Part of Image tools: See all Image tools.
Image Resize: Resize and crop images to custom dimensions or aspect ratios. Supports JPEG, PNG, and WebP. Ideal for social media, thumbnails, or fitting images to specific sizes. Change width and height in pixels, or crop to 1:1 square, 16:9, 4:3, and more.
Quick steps
- Upload your image via the drop zone or drag and drop.
- Set width and height in pixels, or choose a crop aspect ratio…
- Preview the result and click 'Resize & Download' to save your resized…
Image Resize vs desktop software
| Feature | Image Resize | Desktop software |
|---|---|---|
| Install required | No | Yes |
| Works on phone & desktop | Yes | Varies |
| Free to use | Yes | Often paid |
| Signup needed | No | Sometimes |
People also ask
Is this tool free?
Yes. Unlimited resizing at no cost.
What formats are supported?
JPEG, PNG, and WebP. Output matches input format.
Can I crop to a specific ratio?
Yes. Choose from 1:1, 4:3, 16:9, 3:4, 9:16, or custom dimensions.
What's the maximum image size?
We support images up to 1 GB. Very large images may take longer to process.
Will resizing reduce quality?
Scaling down reduces file size and can sharpen. Scaling up may soften the image. Use appropriate dimensions for your use case.
What is Image Resize?
Resize and crop images to custom dimensions or aspect ratios. Supports JPEG, PNG, and WebP. Ideal for social media, thumbnails, or fitting images to specific sizes. Change width and height in pixels, or crop to 1:1 square, 16:9, 4:3, and more. all processing happens for your request online.
How to use Image Resize
- Upload your image via the drop zone or drag and drop.
- Set width and height in pixels, or choose a crop aspect ratio (1:1, 16:9, 4:3, 3:4, 9:16). Use 0 for width or height to auto-calculate and preserve aspect ratio.
- Preview the result and click 'Resize & Download' to save your resized image.
Why use this tool?
Resize image online free when you need exact dimensions for websites, Instagram, or print. Crop to aspect ratio for consistent thumbnails or social posts. Reducing image dimensions also reduces file size. Our tool run online—no upload to external servers. Need preset sizes for Instagram or TikTok? Try our Social Media Resizer.
FAQ
- Is this tool free?
- Yes. Unlimited resizing at no cost.
- What formats are supported?
- JPEG, PNG, and WebP. Output matches input format.
- Can I crop to a specific ratio?
- Yes. Choose from 1:1, 4:3, 16:9, 3:4, 9:16, or custom dimensions.
- What's the maximum image size?
- We support images up to 1 GB. Very large images may take longer to process.
- Will resizing reduce quality?
- Scaling down reduces file size and can sharpen. Scaling up may soften the image. Use appropriate dimensions for your use case.
Image Resize — In-Depth Guide
Website and blog images load faster when sized to match your layout. A 1920px-wide hero image is overkill for a 800px content column. Resize to your actual display size to cut file size and improve Core Web Vitals. Combine with our Image Compressor for best results.
Social media platforms favor specific sizes. Instagram square posts work best at 1080×1080; stories at 1080×1920. LinkedIn recommends 1200×627 for link shares. Resizing before upload ensures your images display correctly without cropping or compression by the platform.
E-commerce product photos often need consistent dimensions. A grid of 800×800 thumbnails looks clean and professional. Use the 1:1 crop for square product images. For full product shots, 1200×1200 or 1600×1600 is common. Keep aspect ratio consistent across your catalog.
Email signatures and avatars need small dimensions. Profile pictures are typically 200×200 to 400×400. Resizing large photos before use prevents slow-loading emails and bloated attachments. Crop to center the face or logo for best results.
Print projects require exact dimensions. A 4×6 photo at 300 DPI needs 1200×1800 pixels. Business cards, flyers, and banners have specific pixel requirements. Resize to match your print specs to avoid blurry or stretched output.
Resize, resample, rescale — the vocabulary actually matters
Three terms get mixed up in image tools, and the mix-up leads to confused results. Resizing changes the pixel dimensions of an image: a 4000×3000 photo becomes a 1200×900 photo. Resampling is the algorithm used to decide what colour each pixel in the output should be, given that the input pixels do not map cleanly onto the output grid. Rescaling (in CSS and in image viewers) is when the stored pixel dimensions do not change, but the display size does — the browser pretends the image is smaller or larger than it is by interpolating at render time.
This tool does the first two: it changes pixel dimensions and uses a high-quality resampling algorithm (Lanczos-3 by default) to produce output that looks sharp, not blurred or pixelated. It does not do the third: once you download the resized image, the file itself is smaller, and the browser displays it at its new native size.
Platform-specific dimensions that matter in 2026
Every social platform re-encodes anything you upload. If you upload a 4000×4000 photo to Instagram, Instagram will re-encode it to 1080×1080 on the server, apply a heavier JPEG quality setting than you would have chosen, and display that compressed version to everyone. The visual loss is noticeable on detail-heavy images — fine text, intricate patterns, subtle gradients. The fix is to resize and export at the exact target dimensions on your own end, using controllable compression, so the platform does minimal additional re-encoding.
Instagram uses three primary sizes: square 1080×1080, portrait 1080×1350 (the 4:5 ratio that dominates modern feeds), and story 1080×1920. Feed images wider than 1080 are scaled down; anything smaller is scaled up with mediocre results. Facebook cover photos are 820×312 on desktop, 640×360 on mobile; upload at 820×312 and the mobile crop is automatic. LinkedIn banners are 1584×396 on personal profiles and 1128×191 on company pages. Twitter/X header image is 1500×500. YouTube thumbnails are 1280×720; channel art is 2560×1440 with a 1546×423 safe area.
