The selected time is a media timeline position, not a guarantee that the extracted frame came from exactly that instant. When you ask a browser to seek to a specific time and then draw the current decoded frame, the browser uses HTMLMediaElement seeking behavior, which snaps to the nearest keyframe and rounds to the closest decoded timestamp. The result is a still drawn from a nearby decoded frame, not necessarily the source frame at the exact moment you typed. For most everyday uses — thumbnails, reference images, slides, social posts — the visual difference is unnoticeable. For frame-accurate editorial work, color-managed output, HDR pipelines, or any task that requires hitting an exact numbered source frame, the answer is no: the displayed time records your requested position, and the decoded frame you receive is the browser's best approximation at that moment.
What the selected time actually represents
The time field in Video Frame Extractor accepts any value from zero through the video's reported duration, including fractional seconds such as 12.75. The number you type is stored as the requested media timeline position. Internally, the tool assigns that value to the browser's HTMLMediaElement currentTime property, waits for the seek to finish, then draws the current decoded frame onto a same-size canvas and encodes it as a PNG. The output filename includes the requested time so you can see exactly what position you asked for.
What you do not get back is a guarantee that the drawn frame came from the exact source frame at that timestamp. The browser decoder is free to land on a nearby decoded frame because video is encoded as a sequence of keyframes (full images) and predicted frames (delta updates). Random access to any timestamp usually means decoding from the most recent keyframe forward. If your requested time falls between keyframes, the browser typically seeks to the nearest keyframe and either decodes forward or returns the cached frame at that point.
Why browser seeking can land on a nearby frame
Several technical factors influence how close the extracted frame is to the time you typed. None of them are bugs — they are the normal behavior of browser video decoding, and the same factors affect command-line tools like FFmpeg when seeking by time rather than by frame number.
| Factor | What it does | Effect on the extracted frame |
|---|---|---|
| Keyframe interval | Video is encoded with I-frames (keyframes) every N frames; predicted frames depend on the previous keyframe. | Browser seeks to the nearest keyframe; longer keyframe intervals mean larger potential gaps. |
| Variable frame rate (VFR) | Some clips encode frames at irregular intervals, such as screen recordings. | Browser timestamps may not align evenly with real playback time. |
| Edit lists and CTTS offsets | Container metadata can shift where t = 0 starts or remap frame timestamps. | The seeked position may not align with the visual moment you expect. |
| Codec timestamp rounding | Decoders round seek times to the nearest sample timestamp they can decode. | Small fractional values may still snap to the closest available sample. |
| HTMLMediaElement currentTime | Per the MDN HTMLMediaElement currentTime reference, the property sets a target time and the browser picks the nearest decodable frame. | You control the requested target, not the exact sample landed on. |
Consider a 30 fps clip with a keyframe every 60 frames. Keyframes fall every 2 seconds. If you request t = 1.3 s, the browser decoder may snap to the keyframe at t = 0 s (about 39 frames back from your intent at 30 fps) or to the keyframe at t = 2 s (about 21 frames forward). That is why fractional seconds alone cannot guarantee frame accuracy in browser-based extraction.
How to extract a frame at a specific time using Video Frame Extractor
- Open Video Frame Extractor and choose one supported local video file (MP4, WebM, MOV, M4V, or Ogg). The file is processed in your browser — nothing is uploaded.
- Wait for the video's duration and preview to load. The displayed duration is the upper bound for the time field.
- Enter the frame time in seconds. Use decimals such as 12.75 for sub-second positions. The time must be between 0 and the reported duration.
- Select Extract PNG frame. The browser seeks to that time, waits for the seek to finish, draws the decoded frame at its native dimensions, and encodes it as a PNG.
- Verify the still. The PNG uses the decoded frame dimensions, preserves transparency only if the decoded video frame provides it, and does not resize, crop, sharpen, interpolate, or apply filters.
- Download the local PNG. The filename includes the requested time so you can trace the file back to your input.
The tool applies shared safety limits: files up to 500 MiB and five minutes long, no more than 4096 pixels per side, and no greater than 3840 × 2160 pixels in area. Files outside those bounds cannot be processed. For files where the codec inside the container is not supported by your browser, the tool reports a decode, seek, canvas, or encoding failure rather than producing an empty download.
How to get closer to the frame you want
Even when the selected time is not frame-accurate, you can usually get within a frame or two of the moment you want with a few simple habits.
- Enter fractional seconds. Stepping by 0.01 s or 0.05 s is often enough to land on the next decoded frame when keyframes are dense.
- Bracket the moment. Extract one frame slightly earlier and one slightly later, then compare. The visual you want is usually one of the two.
- Use shorter source clips. Files within the 5-minute and 500 MiB limits decode more predictably and let the browser cache frames more reliably.
- Re-run the extraction. Because the tool records your requested time in the filename, you can compare two PNGs side by side to see how the browser interpreted each request.
- Verify the still before downloading. The PNG mirrors the decoded frame dimensions — if the resolution looks wrong, confirm that the source file decoded correctly in your browser first.
For help choosing the moment to extract, the guide Extract Frames from Video: How to Pick the Right Time walks through picking timestamps that are likely to be near a keyframe.
Limits of browser-based extraction and when to switch tools
Video Frame Extractor is designed for thumbnails, reference stills, slides, presentations, and quick PNGs from media you own or may process. It is not a substitute for editorial tools when precision matters. The tool does not inspect packet timestamps, does not identify an exact numbered source frame, and does not provide HDR or color-managed output. It also does not support batch extraction, alpha workflows on most decoded content, or video longer than five minutes.
For frame-accurate editorial work — pulling the exact 1247th frame of a master file, matching frames across an original and a re-encode, color-managed stills for grading review, or HDR pipeline frames — use a dedicated desktop video tool. Command-line tools such as FFmpeg with frame-number seeking, or NLE software that decodes with full packet-level timestamp access, can hit an exact source frame. The browser-based approach is the right tool when speed, locality, and simplicity matter more than hitting a numbered frame.
If your extraction result looks wrong in a way the tips above do not fix, the tool reports decode, seek, canvas, or encoding failures instead of producing an empty PNG. Those messages tell you whether the problem is the file, the codec, or the input time — useful information when deciding whether to retry locally or move to a desktop workflow.
For a deeper look, see Is the Video Uploaded When You Extract a Video Frame?.