Upload a landscape photo to an Instagram post, a tall phone photo as a LinkedIn link image, or a screenshot to X, and the platform quietly decides what part of it people see. It fits the image to one of a few fixed shapes and trims whatever does not fit, and it trims from the middle outward. If the subject is in the middle, that works. If the subject is a face near the top of the frame, a product on the left, or a headline along the bottom, the platform crops it away.
The fix is to crop the image yourself, to the exact shape the feed wants, and to choose which part survives. This guide covers the sizes and aspect ratios the main platforms display, the geometry of a crop, how to choose a focal point, what cropping does to quality and file size, and how to avoid the soft, upscaled result that catches people out. The worked examples use the free Social Media Image Cropper on Toolz, which does the whole job client-side.
TL;DR: To crop an image for social media, pick the target aspect ratio for the platform and placement (for example 1080 x 1350 for an Instagram portrait post, 1200 x 630 for a link preview, 1600 x 900 for an X post image), take the largest region of that shape that fits inside your photo, anchor it on the part that matters, then scale it to the exact pixel size. A 4032 x 3024 phone photo cropped to 4:5 keeps 60% of its area; cropped to 16:9 it keeps 75%. The social media image cropper does this on a canvas in your browser with a nine-point focal grid, free, no signup, with no upload.
What does a social media image cropper do?
A social media image cropper cuts a photo to the aspect ratio a platform displays and scales it to that platform's pixel size, so the platform has nothing left to crop. Two separate operations happen, and it helps to keep them apart:
- Crop. Choose a rectangle inside the original that has the target shape. Pixels outside it are discarded. Pixels inside it are untouched.
- Scale. Resize that rectangle to the target dimensions, such as 1080 x 1350. This is where quality can change, because the image is resampled.
A generic image editor can do both, but it makes you find the ratio, draw the box by hand, and remember the output size. A dedicated cropper knows the sizes, draws the largest valid box automatically, and lets you move it with one click. The Toolz version also tells you what the crop cost before you download: how much of the original area survives and whether the result was scaled down or up.
Which image sizes do the main platforms use?
Each network renders feed images at a small set of aspect ratios, and anything else is fitted to the nearest one. The table lists the 15 target sizes built into the cropper, which reflect what each platform currently displays without re-cropping. Platforms revise these from time to time, so treat them as current guidance rather than a permanent specification.
| Platform | Placement | Pixel size | Aspect ratio | Where it is used |
|---|---|---|---|---|
| Square post | 1080 x 1080 | 1:1 | The classic feed post | |
| Portrait post | 1080 x 1350 | 4:5 | Tallest the feed allows, most screen space | |
| Landscape post | 1080 x 566 | 1.91:1 | Widest the feed accepts | |
| Story or Reel | 1080 x 1920 | 9:16 | Full screen; keep text clear of the top and bottom 250 px | |
| Threads | Square post | 1080 x 1080 | 1:1 | Shown uncropped in the timeline |
| Threads | Portrait post | 1080 x 1350 | 4:5 | Tallest Threads displays without cropping |
| Bluesky | Feed image | 1200 x 630 | 1.90:1 | Wide images are pulled toward 1.91:1 |
| Bluesky | Square image | 1080 x 1080 | 1:1 | Shown uncropped |
| X | Post image | 1600 x 900 | 16:9 | Shown in full without a tap |
| X | Summary card | 1200 x 628 | 1.91:1 | Large link preview card |
| Feed post | 1200 x 630 | 1.90:1 | Also the Open Graph size | |
| Feed post | 1200 x 627 | 1.91:1 | LinkedIn feed standard | |
| YouTube | Thumbnail | 1280 x 720 | 16:9 | Minimum 1280 wide for the HD player |
| Standard pin | 1000 x 1500 | 2:3 | Taller pins are downranked | |
| Open Graph | Link preview | 1200 x 630 | 1.90:1 | Default card most sites and chat apps render |
Two things stand out. The "1.91:1" family (1200 x 630, 1200 x 628, 1200 x 627, 1080 x 566) is one shape with slightly different rounding, and it is the shape of nearly every link preview on the web. It traces back to the Open Graph protocol, which defines the og:image tag sites use to supply a preview picture. And the tall formats differ: Instagram and Threads stop at 4:5, Pinterest goes to 2:3, and Stories use 9:16. One crop does not serve all of them.
If you are also setting the preview image for a web page, the Open Graph preview tool shows how that 1200 x 630 image will look once shared.
How is the crop rectangle calculated?
The crop is the largest rectangle of the target shape that fits inside the original image. If the image is wider than the target shape, the crop keeps the full height and trims the sides. If it is taller, the crop keeps the full width and trims the top and bottom.
In numbers, for a source of width W and height H and a target ratio R (width divided by height):
if W / H > R: crop width = H x R, crop height = H (trim the sides)
else: crop width = W, crop height = W / R (trim top and bottom)
Worked example 1: a 4032 x 3024 phone photo to an Instagram portrait post.
- Source ratio: 4032 / 3024 = 1.333. Target ratio: 1080 / 1350 = 0.8.
- The source is wider, so the crop keeps the full 3024 px height.
- Crop width: 3024 x 0.8 = 2419.2 px.
- Area kept: 2419.2 / 4032 = 0.60, so 60% of the photo survives and 40% is trimmed from the sides.
- The crop is then scaled from about 2419 x 3024 down to 1080 x 1350.
Worked example 2: the same photo to an X post image.
- Target ratio: 1600 / 900 = 1.778. The source ratio 1.333 is smaller, so the crop keeps the full 4032 px width.
- Crop height: 4032 / 1.778 = 2268 px.
- Area kept: 2268 / 3024 = 0.75, so 75% survives, with 756 px trimmed from the height.
Worked example 3: the same photo to a Pinterest pin.
- Target ratio: 1000 / 1500 = 0.667. Crop width: 3024 x 0.667 = 2016 px.
- Area kept: 2016 / 4032 = 50%. Half the photo goes.
These percentages are what the tool reports as "Keeping 60% of the original", and they are worth reading before you commit. A crop that keeps half the frame is a composition decision, not a formatting step.
How do you choose what part of the image to keep?
Choose the focal point that holds the subject, and remember that it only matters along the axis being trimmed. The cropper's nine-point grid sets a focus of 0, 0.5 or 1 across each axis, which anchors the crop to the left edge, the center or the right edge horizontally, and to the top, the middle or the bottom vertically.
The position is worked out like this, then clamped so the crop never runs off the image:
crop x = focusX x W - cropWidth / 2 (clamped between 0 and W - cropWidth)
crop y = focusY x H - cropHeight / 2 (clamped between 0 and H - cropHeight)
In worked example 1, only the sides are trimmed, so only the horizontal choice changes anything:
| Anchor column | Crop x | Kept columns of the photo |
|---|---|---|
| Left | 0 | 0 to 2419 |
| Center | 806.4 | 806 to 3226 |
| Right | 1612.8 | 1613 to 4032 |
The top, middle and bottom rows give identical results for that crop, because the full height is already kept. It is a useful check when a click seems to do nothing: the crop is only free to move in one direction.
Worked example 4: a portrait phone photo as a LinkedIn image. A 3024 x 4032 photo cropped to 1200 x 627 (ratio 1.914) keeps the full width and a strip 3024 / 1.914 = 1580 px tall. That is 1580 / 4032 = 39% of the photo. With the default center anchor the strip runs from y = 1226 to y = 2806, which on a typical head-and-shoulders portrait is the chest, not the face. Clicking top-center moves the strip to y = 0 to 1580 and keeps the head. This case is the reason the focal grid exists: centering is a reasonable default and the wrong answer for most portraits turned into banners.
A few rules of thumb for picking an anchor:
- People: top row, since faces usually sit in the upper third of a frame.
- Headlines or captions baked into a graphic: the row or column where the text sits.
- Products and off-center subjects: the matching corner or side.
- Symmetrical scenes and centered logos: leave it at center.
Will cropping reduce image quality?
Cropping alone does not reduce quality, but the scaling step can. Discarding pixels outside the frame leaves the remaining pixels exactly as they were. The resize to the target size is where the result can improve or degrade.
Scaling down is safe and often makes a photo look crisper, since several source pixels are averaged into each output pixel. The cropper draws with the canvas's high smoothing quality setting, using the drawImage method defined in the WHATWG HTML canvas specification. Scaling up cannot add detail that was never captured. The output gets bigger but not sharper, and edges look soft.
Worked example 5: an upscaled screenshot. An 800 x 600 screenshot cropped for an X post image keeps the full 800 px width and a 450 px tall strip, 75% of the area. That 800 x 450 region is then enlarged to 1600 x 900, twice its size in each direction. The tool says so under the preview: "Upscaled - the source is smaller than the target, so it will look soft."
The quick test is to compare the crop width with the target width. If the crop is wider than the target, the image is downscaled and will look fine. If it is narrower, it is being enlarged. For photos, a phone camera almost always has pixels to spare. Screenshots, old images and pictures saved from other posts are the usual upscaling cases, and the only real fix is a larger source.
Should you export PNG, JPEG or WebP?
