Proportional downscale
Halving both dimensions keeps the shape and removes three-quarters of the pixels.
Input
3000 × 2000 photo · target 1500 × 1000Output
A 1500 × 1000 image with identical framing, at a fraction of the file weightResize any image to exact pixel dimensions in your browser — upload, set width and height, compare against the original, and download.
Content last reviewed
Drop an image or click to upload
JPG, PNG, WebP — never leaves your browser
Drop a JPG, PNG, or WebP file — its original dimensions load into the fields.
Set the width and height in pixels and click Resize.
Check the resized preview against the original, then download the result.
Halving both dimensions keeps the shape and removes three-quarters of the pixels.
Input
3000 × 2000 photo · target 1500 × 1000Output
A 1500 × 1000 image with identical framing, at a fraction of the file weightForcing a square target on a rectangular source distorts it — crop first instead.
Input
3000 × 2000 photo · target 1000 × 1000Output
A visibly squashed image — the side-by-side preview makes it obvious; crop to square in the Quick Image Editor first, then resizeEvery destination wants images at its own size: the form says maximum 800×800, the listing wants 1200×628, the print shop asks for exact pixels — and your source photo is whatever the camera produced. This tool scales any image to the precise dimensions you type, entirely in your browser: upload, enter width and height, click Resize, and compare the result against the original before downloading.
The width and height fields start at the image's real dimensions, which makes proportional resizing straightforward arithmetic: to halve a 3000×2000 photo, enter 1500×1000. The fields are independent — there is no automatic aspect-ratio lock — which cuts both ways: you can hit any exact target a platform demands, but entering a shape the original doesn't have will stretch or squash the picture. When the target shape genuinely differs from the source (a square avatar from a rectangular photo), crop to the new shape first in the Quick Image Editor, then resize.
One direction of resizing is free and the other is not: shrinking discards pixels and looks clean, while enlarging invents pixels the camera never captured and comes out soft — always start from the largest original you have. Resizing is also the single biggest lever on file weight (halving both dimensions removes 75% of the pixels), which is why resize-then-compress — finishing in the Image Compressor — is the standard web-image workflow.
Drop an image onto the upload zone or click to select one. Its original dimensions are shown and pre-filled into the fields.
Enter the target Width and Height in pixels. To keep proportions, scale both by the same factor (half of 3000×2000 is 1500×1000) — the fields do not lock the ratio for you.
Click Resize. The resized preview appears next to the original for comparison.
Adjust the numbers and resize again if needed, then click Download to save the result.
If the image is headed for the web, finish in the Image Compressor for the byte savings that resizing sets up.
Type any width and height — the output matches to the pixel, which is what upload forms and print shops require.
The source size is displayed and pre-filled, making proportional math trivial.
The resized preview renders next to the original so distortion or softness is visible before download.
The image is scaled on a canvas on your device — nothing uploads.
Hit the exact pixel dimensions a form, marketplace, or print service demands.
Produce a small version of a large source for previews and listings.
Scale images down to their actual display size — the biggest single lever on page load.
Bring a set of images to uniform dimensions one by one for a tidy grid.
The uploaded file is decoded into an image, drawn onto a canvas sized to your target width and height, and re-encoded in the source's format. Downscaling merges neighbouring pixels (which looks clean); upscaling interpolates new pixels between existing ones (which reads as softness — the information simply is not there). Because the canvas is set to exactly the numbers you typed, the output shape follows your input literally: matching proportions preserve the picture, mismatched ones stretch it. The preview renders the actual output blob, so what you compare is what you download.
Common web formats via drag-and-drop or file picker; the output keeps the source's encoding.
Free numeric fields, pre-filled with the original dimensions. No aspect-ratio lock — scale both values by the same factor to keep proportions.
Resizing runs on click; the result downloads from the preview.
The target shape differs from the original's. Keep the width-to-height ratio identical (scale both by one factor), or crop to the new shape first in the Quick Image Editor.
Expected — upscaling interpolates pixels that were never captured. A 2× enlargement is usually passable; 4× rarely is. Start from a larger original when one exists.
Check that both fields contain positive numbers — a zero or emptied field produces no usable canvas.
The width and height fields are independent — there is no aspect-ratio lock, percentage mode, or batch processing, so proportional resizing means scaling both numbers yourself and multiple images mean multiple passes. Enlarging beyond the source size produces soft results by the nature of interpolation, not as a defect of the tool. Resizing cannot change an image's shape without distortion; cropping (Quick Image Editor) is the right tool for shape changes.
The image is decoded, scaled, and re-encoded on your device — it never uploads to a server. Close the tab and nothing persists.
Scale width and height by the same factor to preserve proportions — there is no automatic lock.
Resize down, not up; enlargement past the original size always costs sharpness.
Crop to the target shape first when it differs from the original, then resize.
Size web images to their largest actual display size (2× for retina), not the camera's output.
Follow resizing with compression for web use — the two savings multiply.
Start from the destination, not the source. Find the real display size — the CSS slot, the platform spec, the print dimensions — and make that (or 2× it for high-DPI screens) your target. Camera-native dimensions belong in your archive, not on your pages.
Do shape before size. If the destination's aspect ratio differs from the photo's, crop deliberately first so you choose what the new frame keeps; resizing into a different shape is how heads get squashed. The Aspect Ratio Calculator gives you matching numbers when you know one dimension and the ratio.
And keep the pipeline order: original → crop → resize → compress → publish, with the original retained. Every step is cheap to redo from the master and expensive to undo from the output.
The file is loaded via an object URL into an Image element and drawn with drawImage onto a canvas sized to the entered width and height — a single-step scale using the browser's built-in interpolation. Output is produced by canvas.toBlob in the source's MIME type and previewed from the resulting object URL. Processing is on-demand (button-triggered) and entirely client-side.
Anyone who has been told "the image must be exactly W×H pixels" — form fillers, marketplace sellers, print customers — and site owners scaling photos to their display size before compressing. Users who want guided platform presets should use the Social Media Image Resizer; this tool is the precise, any-dimensions version.
Keep the width-to-height ratio the same as the original: divide both dimensions by the same factor. The fields don't enforce this for you, so 3000×2000 → 1500×1000 is safe while 3000×2000 → 1000×1000 will squash — for shape changes, crop first.
Enlarging cannot create detail — the browser interpolates new pixels from existing ones, which reads as softness. Two-times upscaling is usually acceptable, four-times rarely. The only real fix is starting from a larger original.
Dramatically — file weight scales with pixel count, so halving both dimensions removes about 75% of the data before compression even starts. For oversized photos it is a bigger lever than any quality slider; use both together.
The largest size they actually display: hero images around 1600–2000 px wide, content images 800–1200 px, thumbnails at their rendered size (doubled for retina screens). Shipping camera-native 4000 px files into small slots wastes bandwidth for nothing.
Not in this tool — it processes one image at a time for precise control. Batch workflows are better served by compressing in bulk after sizing, or by the Social Media Image Resizer when the targets are platform presets.
No — decoding, scaling, and re-encoding all happen on a canvas in your browser. The image never leaves your device.