Both numbers are used exactly as entered
There is no aspect ratio lock. Enter a width and height that do not match the original proportions and the image stretches or squashes to fit them, without warning.
That is worth stating first, because the [msm_tool slug="image-resizer"] will do exactly what you asked rather than what you probably meant. A 1600×900 photograph resized to 800×600 comes out visibly squashed, and it is the kind of distortion that is obvious to everyone except the person who made it.
Work out the second dimension first
The arithmetic is one line: divide the original height by the original width, then multiply by your new width.
A 1600×900 image resized to 800 wide needs a height of 900 ÷ 1600 × 800 = 450. The Aspect Ratio Calculator in this category does exactly that if you would rather not.
Deliberate distortion is occasionally useful — stretching a texture, squashing a background — but it should be a decision rather than an accident.
Reducing is safe, enlarging is not
Making an image smaller discards information cleanly. The result is genuinely sharp, because there was more detail available than the new size needs.
Making it larger cannot invent detail that was never captured. The same information is spread over more pixels, and the result is soft. No setting changes this, and the tools that appear to do it well are guessing at plausible detail rather than recovering real detail.
The practical rule: always start from the largest version you have. Resizing down from an original is fine; resizing up from something already reduced compounds the loss.
Large reductions benefit from two steps
Shrinking to a small fraction of the original in one pass can leave fine detail muddy, because the resampler is averaging across a large number of source pixels for each output pixel.
Two moderate steps — say 100% to 50%, then 50% to 25% — frequently look better than one drastic one. It is not a rule for every image, but if a heavily reduced version looks worse than expected, this is the first thing to try.
PNG output, and what it costs
The result is a PNG: lossless, and transparency survives. It is also large. A resized photograph will often be several times the size of the JPEG it came from, despite containing fewer pixels.
Resize first, convert last. One re-compression at the end of the process costs far less than compressing at every step.
The sizes worth knowing
Rather than resizing to whatever looks right, resize to what the destination actually uses:
- Full-width hero images — 1920 or 2400 wide is ample for almost any site. Larger is bandwidth nobody sees.
- Content images — twice the display width, so 1600 for an 800-pixel column, which covers high-density screens.
- Thumbnails — twice the display size, and no more. This is where oversized images most often hide.
- Email — 600 wide is the long-standing convention and still holds.
Serving a 4000-pixel photograph into a 400-pixel slot is one of the most common performance faults on the web, and it is entirely invisible until someone measures it on a phone connection.
Resizing is not the only lever
If the goal is a smaller file rather than smaller dimensions, converting to WebP usually achieves more than resizing does, and without losing any detail. The two together — correct dimensions, then a modern format — typically reduce a photograph by an order of magnitude with no visible difference.