel dimensions. Name the widest printed width in the size run. Name the size that width belongs to, so the next person can check the arithmetic rather than inherit it. Written that way, the requirement is verifiable by anyone with a calculator, and it stays correct when the person who wrote it has moved on.
If the arithmetic comes out short, there are two honest routes and one that only looks like a route. Generating again at the required pixel count is the first. Reducing the printed width is the second, and it is more often available than teams assume. The third is resampling upward, which changes the arithmetic without adding anything to the image — worth understanding precisely before relying on it: what upscaling actually does.
Questions teams ask about print resolution
Is a 300 DPI file good enough for apparel printing? As written, the question cannot be answered, because it describes a ratio without giving the width the ratio was calculated at. Ask instead how many pixels wide the file is and how wide the print will be on the largest size in the run. Those two numbers settle it and the DPI figure follows from them.
What should a print spec say instead? Pixel dimensions, plus the widest printed width those dimensions were chosen for, plus which garment size that width belongs to. Everything else can be recalculated from those three facts by anyone who receives the file. A spec stated as a bare ratio cannot be recalculated by anyone, because the information needed to check it was never written down.
If I set the DPI field in my editor to a higher number, does the file improve? No. That field records a preferred print width and editing it moves no pixels. The image that comes out of the printer is determined by the pixel count and the width it is printed at, neither of which that field controls.
Why does the same print look fine on a small tee and soft on a 3XL? Because it is being printed wider on the larger garment while carrying the same pixels, so its effective resolution is lower there. This only happens when the print scales with the size run rather than staying a fixed width. It is the expected outcome of a spec written against an average size.
If the pixel count comes out short, can upscaling fix it? Upscaling changes the numbers to the ones the spec asked for without adding detail that was not captured. Sometimes that is acceptable, particularly on a surface that could not have resolved the missing detail anyway. It is a judgment about the substrate, not a repair of the file.
Is higher resolution always safer? Not past the point where the surface can resolve it, after which extra pixels are paid for in file size and processing and returned as nothing. The ceiling belongs to the fabric and the decoration process, so it comes from the supplier rather than from a rule of thumb. Below that ceiling, more headroom is genuinely useful, because it leaves room for the print width to grow later.
Where this leaves you
A resolution requirement stated as a ratio is missing the term that decides everything, and it will be read as complete by everyone who sees it. Pixels are a fact about the file; inches are a decision about the product, and the decision is what the spec exists to record. Write the pixel dimensions and the widest width they were chosen for, get the surface’s ceiling from whoever is printing, and the question of what DPI the file is becomes something anyone can work out and nobody has to ask.
Write the width into the spec, not just the ratio
Take your current print spec and check whether it states a printed width. If it names only a ratio, add the widest width the graphic will reach across the size run, and the size that width belongs to. Then confirm which convention your line uses — fixed print width across sizes, or a print that scales with the garment — because that choice determines which size governs the requirement. Set the placement and print size on the garment first, then check the pixels against it.
Written by
