ThumbCraftedTry Free Editor
Free · Private · No Upload

Video to JPG, and the sharpest frame found for you

Step frame by frame, or let it scan the whole video and rank the clean ones.

Most frame grabbers make you scrub around hunting for a shot that is not a blurred mid-motion mess. This one measures every sampled frame for focus and contrast and puts the best candidates in front of you. Your video never leaves the browser.

Drop or click to upload

Drag & drop, or pick from your device

MP4, WebM, MOV, MKV — 500MB max

Nothing uploaded No file size queue Full quality frames
0:00.00 / 0:00

Keyboard: step one frame · Space play or pause · G grab this frame

Samples across the whole video and ranks each frame by focus and contrast. It cannot judge composition, only clarity.

Every few seconds

Export

Frames are letterboxed to fit the chosen size without stretching.

Grabbed frames

No frames yet. Grab one, or scan for the sharpest.

Why hunting for a good frame by hand rarely works

Video is full of frames you would never choose. Anything captured mid-motion is smeared, because a camera exposes each frame over a slice of time rather than freezing an instant. Scrub to a random point in a talking-head video and there is a decent chance you land on a blink, a half-formed word, or a hand crossing the face.

Scrubbing around looking for a clean one is slow, and worse, it is unreliable: a frame that looks acceptable in a large player can fall apart at the size a thumbnail is actually seen. So this tool does the searching by measurement instead.

What "sharpest" actually measures

Pressing Find the sharpest frames samples the video at regular intervals and scores each sample on two things:

The two are combined and the top candidates are surfaced with their scores shown. It cannot judge composition, expression, or whether the shot suits your topic. No measurement can. It narrows a ten-minute video down to a handful of technically clean frames so you are choosing between good options rather than scrubbing blind.

The frame you grab is not the frame you publish

A still pulled from video is a starting point, not a finished thumbnail, for reasons worth knowing:

IssueWhy it happensWhat to do
Softer than a photoVideo compression allocates far fewer bits per frame than a still camera does, and inter-frame compression means most frames are stored as differences rather than complete images.Grab from the highest-resolution copy you have. Sharpening recovers some apparent detail.
Wrong aspectVertical or ultrawide footage does not fit a 16:9 thumbnail frame.Export at 1280 × 720 here, then crop to place the subject deliberately.
Nothing to readA raw frame carries no title, no focal point, no text.Send it to the editor and add what the frame is missing.
Fails at feed sizeA frame that reads fine at full size can turn to mush at 168px.Run the finished thumbnail through Thumbnail Autopsy, which measures exactly that.

Frame stepping, and why it is approximate

The arrow buttons and the ← → keys move roughly one frame at a time. "Roughly" is honest: a browser exposes a video's playback position in seconds, not in frame numbers, and it does not report the true frame rate. The step is derived from the video's measured cadence where the browser supports frame callbacks, and falls back to 1/30 of a second where it does not. For choosing a still it is more than close enough; for genuine frame-exact editing you want a proper NLE.

Nothing is uploaded, and with video that matters more

The video is read from your device into a blob URL and decoded by the browser's own player. Frames are drawn to a canvas and encoded locally. No part of the file is ever transmitted, which is why there is no upload progress bar, no size queue, and no wait behind other people's jobs. A 500MB video costs you nothing but local decode time, where an upload-based tool would spend several minutes moving it to a server before doing any work.

Frame Extraction Questions

Is my video uploaded anywhere?
No. It is read straight from your device into the browser's own video player, and frames are drawn to a canvas locally. Nothing is transmitted, which is why there is no upload bar and no queue. This matters more with video than with images simply because the files are so much larger.
How does it decide which frames are sharpest?
It samples across the video and scores each frame on focus, using the variance of a Laplacian edge filter, and on contrast, using the spread of luminance. Sharp frames have abrupt brightness changes at edges; motion-blurred ones do not. It cannot judge composition or expression, only clarity, and the page says so rather than pretending otherwise.
Why is my extracted frame blurrier than I expected?
Two reasons. Video compression gives each frame far fewer bits than a photo would get, and anything captured mid-motion is genuinely smeared because each frame is exposed over a slice of time rather than an instant. Use the sharpest-frame scan to avoid the worst of it, and sharpen afterwards if needed.
What video formats work?
Whatever your browser can play, which in practice means MP4 (H.264), WebM and usually MOV. MKV and some codecs depend on the browser. If a video loads but shows a black frame, the container is readable but the codec is not, and converting it first is the fix.
Can I get every frame of the video?
Not in one action, and deliberately. A one-minute 30fps video is 1,800 frames, which would exhaust browser memory and produce a download nobody wants. Use "grab one frame every N seconds" for an even spread, or the sharpest-frame scan to get the handful worth keeping.
Why is the frame step not exactly one frame?
Browsers expose playback position in seconds and do not report a video's true frame rate. The step is measured from the video's own cadence where the browser supports frame callbacks and falls back to 1/30 second otherwise. Close enough for picking a still; not a substitute for a real editor if you need frame-exact work.
What do the safe zones show?
The dashed rectangle is the area that survives cropping across platforms, and the solid orange box in the bottom-right is where YouTube stamps the video duration. If the moment you want has your subject in that corner, pick a different frame or plan to move them in the editor.
Does the ZIP compress my images?
No, it stores them. JPG, PNG and WebP are already compressed, so running them through ZIP compression again would cost time and save almost nothing. The archive is a container for convenience, and every frame inside is byte-identical to downloading it on its own.

Continue crafting

Keep going without re-uploading — your image carries over to the next tool automatically.

  1. 1AutopsyGrade it before you publish
  2. 2EditAdd text, effects, or a fresh background