Frame1280×720
Kept—
Weight—
Scale—
Turn any picture into a fixed-size thumbnail for YouTube, Open Graph, LinkedIn, stories, or an avatar. Click the preview to say which part of the photo must survive the crop, and see the same settings applied to every preset at once before you export.
Step by step
01Add the source image
Drop in the frame, screenshot, or photograph the thumbnail is made from. Everything from here is drawn on a canvas in your browser.
02Set the canvas the platform expects
1280x720 is the default because it is what video platforms want. Change it for a card, an open-graph image, or a listing tile — the output is exactly these dimensions, whatever the source was.
03Move the focus point to the subject
Cover fills the canvas and crops the overflow, and the focus point decides which part survives. Drag it onto the face or the product instead of accepting the centre.
04Add padding, rounding, and a background if the frame needs them
Contain plus padding gives a card-style thumbnail that never crops the subject. The radius clips the output itself, so the corners are real rather than a preview effect.
Worked example
Both fit modes on the same source, since the choice between them is the whole job. The numbers are what the canvas does with a 16:9 source going into a 16:9 box that is slightly smaller.
Given
Same ratio, so the fit modes agree - until you add padding
cover scale x0.667 -> 1280x720, nothing cropped
contain scale x0.667 -> 1280x720, nothing padded
contain + 8% pad scale x0.613 -> 1178x662 centred
51px of background on each sideWhen source and canvas share a ratio the fit mode does nothing and the focus point is inert — which is the quiet case. Feed the same canvas a 4:3 screenshot and cover silently removes 25% of the height; that is when the focus point stops being decoration and starts deciding whether the face survives.
Before you publish it
A thumbnail is read at a fraction of its size
The file is 1280px wide; the reader sees it 210px wide in a sidebar or 168px on a phone. Detail that only resolves at full size is not doing any work. Test by shrinking the preview until it is the size it will actually be seen at.
The focus point is the whole feature of cover
Every automatic crop takes the centre, which is where subjects least reliably are. Moving the focus decides which part of an oversized source survives the crop — the difference between a face and a shoulder in a 16:9 slot.
Contain plus padding is the safe frame
When nothing may be cropped — a full product, a chart, a slide — contain fits the whole image and pads the rest. Set the background to something deliberate, because the padding is now part of the composition rather than an accident.
Rounded corners belong to the file, not the page
The radius clips the exported bitmap. That matters when the thumbnail lands somewhere you do not control the CSS — a feed, an embed, an email — and it means JPEG will fill those corners with the background rather than leaving them clear.
Export dimensions are exact
The canvas is the size you typed regardless of what went in, so a set of thumbnails from mixed sources comes out uniform. That consistency is most of what makes a grid of cards look designed rather than assembled.
Nothing is uploaded
Compositing and export both run locally, which matters when the thumbnail is made from an unreleased frame or an internal screenshot. The share link carries the recipe — canvas, fit, focus, padding — never the picture.
The judgement call
The one decision that changes what the reader sees. Everything else is trim.
A photo or video frame, subject near the middle
Cover
Fills the canvas edge to edge. Cropped margins on a photograph are invisible; empty bars are not.
A photo with the subject off to one side
Cover + focus
Keep the full bleed, but move the focus point onto the subject so the crop takes the background instead.
A chart, slide, or full product shot
Contain
Nothing may be lost. Pad to the canvas and choose a background that reads as deliberate.
A tall screenshot into a wide canvas
Contain
Cover would keep a narrow horizontal strip of it. Padding looks intentional; a random slice looks broken.
A set of cards that must look uniform
One canvas, one fit
Consistency across the set matters more than the best framing of any single image.
Text you want legible in a feed
Not from a screenshot
Downscaled UI text turns to mush. Add large type over the image rather than relying on type inside it.
Reference
FAQ
1280 × 720 pixels, a 16:9 frame, is the size YouTube asks for, and it is one of the presets here. Keep the file under 2 MB — exporting as JPEG at 80–90% quality, or as WebP, comfortably clears that for a photographic thumbnail.
1200 × 630 is the widely supported Open Graph size and works across Facebook, Slack, and most link unfurlers. LinkedIn prefers 1200 × 627, which is close enough that the same crop usually serves both. Both are presets, and the preview strip lets you check one framing against the other.
Use Cover and move the focal point. Click the spot that has to stay visible — a face, a product, a logo — and the crop pans towards it. If nothing may be lost at all, switch to Contain: the entire picture is scaled to fit and the leftover space is filled with your background colour.
Cover scales the picture until it fills the frame in both directions, so the excess on the longer axis is cropped. Contain scales until the whole picture fits inside, which leaves bars of background on one axis. Cover is the usual choice for thumbnails and hero images; Contain suits logos, product shots on white, and anything where the full frame carries meaning.
Yes. Choose the 512 square preset, set the corner radius to 50%, and export as PNG or WebP so the area outside the circle is transparent. If the destination flattens transparency, set the background colour to match the surface the avatar will appear on instead.
No. The picture is decoded, framed, and re-encoded by your browser on a local canvas. Nothing is transmitted or stored, and the share link carries only the frame size, focal point, padding, radius, colour, and format.
Thumbnails are rendered in your browser on a local canvas. The source picture is never uploaded, and share links carry only the framing settings.
Keep going
Image cropper
Draw the exact crop by hand instead of fitting to a preset frame.
Image resizer
Change pixel dimensions with the same cover and contain fitting.
Watermark tool
Stamp a logo or a line of text onto the finished thumbnail.
Image compressor
Bring the exported thumbnail under a size budget in kilobytes.