Local Video Compressor & Trimmer

Define the job before changing the file

A useful compression decision begins with a viewing task. A customer checking whether a hinge closes needs the movement and the final fit. A colleague reviewing a dashboard needs readable labels and the result of the click. Write that requirement in a sentence before choosing settings. It gives you a concrete reason to keep detail or duration instead of treating the smallest download as the automatic winner.

Select one local source rather than a remote address. The source can be MP4, MOV, or WebM, but its container name does not establish whether it can be processed. The supported video codecs are AVC, VP8, and VP9, subject to native browser decoding. A MOV containing another codec can therefore be unsuitable even when it plays in a different application on the same computer.

The source budget is 150 MiB and five minutes, with one video track and at most one audio track. Choose a working interval between 0.25 and 120 seconds within that recording. These are this tool’s operating limits, not destination requirements. For a longer recording, make an appropriate working source elsewhere while retaining the original, rather than assuming this page can assemble multiple excerpts or process an entire long session.

The platform selector records the preparation context. It does not change the converter into a native export preset, adjust a destination’s account settings, or certify acceptance. An interview and a product demonstration can need different editorial choices while using identical encoding mathematics. Keep the destination review separate from the local file review, and do not interpret a successful MP4 conversion as evidence that every later use will accept it.

Read bitrate and duration as one size budget

Bitrate describes a rate of data over time, not a quality score. The video target here is an integer from 500 to 8000 kbps. One kbps means 1000 bits per second, and 1000 kbps equals 1 Mbps. At a fixed target, a longer interval needs more estimated bytes. At a fixed duration, lowering the target reduces the estimate but gives the encoder less room to describe movement and detail.

The planning formula is (video kbps + kept audio kbps) * 1000 * seconds / 8. Division by eight converts bits into bytes. Kept source audio contributes a 128 kbps AAC target; a source without audio contributes zero. The formula estimates the selected interval, not the full source, and cannot predict exactly how the native encoder will distribute data or how many bytes the MP4 container will add.

Use the correct unit when discussing the result. One MB is 1,000,000 bytes, while one MiB is 1,048,576 bytes. A 10,000,000-byte file is 10 MB but approximately 9.537 MiB. The bytes did not change when the label changed. If a reviewer gives a size budget, clarify its unit and leave room below it; a rounded display is not sufficient evidence for a strict byte limit.

Try a planning comparison at 1800 kbps video with audio kept. Twenty seconds gives 1928 * 1000 * 20 / 8, or 4,820,000 estimated bytes: 4.82 MB, approximately 4.597 MiB. Forty seconds at those settings doubles the estimate to 9.64 MB. That relationship helps you choose a test, but it does not establish which version tells the story adequately or what either finished file will actually weigh.

Work through the four-second demonstration

The demonstration is a synthetic four-second clip, not somebody’s private recording. Its source is 960 by 540 pixels at 24 FPS with stereo AAC audio. The example selects start 0 and end 4, a maximum edge of 1280, a 2500 kbps video target, and keepAudio enabled. Those settings select the whole demonstration, allowing you to examine the controls without confusing a shortened story with a compression-only comparison.

A maximum edge is a ceiling, not a request to enlarge the picture. The 960 by 540 source already fits below 1280, so the planned dimensions remain 960 by 540. Its estimated source rate of 24 FPS also remains below the output ceiling of 30 FPS. Neither setting invents extra captured detail. The kept stereo track is reencoded as stereo AAC at a target of 128 kbps and 48 kHz.

Add the targets before multiplying: 2500 + 128 equals 2628 kbps. For four seconds, 2628 * 1000 * 4 / 8 equals 1,314,000 bytes. That is 1.314 MB, approximately 1.253 MiB. It is the target-based estimate, not a promised download size or a measured result from your device. The actual native encoding must finish and the resulting MP4 must be checked before its byte count can be reported.

For a controlled second attempt, keep the same source, interval, and dimensions while changing video to 1500 kbps. The estimate becomes 814,000 bytes, or 0.814 MB. Do not assume the lower number makes that version better. Compare moving edges and fine detail at the same viewing size, then listen to both. Record actual bytes only after each conversion, and keep the chosen settings alongside the exported file for later review.

Trade picture detail against the bitrate target

The maximum-edge choices are 640, 1280, and 1920 pixels. Downscaling reduces the amount of picture information the encoder must describe, which can make a modest bitrate more usable. It also removes spatial detail. A sign that remains readable at 1280 may become ambiguous at 640. Judge the detail that carries the message, such as a measurement, a product seam, or a small control label, rather than the general attractiveness of the image.

Consider a 1920 by 1080 workshop clip. Choosing 1280 plans a 1280 by 720 output; choosing 640 plans 640 by 360. Neither choice intentionally crops the frame. The tool preserves the overall proportions while respecting its pixel budget and even-dimension requirements. Unusual source dimensions can produce slight rounding differences, so inspect the output dimensions and outer edges rather than promising mathematically identical borders for every recording.

Movement makes comparisons more revealing. A stationary title can look acceptable while a quick pan breaks edges into blocks or smears small textures. Test a section containing the hardest important action, not only a calm opening. At the same duration and target bitrate, a lower resolution has the same planning estimate; any practical benefit is a different allocation of limited data, not an additional guaranteed saving built into the formula.

Watch the output at its likely viewing size, then inspect the critical region more closely. Avoid comparing a tiny preview of one attempt with an enlarged preview of another. If text disappears, first reconsider whether the chosen dimensions support the task; raising bitrate cannot recreate pixels removed by downscaling. If framing itself needs to change, use a separate editing workflow. This tool does not crop, create portrait reframing, or rebuild the composition.