These dimensions change roughly every eighteen months as platforms redesign. The specific numbers above are current as of early 2026, but always confirm against the platform's current documentation for anything high-stakes.
Aspect ratio locks, cropping, and non-square inputs
Most resize operations should preserve aspect ratio — otherwise people and objects stretch into unrecognisable shapes. The aspect-ratio lock in this tool defaults to on. When you change one dimension, the other updates automatically to maintain the original proportions.
What do you do when the target is a fixed aspect ratio (1080×1080 square) and the source is 4000×3000 (4:3)? Two options. First, fit: the image is resized so it fits entirely inside the target, with transparent or coloured padding on the short axis. A 4:3 photo fitted into a 1:1 square gets bars on the top and bottom. Second, fill: the image is resized so it fills the target, with the overflow cropped. A 4:3 photo filled into a 1:1 square loses some of the left and right edges. For portraits, fit preserves the subject; for landscapes posted to square feeds, fill usually produces a more attractive crop. Choose deliberately.
Upscaling: what you can and cannot do
Resizing a 400×400 image up to 1600×1600 multiplies pixel count by 16 but does not add any real information. The new pixels are invented by the resampling algorithm, which interpolates between the existing ones. The result looks blurry compared to a native 1600×1600 source, because the detail is not there to reconstruct. Lanczos-3 looks sharper than bilinear or bicubic interpolation, but none of them can manufacture detail that was never captured.
Neural-network "super-resolution" models (Real-ESRGAN, Topaz Gigapixel) genuinely hallucinate believable detail, and for faces and textures they often look remarkable. This tool does not include a neural upscaler — the compute cost is large enough that offering it free at scale is impractical. For most practical cases, the right answer is: find a higher-resolution source, or accept that the small image will be displayed small.
Downscaling: aliasing and how to avoid it
Downscaling is the common case, and it is lossy in a specific way. When a 4000-pixel-wide image is resized to 1000 pixels, every output pixel is a weighted average of roughly 16 input pixels. If the source contains high-frequency detail — thin lines, small text, fine patterns — that detail can alias (produce visible moiré patterns or jagged edges) unless the resampling algorithm applies a low-pass filter first. Lanczos-3 handles this well; simple nearest-neighbour or bilinear do not, which is why naive downscalers produce visibly worse output than proper ones.
The practical upshot: for downscaling, do not downscale in multiple steps. Go from 4000 to 1000 directly, not 4000 to 2000 to 1000. Each step compounds rounding error. This tool does it in one pass.
Retina displays, pixel density, and when 2× matters
Modern phone and laptop displays have high DPI — roughly twice the old standard. A 100-pixel-wide image on a Retina screen visually occupies the same space as a 100-pixel-wide image on a standard screen, but the Retina screen dedicates 200 actual hardware pixels to render those 100 logical pixels. If the image file is only 100 pixels wide, the browser has to upscale it at render time, and it looks blurry.
For any image that will be displayed on Retina, export at twice the logical size. A blog post header that sits in a 600-pixel-wide column should be exported at 1200 pixels. An avatar displayed at 48 pixels should be exported at 96. This doubles file size, but only roughly — since compression handles repeated detail well, the gain is usually 50–70% rather than 100%. Modern frameworks handle this automatically with srcset; if you are posting manually, just export at 2× and let the browser scale down.
Resizing before uploading vs resizing after
Every major upload pipeline — WordPress, Shopify, Squarespace, Webflow, Ghost — performs its own resize pass on incoming images. They generate responsive variants, apply their own compression, and serve the result through their CDN. This raises a fair question: if the platform is going to resize anyway, why resize before upload?
Two reasons. First, upload bandwidth. Uploading a 12 MB original over a hotel Wi-Fi takes minutes; uploading a 400 KB pre-resized version takes seconds. For someone publishing twenty images, that adds up. Second, control over quality. Platforms tune their encoders for the average case, not your case. A colour gradient that looks clean in your source but develops banding after the platform's re-encode is a common complaint; exporting at the target dimensions with your own compression settings often avoids the re-encode, because the platform sees a file that is already the right shape and leaves it alone.
The one case to skip pre-resize: platforms that strictly require originals for cropping or archival (Flickr, some photo-sharing services). For those, upload the source and let the platform handle responsive variants itself.
Resizing for print: DPI, bleed, and the 300 rule
Screens measure in pixels; print measures in inches or millimetres and a target dots-per-inch. The conventional print target is 300 DPI — a 4×6 inch photo printed at 300 DPI needs 1200×1800 pixels. At 150 DPI the same photo looks soft; at 600 DPI the print cannot meaningfully show more detail than 300 already provides. For small prints (business cards, postcards) 300 DPI is a must. For large-format prints (banners, posters viewed from a distance) 100–150 DPI is plenty because viewers are not inspecting at arm's length.
Print production usually requires 3 mm of bleed — extra image extending past the final trim line, so that small misalignment in the cutter does not leave white edges. When resizing an image for print, add the bleed to the target dimensions: a 100×150 mm postcard at 300 DPI needs 1240×1832 pixels total to cover the 3 mm bleed on each edge, even though only 1181×1772 of those pixels end up visible after trim.
Also try
Related tools that work well with this one: