Social Media Image Sizes in 2026: A Cheat Sheet for Instagram, X, Facebook, LinkedIn, and YouTube

Two numbers decide how your image survives an upload: the aspect ratio (the shape, like 1:1 or 16:9) and the pixel dimensions (the size, like 1080 × 1080). Get the shape wrong and the platform crops for you, usually from the center, which is how heads and logos end up chopped off in feed previews. Get the size wrong in the too-small direction and the platform scales it up, and upscaling always looks soft because there is no detail to invent.

The too-big direction is much safer. If you upload something larger than recommended at the right shape, the platform scales it down, and downscaling is the quality-friendly direction. So the practical rule is: match the aspect ratio exactly, and hit the recommended pixel size or slightly above it. What you should never do is start from an image smaller than the target.

These are the widely agreed recommended sizes as of early 2026. Platforms tweak their specs without much notice, so treat these as recommendations rather than hard limits, but all of them are safe upload sizes today. Every size on this list also exists as a FileTinker preset page (for example, resize-image-for-instagram-post opens the resizer with 1080 × 1080 already filled in, and resize-image-for-youtube-thumbnail with 1280 × 720).

Instagram standardizes feed uploads to about 1080 pixels wide, so 1080 is the width to aim for. The classic square post is 1080 × 1080. The portrait post at 1080 × 1350 (4:5) is the classic recommended portrait size, and every current size guide still lists it as safe. It is worth using: a portrait image takes up noticeably more of the screen than a square as someone scrolls. Instagram has also been shifting toward taller formats across the app — profile-grid previews are now 3:4, and the feed natively accepts 3:4 uploads (1080 × 1440), which are slightly taller than 4:5. If you post one shape in 2026, make it 4:5 or taller.

Stories and Reels are full-screen vertical: 1080 × 1920, a 9:16 ratio. Anything that is not 9:16 gets letterboxed or cropped to fill. Two placement warnings: Instagram's own interface overlays the top and bottom of a Story (your profile ring at the top, reply box at the bottom), so keep text and faces away from the outer edges. And do not reuse a landscape photo as a Story without recropping; squeezing a 4:3 photo into a 9:16 frame distorts it badly.

On X, a 16:9 image at 1600 × 900 displays cleanly in the timeline without surprise cropping. The header (banner) is 1500 × 500, but different screens show slightly different crops of it, and your circular profile photo overlaps the bottom-left corner, so keep anything important centered and away from the edges.

Facebook's most useful size is 1200 × 630, the roughly 1.91:1 shape used for link previews and standard feed images. The cover photo is quirkier: 820 × 312 is the desktop display size, but phones show a taller, narrower crop of the same image, so put the content that matters in the middle and treat the left and right thirds as expendable. Event covers are larger, at 1920 × 1005.

LinkedIn sits close to Facebook: 1200 × 627 for a post or link image, and 1584 × 396 for the personal profile banner, a very wide 4:1 strip. On the banner, remember that your profile photo sits over the lower-left area on most screens, so avoid putting text there.

The YouTube thumbnail spec is simple: 1280 × 720, a 16:9 ratio, at least 640 pixels wide, and under 2MB as a JPG or PNG. The harder part is design. Most people see your thumbnail at around 200 pixels wide in a mobile list of recommendations, roughly one-sixth of the upload size. Zoom your draft out to that size before you commit; if the text is unreadable or the subject vanishes, it will not earn clicks no matter how sharp the 1280 × 720 original is.

Channel art works the opposite way: you upload one big 2560 × 1440 canvas and different devices show different slices of it. TVs display the whole thing, but phones and desktop browsers only show a central strip, so YouTube defines a safe area of 1546 × 423 in the middle. Anything you actually need people to see, like your name, upload schedule, or logo, must live inside that strip. The rest of the canvas is decoration for TV screens.

Getting from a camera photo to a platform-ready image is a two-step job: fix the shape, then fix the size. The order matters. A resizer's job is to hit exact pixel dimensions, so if you feed a 4:3 photo straight into a 1080 × 1920 (9:16) resize, the image has to be squeezed to fill that box, and everyone in the photo gets taller and thinner. FileTinker's resize tool works exactly this way by design: it scales the image to fill the box you ask for, and on preset pages the aspect-ratio lock starts off so the exact target size is honored. That is what you want when the shapes already match, and distortion when they do not.

So crop first. FileTinker's crop tool lets you drag a selection box over the image with the ratio locked to 1:1, 16:9, 9:16, and other presets, so you choose what gets kept instead of letting a platform's center-crop decide. Cropping does not resample anything — the pixels inside the box are copied out at their original resolution — though the file is still re-saved in its source format (PNG stays lossless; a JPG is re-encoded at the same high 0.92 quality), so crop-then-resize on a JPG costs one extra high-quality save. In practice that is invisible, but it is why you should not loop through crop and resize repeatedly. Once the shape matches the target, open the matching resize preset (or type the numbers into the free resizer, which starts with the aspect lock on) and scale to the exact pixels.

Both tools run entirely in your browser. The image is decoded, redrawn on a canvas on your own machine, and offered back as a download; nothing is uploaded to a server. For a photo you have not posted yet, that is worth something.

Downscaling, the normal direction for social media, is the safe one. Going from a 4000 × 3000 camera photo to 1080 × 1350 throws away pixels by definition, but done with high-quality resampling (which FileTinker uses), the result looks crisp, because you are condensing real detail rather than inventing it.

The re-encode is the part worth understanding. After redrawing your image, FileTinker saves it back in the source format where it can: a PNG comes out as a lossless PNG, and a WebP comes out as WebP at a high quality setting of about 0.92. Everything else, including JPGs, comes out as a JPG at that same 0.92 quality, and since JPG has no transparency, any transparent areas are flattened onto white. One high-quality lossy re-encode is essentially invisible, but they stack: resize a JPG, edit it, save it, resize again, and generation loss creeps in. Always do the final resize once, from your best original.

Enlarging is the trap. The resize tool will not stop you from typing 2560 into the width field for a 640-pixel-wide image; it will scale it up, and the result will be soft, because no resizer can create detail that was never captured. If you genuinely have to enlarge, FileTinker's upscale tool does it more cleanly by resampling in doubling steps instead of one big jump, but it is honest interpolation, not AI reconstruction, and it says so. The real fix for a too-small image is a bigger source: the original export, the camera file, or a re-download at full resolution.