Trim a complete thought instead of chasing a number

Start and end describe positions in the source. A start of 12.5 and an end of 30.5 select eighteen seconds, not thirty and a half seconds. Enter seconds with up to three decimal places, or use an accepted minute-and-second time form. Recheck the subtraction after changing either boundary. A mistyped position can retain the wrong action while still producing a valid interval and a plausible size estimate.

Suppose a demonstration says, This only works after the adhesive cures, before showing a successful repair. Starting after that condition creates a different claim even if the repair is clearly visible. Retain the condition or choose another complete explanation. At the end, allow enough time to see the result rather than stopping at the last hand movement. A concise clip should remove waiting, not remove the evidence that makes its conclusion fair.

Requested times are not a frame-exact editing guarantee. Video consists of frames with timestamps, and sound has its own packet and sample boundaries. Reencoding can produce a duration that differs slightly from the requested span. Three decimal places let you describe a boundary precisely; they do not create captured frames between existing ones. Watch and listen to the actual first and last moments instead of approving an endpoint solely because its typed number looks exact.

If an eighty-second source becomes a twenty-second output, the file comparison includes sixty seconds of removed content as well as encoding changes. Describe it as output versus the full source file, not a pure compression ratio. For a compression experiment, keep the interval constant across attempts. For a pacing experiment, hold encoding settings steady while checking which complete explanation works best. Separating those tests makes your conclusion much easier to defend.

Make an explicit decision about the soundtrack

Keep audio when it contributes meaning: an instruction, an interview answer, a warning, or the sound demonstrating a fault. The output target is AAC at 128 kbps and 48 kHz, retaining the source’s one or two channels. Keeping audio means reencoding it, not copying the original audio bit for bit. Listen to the result because a listed track confirms its presence but does not tell you whether the important words remain intelligible.

Mute is an explicit choice for a genuinely silent visual explanation or an independently prepared use that does not need the source sound. At the same video target, removing 128 kbps of audio from a thirty-second clip reduces the planning estimate by 480,000 bytes, or 0.48 MB. That may be useful, but it is a poor trade when the omitted sentence explains what the viewer is seeing or limits an otherwise misleading claim.

A quiet preview is not the same as a muted export. Player volume and the keep-audio setting answer different questions. Before converting, verify the requested track state; afterward, check the reported output state and listen at an appropriate volume. A source with no audio cannot gain a soundtrack by enabling keep audio. An unsupported source audio decoder may also prevent a kept-audio conversion rather than silently yielding a successful but incomplete result.

Subtitles and other additional tracks are not preserved as a subtitle workflow here. If an explanation depends on captions, retain the caption source separately and review it in the subtitle editor as a separate task. Check timing against the exported clip, especially after removing an introduction. This page does not generate a transcript or automatically move text between tools, and muting an instructional recording does not make its information accessible by itself.

Understand capability checks and unfinished conversions

Local processing uses the browser’s native decoding and encoding capabilities. The same file may be supported on one browser and refused on another device because the required path differs. Both the source decoder and the H.264 output encoder must be available, along with the necessary audio support when keeping sound. A familiar extension or successful playback elsewhere is not a promise that this conversion can complete in your current environment.

Known HDR and known wide-gamut BT.2020 or P3 sources are rejected. This is not an HDR tone-mapping or archival color-conversion tool. Unknown color metadata remains unknown rather than proving that the recording is standard dynamic range. For color-critical work, use an appropriate managed workflow and retain the source. A small file with acceptable geometry can still be unsuitable when faithful color interpretation is essential to the decision it supports.

The finished output is limited to 64 MiB. Planning refuses an estimate above 85 percent of that cap to leave headroom, but the margin is not a promise about actual bytes. A busy scene or native encoding behavior can differ from the target estimate. Reduce the selected duration or target if planning exceeds the budget, then reassess whether the remaining clip still serves its purpose. Never treat a partial buffer as a finished video.

Wait for completion and validation, not just a progress indicator reaching its end. The conversion checks requested tracks and then inspects the resulting MP4 for its dimensions, duration, and audio state before making a download available. If processing fails, times out, or is cancelled, no partial output should be treated as usable. Retrying a supported, shorter selection can help isolate a problem, but cannot establish universal browser or device compatibility.

Approve measured results and preserve the right files

Review actual bytes, duration, dimensions, and requested audio state together. The reported actual video-and-audio bitrate is derived from complete output bytes divided by actual duration, so it includes container overhead. It is not a separate measurement of video alone and need not equal the requested video target. A larger output is possible, especially when the source was already efficient. Lossy reencoding should therefore earn its place through a useful, inspected result.

Play the exported clip from beginning to end, seek into its busiest passage, and revisit both boundaries with sound. Check the first visible action, the final held result, any small text, and any spoken qualification. A successful technical check cannot assess those editorial details. If overlays will be added later, the video safe zone checker can support a separate placement review; it does not certify this output’s destination acceptance.

Download the actual MP4 separately from its review. Copy and CSV provide review information; the raster PDF is a review document, not a playable video or a collection of frames. It excludes private filenames. Settings JSON retains editable choices and notes, not the source file, its private name, or the output bytes. Keep the media yourself, because reloading settings cannot restore a recording that was never included and there is no autosave.

For a handoff, record the purpose, chosen interval, settings, actual byte count, and remaining checks in concise notes. Keep the untouched source and avoid repeatedly recompressing an already compressed attempt when you can return to it. Media preparation remains separate from recurring account work. A five-day SMTasker trial can help evaluate that separate routine on connected Android phones from Windows or macOS; it does not import this video or complete the clip review for you.

Questions about local video compression