Image Resizer

Pixels set the size. DPI only labels the print.

An image file is a grid of pixels. A 4000 by 3000 phone photo holds 12 million of them, whatever DPI number sits in its metadata. A browser draws one image pixel per screen pixel, or one per two on sharp phone screens, and ignores the DPI tag completely. Change 72 DPI to 300 DPI and a web page shows the photo at the same size as before.

DPI matters on paper. The printer spreads your pixels across the page at a chosen density, so print size comes from one division:

print width (inches) = pixel width / DPI
1800 px / 300 DPI = 6 in
1800 px / 150 DPI = 12 in, softer up close

The Print at 300 DPI line under the preview runs this math for every result. Photo labs aim for 300. Posters viewed from across a room get away with 150. Below 150, prints look soft at arm's length.

Sizes worth knowing before you type numbers

Platforms revise their numbers every year or two. For feed and story sizes with crop guides per network, the social media image resizer keeps a dedicated preset list.

When the new box has a different shape

Switch the ratio lock off and type a width and height with a different shape from the photo, say a 4 by 3 picture into a 1200 × 630 box. Something has to give. New shape decides what.

Shrinking keeps real detail. Enlarging invents none.

Take a photo from 4000 pixels wide down to 1000. Each output pixel becomes the average of a 4 by 4 patch of source pixels. Fine detail drops away, yet everything left on screen came from the camera. Go the other direction, 1000 up to 4000, and every source pixel gets smeared across 16 output pixels while the browser blends the gaps. Edges go soft and small text turns fuzzy. The resizer flags every enlargement in amber under the preview so the trade is visible before you download.

An enlarged photo tends to look acceptable on a phone and poor on a monitor. If a form demands a minimum size bigger than your file, enlarge by the smallest amount the form accepts, then check the result with Actual pixels.

Why big reductions happen in halves here

Ask a browser to draw a 6000 pixel image straight into 600 pixels and some engines sample only a fraction of the source. Fabric, brick, window blinds and screen photos break into rippling bands called moire. This page halves the image repeatedly instead, 6000 to 3000 to 1500, and makes one final draw to 600. Each halving averages neighbors, so the bands never get a chance to form.

Load test pattern puts a 2400 by 1600 chart in the tool. Set Longest side to 300, switch to Actual pixels, and compare the stripe block with the same chart through any other resizer. A clean result shows flat grey. A poor one shows stripes or curved bands.

Getting under a KB limit

Job portals, exam registrations and visa forms often reject anything over a fixed size, commonly 100 KB, 200 KB or 500 KB. Type the limit into Max size and the page searches eight quality levels between 10 and your Quality setting, then keeps the highest one under the limit. The Quality used readout shows where the search landed.

Quality alone has a floor. Below about 40, JPEG blocks show on skin and smooth skies. When the readout says the file is still over the limit at the lowest level, cut pixels instead. File size follows pixel count, and pixel count follows the square of the width:

• 2000 × 15003.0 megapixels
• 1000 × 7500.75 megapixels, a quarter of the data
• 600 × 4500.27 megapixels, under a tenth

For a typical phone photo, 1000 pixels wide at quality 70 lands near 100 KB. PNG has no quality setting, so Max size works with JPEG and WebP only. Screenshots and flat graphics are the exception where PNG often comes out smaller than JPEG anyway. For squeezing an image without touching its dimensions, the image compressor fits better.

Where this resizer stops