Image Compressor
Shrink JPG, PNG and WebP with a quality slider and live size readout.
Image & Photo
Resize to exact pixels or a social preset with aspect ratio locked.
Every platform re-compresses what you upload, so supplying the exact native dimension avoids a second lossy resize on their servers.
| Platform / placement | Pixels | Aspect |
|---|---|---|
| Instagram square post | 1080 × 1080 | 1:1 |
| Instagram portrait post | 1080 × 1350 | 4:5 |
| Instagram / TikTok story | 1080 × 1920 | 9:16 |
| YouTube thumbnail | 1280 × 720 | 16:9 |
| YouTube channel banner | 2560 × 1440 | 16:9 |
| X / Twitter in-stream | 1600 × 900 | 16:9 |
| LinkedIn shared image | 1200 × 627 | 1.91:1 |
| Facebook shared link | 1200 × 630 | 1.91:1 |
| Pinterest pin | 1000 × 1500 | 2:3 |
| Open Graph / social card | 1200 × 630 | 1.91:1 |
| Email header | 600 × 200 | 3:1 |
Downscaling in one step from 4000px to 400px throws away 99% of the pixels, and a naive nearest-neighbour sample will alias badly — fine textures turn into moiré, and thin lines vanish or double.
This tool sets imageSmoothingQuality: "high", which selects the browser’s best resampling filter, and applies stepped downscaling for reductions greater than 2× — halving repeatedly until within 2× of the target, then doing the final step. That is the same technique high-end image libraries use and it preserves noticeably more detail than a single large jump.
Upscaling is a different matter. Enlarging beyond the original resolution invents pixels that were never captured. The result looks soft, and no amount of sharpening restores information that is not in the file. Where you need genuine upscaling, an AI super-resolution model is the only real answer.
Resize first, then compress. File size scales roughly with pixel count, so halving both dimensions cuts pixels by 75% before any compression is applied.
The most common web performance mistake is serving a 4000×3000 camera image in a slot that displays it at 800×600. The browser downloads all 12 megapixels, decodes them into about 48 MB of memory, and then throws away 96% of them at paint time. Resizing to 1600×1200 — 2× the display size, for high-DPI screens — before compressing typically turns a 4 MB request into 180 KB.
With the lock on, typing a width recalculates the height so the proportions are preserved and nothing is distorted. Turn it off and you can set both independently, which stretches the image.
When you need a specific aspect ratio that does not match the source — a 16:9 thumbnail from a 4:3 photo — stretching is the wrong answer. Use the Image Cropper to cut the frame to the right shape first, then resize. Cropping loses content at the edges; stretching makes every face in the picture look wrong.
Yes, substantially. File size scales with pixel count, so halving width and height typically cuts the file to a quarter or less — usually a bigger saving than compression alone.
Not really. Upscaling interpolates between existing pixels; it cannot recover detail that was never recorded. Modest enlargement up to about 150% is acceptable; beyond that the softness is obvious.
Yes, if you output to PNG or WebP. Exporting a transparent image to JPEG fills the transparent areas with white, because JPEG has no alpha channel.
Yes. Drop a batch, set the target once, and every image is processed with the same settings. Downloads are offered individually or as a ZIP.
Resizing scales the whole image to new dimensions and keeps all the content. Cropping cuts away part of the frame and keeps the remaining pixels at their original scale.
Most often because it was enlarged past its native resolution. Check the original dimensions shown in the panel — if the target is larger, the softness is unavoidable.
Only your device memory. Images up to roughly 60 megapixels resize comfortably on a laptop.