Image resolution is often used to mean several different things. For screen work, the most concrete measure is pixel dimensions: width by height in pixels. For print, resolution usually describes pixel density, such as pixels per inch, together with a physical print size. Changing a metadata number does not necessarily create new visible detail.
The useful question is not “Is this 300 DPI?” by itself. Ask how many pixels the file contains, how large it will be displayed or printed, whether resampling is enabled, and how closely the result will be viewed.
Separate three measurements
Pixel dimensions are the grid of samples in the file, for example 2400 × 1600 pixels. Physical dimensions describe a print size, such as inches or centimeters. Resolution links the two by expressing how densely the available pixels are placed. The same pixel grid can be printed small at a high density or large at a lower density.
On a web page, layout CSS can display the same source at many sizes. Responsive-image markup may also let the browser choose among multiple files. That is why inspecting only the visible width in a page builder does not tell you the intrinsic pixel dimensions of the downloaded asset.
What resampling changes
With resampling off, changing print resolution redistributes the same pixels across a different physical size. With resampling on, software adds or removes pixels. Traditional interpolation estimates new values from neighboring pixels; learned super-resolution uses patterns learned from training data to propose plausible high-frequency detail under a degradation model.
Both approaches can make a file larger, but neither gives direct access to information that the camera never captured. A learned model may produce a more convincing texture than ordinary interpolation, yet that texture is an inference. For evidence, identification, text, or product-condition details, compare against the original rather than treating the enhanced output as ground truth.
Choose dimensions for the destination
For screens, start from the largest expected rendered size and consider higher-density displays. Avoid sending a huge master file when a smaller derivative will look the same in the layout, because unnecessary pixels increase transfer and processing cost. Keep an archival source separately from web derivatives.
For print, determine physical size, required pixel density, viewing distance, and the print provider's specification. A billboard and a photo book can tolerate very different pixel densities because they are viewed differently. Judge a test at realistic scale instead of relying on a single universal number.
Resolution decision checklist
Record the original pixel dimensions, intended output size, crop, and whether any resampling step is enabled. Inspect edges, text, faces, and repeated textures after resizing. If an AI upscaler is used, keep the original beside the output and label the result as enhanced when fidelity matters.
Do not repeatedly resize the same derivative. Return to the best available source, make one intentional transformation for the destination, and export in a suitable format. This limits accumulated blur, ringing, compression artifacts, and uncertainty about which version is authoritative.
Conclusion
Pixel count, physical size, and pixel density answer different questions. Check all three, identify whether resampling occurs, and evaluate the real destination. For the next step, read whether AI upscaling can recover detail or the image-upscaling topic guide.
Editorial note: This explainer was drafted with AI assistance, then checked and edited against the sources listed below on September 13, 2026. It does not report an unperformed product test.
Sources
Improve this guide
Report a problem or suggest the next question
Submissions enter a private editorial queue. They are reviewed and deduplicated before any public update or new page is created.