Resizing
How to Resize an Image Without Losing Quality
Resizing is the most misunderstood operation in image editing. People lower the DPI number and expect a sharper file, or enlarge a small photo and expect detail to appear. Neither happens. What resizing changes is the number of pixels; what it costs depends on direction and method.
Downscaling and upscaling are not symmetrical
Downscaling removes pixels. The software has more information than it needs, and the job is to combine neighboring pixels without creating aliasing: jagged edges, moiré patterns, or shimmering in fine textures. A good resampler averages across the source area rather than sampling a single pixel.
Upscaling invents pixels. The software has less information than the output requires, so it interpolates between existing values. Edges soften, texture turns plastic, and no amount of sharpening recovers detail that was never captured. A 600-pixel-wide photo enlarged to 2400 pixels is still a 600-pixel photo; it just takes up more space.
This site’s resizing pipeline uses high-quality smoothing and, for large reductions, steps the image down in halves before the final draw. That matters because reducing 4000 pixels to 400 in one operation can skip detail and produce aliasing. Halving repeatedly averages more of the source and gives a cleaner result. The trade-off is a little processing time, not quality.
DPI is metadata, not resolution
A JPEG or PNG may carry a density value, often expressed as DPI or PPI. It tells a print application how large to place the image on paper. It does not change how many pixels exist.
The formula is simple:
print size in inches = pixels ÷ pixels per inch
A 3000 × 2400 image printed at 300 PPI is 10 × 8 inches. The same file printed at 150 PPI is 20 × 16 inches, with the same pixels spread over four times the area. Changing the embedded DPI from 72 to 300 does not add a single pixel; it only changes the default size the printer assumes.
That is why “set it to 300 DPI” is incomplete advice. Decide the physical print size and the required PPI, then calculate the pixel dimensions you need.
| Print size | At 300 PPI | At 150 PPI |
|---|---|---|
| 4 × 6 in | 1200 × 1800 | 600 × 900 |
| 5 × 7 in | 1500 × 2100 | 750 × 1050 |
| 8 × 10 in | 2400 × 3000 | 1200 × 1500 |
| 8.5 × 11 in | 2550 × 3300 | 1275 × 1650 |
| A4 | 2480 × 3508 | 1240 × 1754 |
For a poster viewed from a meter away, 150 PPI is often enough. For a hand-held print, 300 PPI is a safer target. For a screen, pixels are the only thing that matters; a 1600-pixel-wide image is 1600 pixels wide on every monitor regardless of its DPI tag.
Resize a batch in the browser
The image resizer exposes the three modes most tasks actually need.
- Open the tool and add your files. Batches are supported, and nothing is uploaded.
- Choose a mode:
- Exact dimensions when a form or a layout specifies width and height.
- Percentage when you want a consistent scale across a mixed batch.
- Longest side when you want every image to fit inside a maximum, regardless of orientation.
- Leave aspect ratio locked unless you deliberately want to stretch the image. Unlocking it changes the shape of faces and objects.
- Run the batch, then check the reported output dimensions. A landscape and a portrait image given the same longest-side limit will have different width and height, which is usually what you want.
- Download files individually or as a zip.
If the destination also has a byte limit, resize first and compress second. Compressing a full-resolution file and then resizing it wastes the first encode and adds a generation of loss. The 100 KB compression guide covers the second step.
Sharpen after downscaling, lightly
Downscaling averages adjacent pixels, which reduces local contrast. A photograph that looked crisp at full size can look slightly flat after a large reduction. A small unsharp mask restores the impression of edge definition without adding real detail.
Reasonable starting values for a downscaled photograph are a radius of 0.5 to 1 pixel and an amount around 50 to 100 percent. Apply it at the final size, not before resizing. Overdoing it produces bright halos along high-contrast edges, which look worse than mild softness. The editor has filter controls if you need this step. The preset resizer is built for dimension changes, not retouching, so plan on a separate sharpening pass when a large reduction looks flat.
Common mistakes
Resizing a JPEG twice. Each save re-encodes. Resize once, at the final dimensions, and export once.
Using percentage on a mixed batch. Fifty percent of a 6000-pixel photo and 50 percent of an 800-pixel graphic produce very different results. Use longest side when the goal is a consistent maximum.
Upscaling a small image for print. A 1000-pixel-wide file cannot become a crisp 300 PPI A4 print. The pixels are not there. Choose a smaller print size or a lower PPI.
Assuming PNG is always better. For a photograph, PNG preserves pixels but produces a much larger file than JPEG at a visually similar quality. Use PNG when lossless pixels or transparency matter, not as a default.
Resizing pixel art with smoothing. Hard-edged sprites and some UI screenshots need nearest-neighbor scaling to keep crisp blocks. This pipeline uses smoothing, which blurs them. If that is your task, use an editor with a nearest-neighbor option instead.
Match the output to the destination
Screen layouts rarely need more than 1600 to 2000 pixels on the long side. Social platforms have their own recommended dimensions, and uploading a much larger file usually results in the platform downscaling it with its own algorithm, which you do not control. The 2026 social media size guide lists the common presets and the safe areas that platforms crop.
For a form that specifies both dimensions and a byte limit, do the operations in this order: crop to the required aspect ratio, resize to the required pixels, then compress to the byte target. Each step has one job, and you can verify the result before moving on.