Export JPEG for photographs, PNG for screenshots, text and graphics with hard edges, and WebP when you want the smallest file and know the destination accepts it. The cropper defaults to PNG because it is lossless, and offers all three.
| Format | Compression | Best for | Watch out for |
|---|---|---|---|
| PNG | Lossless, as specified by the W3C PNG standard | Screenshots, UI, text, logos, transparency | Large files for photographs |
| JPEG | Lossy, the method in ITU-T T.81 | Photographs | Halos around sharp text and lines; no transparency |
| WebP | Lossy or lossless, described in Google's WebP documentation | Smallest files at similar quality | A few older tools cannot open it |
JPEG and WebP exports use a quality setting of 0.92 on the canvas encoder, which is high enough that compression artifacts are hard to see in photographs. The downloaded file is named after the original plus the preset, so team.jpg cropped for an Instagram portrait post as JPEG saves as team-ig-portrait.jpeg.
Most platforms re-encode what you upload anyway, so a lossless PNG of a photograph does not survive as PNG once posted. It mainly adds upload size. For choosing formats in more depth, the image format converter guide compares them across more use cases.
Step by step: cropping an image with the tool
The image cropper for social media has four steps, and the preview canvas is the same canvas that produces the file, so what you approve is what you download.
- Add the image. Drop a file, click to choose one, or paste an image from the clipboard with Cmd+V or Ctrl+V. The paste listener is on the whole page, so a screenshot copied to the clipboard goes straight in without being saved first. The drop zone accepts PNG, JPEG, WebP, GIF or AVIF, or anything else your browser can decode.
- Pick the target size. The Target size menu groups the 15 presets by platform, each labeled with its pixel size and ratio, and shows a one-line note on where the size is used. The default is the Instagram portrait post.
- Choose what to keep. Click one of the nine points in the Keep grid. The preview redraws immediately.
- Check the numbers and download. Under the preview, read the line that states the kept percentage, the crop dimensions and whether the result is downscaled or upscaled. Pick PNG, JPEG or WebP and click Download.
Everything runs on an HTML canvas in the browser tab. The image is not uploaded, stored or logged, and the page keeps working with the network off. Like every tool on the site, the download step checks a daily usage limit.
What errors can you run into?
Most problems come from the file format or from the size of the source. The messages are specific.
- "That file is not an image." The file's type does not start with
image/, for example a PDF or a video. The message names the file. - "That image could not be read. It may be corrupt or in a format this browser does not decode." The file claims to be an image but the browser cannot decode it. The common case is an iPhone HEIC photo in a browser without HEIC support. Convert it to JPG first with the image format converter.
- "Upscaled - the source is smaller than the target, so it will look soft." Not an error, a warning. Use a larger original if you have one.
- The focal grid seems to do nothing. The crop is only trimmed along one axis, so only the anchors along that axis move it.
- Daily limit reached. The free daily usage allowance has been used.
One more effect is worth knowing. Because the output is re-encoded from canvas pixels, the downloaded file carries no EXIF metadata: no camera settings, capture date or GPS location. For photos posted publicly that is usually a privacy gain. The EXIF remover guide explains what that metadata contains and why it matters when sharing photos.
Crop, resize, or fit on a background?
Cropping is one of three ways to get an image into a fixed shape, and the right one depends on whether you can afford to lose any of the picture.
| Approach | What happens | When to use it | Tool |
|---|---|---|---|
| Crop to the ratio | Trims the edges, keeps the shape exact | The subject is a part of the frame and the edges can go | Social Media Image Cropper |
| Fit on a matte | Shrinks the whole image onto a colored or blurred background | Nothing may be cut, such as artwork or a group photo | Matte Generator |
| Resize only | Changes pixel dimensions, keeps the original ratio | The ratio is already right, or the platform accepts any shape | Image Resize |
| Split into slides | Cuts a wide image into consecutive carousel slides | A panorama too wide for one post | Carousel Scroll Generator |
If the problem is file size rather than shape, the image resize guide covers changing dimensions without distortion. When a cropped image is going out publicly and should carry your name, the watermarker adds a logo or signature in a corner or tiled across the frame. The full set of image tools for posting lives in the social media tools hub.
Why crop before uploading at all?
Because the platform crop is decided by software that cannot see your subject. Every feed renders at fixed ratios, and an upload in any other shape is fitted automatically, usually by trimming evenly from both sides of the long dimension. Some platforms, Bluesky among them, apply that crop only on display in the timeline, so the full image exists but most people never see it.
Cropping first moves three decisions back to you: which part of the frame is visible, how sharp the result is at the exact display size, and how large the file is. It takes a few seconds per image, and it matters most for exactly the images people care about: portraits, product shots, and graphics with text near an edge. For those, crop the image to the platform size with the focal point on the subject, free, no signup, and upload a file the feed has no reason to change.
