Compress video

Video compression is easier to judge when the destination is known. An upload form may impose a hard byte limit. A product page may have a playback budget based on mobile loading time. A chat service may accept the original but take too long to send it. Record that constraint before changing quality, dimensions, or frame rate. This tool reports the downloaded result as a separate MP4, so compare its actual byte size with the destination limit rather than relying on the quality number alone.

The quality control is a constant rate factor for H.264 encoding. Lower values retain more detail and usually create larger files. Higher values discard more detail and usually create smaller files. The default is 28, which is a practical compression starting point rather than a promise about size or appearance. Content matters. Fast motion, camera noise, film grain, water, foliage, and confetti need more data than a static slide or talking head at the same setting.

The web page checks the selected file type and the 100 MB input limit before sending one MP4 to the private conversion worker. The worker checks the MP4 signature again, creates a unique temporary job directory, and calls FFmpeg with fixed argument arrays. It does not build a shell command from the filename or form fields. The job directory is removed after the response succeeds, fails, times out, or is cancelled. PWRKIT does not write the video bytes to its application database or analytics events.

FFmpeg decodes the source video and encodes a new H.264 stream. Audio is encoded as AAC unless Remove audio is selected. Source metadata is omitted, and the MP4 is arranged for progressive download so compatible players can begin playback before the full file has arrived. Re-encoding takes time and consumes worker capacity. The progress indicator covers upload and response transfer, but it cannot report every internal encoder frame as a precise percentage.

Run the default first. Download the result and inspect it at the size and playback speed the audience will use. If it is larger than the limit but still looks clean, raise the quality value in small steps and process the original again. If detail is already damaged, lower the value. Always return to the untouched source for each comparison. Compressing a previous result adds another lossy generation and makes it harder to identify which setting caused the visible loss.

Look at difficult frames rather than only the opening image. Pause on movement, faces, text, gradients, smoke, and high-contrast edges. Block patterns, smeared texture, color banding, ringing around text, or unstable fine detail indicate that the encoder is discarding too much information. Listen as well. AAC audio is also re-encoded, so speech, music, and sharp transients should be checked in the downloaded file. A smaller result is useful only if it still meets the purpose of the video.

Reducing frame dimensions often saves more bytes than pushing quality to an extreme. Enter an even output width, height, or both when the destination does not need the original resolution. Supplying only one dimension keeps the source aspect ratio. Supplying both sets an exact output frame and can distort the picture if the ratio changes. Check the downloaded width and height with a media inspector or the publishing system before discarding the source.

Frame rate is another material choice. A 60 frame-per-second source can often be delivered at 30 frames per second when the destination does not need high-motion smoothness. The available values are 15, 24, 25, 30, 50, and 60. Reducing frame rate removes temporal samples and cannot be reversed later. Animation, screen recordings, sports, and camera movement respond differently, so watch the whole result for judder, unreadable pointer motion, or uneven pans.

A crop removes pixels outside the rectangle before rotation and output resizing. Enter all four source-pixel values: X from the left, Y from the top, crop width, and crop height. Width and height must be even. The rectangle must remain inside the uploaded frame. Cropping can remove letterboxing, unused screen areas, or a distracting edge, but it also changes composition. Confirm that captions, faces, logos, and other important content remain visible throughout the clip.

Trim start and trim end use seconds. The end must be later than the start. Cutting an unused introduction or tail reduces duration directly, which is often the cleanest size reduction available. Seek points can interact with source keyframes, and the output duration can differ by a small fraction because video and audio are divided into encoded frames and packets. Check the first and last moments of the download for clipped words, missing transitions, or an unintended blank frame.

Rotation supports 90 degrees clockwise, 90 degrees counterclockwise, and 180 degrees. Horizontal or vertical flipping can be applied separately. These operations change the encoded pixels; they are not saved as a player instruction. A 90-degree rotation swaps the frame axes before an optional output resize is applied. Verify portrait videos in the target player because some source files also contain orientation metadata that other applications interpret differently.

Remove audio is useful when the destination needs silent background footage, when the source track contains private conversation, or when audio bytes are unnecessary. It removes the output audio stream rather than lowering its volume. The visual track remains. Confirm silence with a media inspector, not only the player interface, because some players hide an empty or unsupported audio track. Keep a source copy when the original sound may be needed later.

Constant rate factor controls visual compression, not a fixed final byte count. Two videos with the same duration, dimensions, frame rate, and quality value can produce different sizes because their motion and texture differ. Audio duration and content also contribute. This tool therefore does not label a quality setting as a guaranteed megabyte target. If a service requires a strict limit, process the source, measure the download, adjust one control, and repeat until the actual file clears the limit with acceptable quality.

The result can also be larger than the source. A short MP4 may already use an efficient H.264 configuration, low frame rate, small dimensions, or aggressive compression. Re-encoding it with a higher fidelity setting can add data without restoring details that were previously lost. In that case, keep the original or make a structural reduction such as trimming, cropping, lowering dimensions, removing audio, or reducing frame rate. Do not assume that running a compressor always produces fewer bytes.

Open the downloaded MP4 in a separate player. Confirm duration, width, height, orientation, frame rate, audio presence, and visual quality. Scrub through the full timeline and replay transitions or movement that looked difficult in the source. Then test the actual upload, content system, device, or browser that will receive it. A valid local H.264 MP4 can still be rejected by a destination with its own duration, profile, dimension, or byte limits.

Keep the original until the destination accepts the new file and playback has been checked. The tool never overwrites the source on disk. Use clear filenames when comparing several settings, and record the quality value plus any resize, crop, trim, frame-rate, rotation, flip, or audio changes. That record prevents a smaller result from being mistaken for a full-quality master and makes later compression decisions easier to reproduce.

If the file is rejected immediately, confirm that it is an MP4 no larger than 100 MB and that the bytes match the extension. Renaming another container to .mp4 does not convert it. Export a clean MP4 from a trusted editor if the source is damaged or uses an unsupported structure. If processing fails after upload, reduce the duration or dimensions in desktop software and try again. The worker has bounded execution time and output size so one request cannot occupy it indefinitely.

If the output looks stretched, remove either the custom width or height so the aspect ratio is preserved. If a crop fails, check that all four fields are present and that the rectangle fits within the source frame. If motion looks uneven, keep the source frame rate or choose a value that divides it sensibly. If file size barely changes, remove unused duration or pixels before raising the quality number further. Structural reductions usually preserve more useful detail than forcing the encoder to solve the entire size problem alone.