Before you click resize, compare five concrete things: your source file's actual dimensions, codec, duration, and file size against the tool's input limits; the target dimensions you want and whether they should preserve or change your aspect ratio; the output format the tool produces (WebM, not MP4); the privacy model (local processing in your browser tab versus upload to a remote server); and the time cost, since the re-encode runs in real time inside the tab. Each comparison either confirms the tool is the right fit for your clip or redirects you to a different method before you waste effort on a bad output. Skipping any of these comparisons is the most common reason a resize looks wrong on download — the dimensions don't match the platform, the aspect ratio looks stretched, or the file ends up larger than expected. A few minutes spent on these comparisons before you press resize turns the job from guesswork into a predictable outcome, and the result lines up with where the video is actually going to play.

what should i compare before i resize video
What to Compare Before You Resize Video

Source File Specs to Check First

Start by comparing your source clip against the Video Resizer input limits. The tool accepts a local file up to 500 MiB, no longer than five minutes, no more than 4096 pixels on any single side, and no greater than 3840 × 2160 in total pixel area. It also needs the browser to decode the source, which means MP4, WebM, MOV, M4V, or Ogg containers with codec combinations the browser actually supports. Unsupported codecs are not converted silently — they fail visibly. That is the right behavior, because a "successful" download of unreadable frames would be worse than a clear error. The relevant limits in one place:

LimitValue
Source containersMP4, WebM, MOV, M4V, Ogg
Maximum file size500 MiB
Maximum duration5 minutes
Maximum pixels per side4096
Maximum total area3840 × 2160 pixels
Target dimensionsWhole numbers 2 to 1920 on each axis, rounded down to even
ModesFit (aspect-preserving bounding box) or Stretch (exact dimensions)
Output containerWebM (VP9 or VP8)

If your source is a vertical phone recording, a long screen capture, or a 4K master that exceeds 3840 × 2160, you have already found the constraint that disqualifies the tool for this job. Move on without resizing rather than hoping for a graceful fallback.

Target Dimensions and the Fit vs Stretch Decision

Once the source is confirmed compatible, the next comparison is target dimensions versus mode. You enter a maximum width and a maximum height, then pick Fit or Stretch. Fit treats those numbers as a bounding box and applies the smaller of the two scales, so the output stays inside the box, preserves the source's aspect ratio, and never crops. Stretch uses the requested width and height independently, so the output is exactly what you typed, but the picture can be distorted if your source ratio and target ratio do not match. In both modes, dimensions are rounded down to the nearest even pixel because common browser video encoders are more reliable with codec-safe even pixel sizes.

PropertyFit modeStretch mode
Aspect ratioPreservedCan change
Output size relative to boxFits inside the box; may be smaller than the max on one axisExactly the requested width × height
CroppingNoneNone, but content can be distorted
Bounding box behaviorWidth and height treated as maximaWidth and height applied independently
Typical useSocial posts, preserving the source look, letterbox-style framingPixel-exact targets, fixed-size placeholders, intentional stylization
Distortion riskNoneYes, when source ratio differs from target ratio
Odd-dimension handlingRounded down to the nearest even pixelRounded down to the nearest even pixel

The comparison to make here is about intent. If you want the output to fill the requested box at any cost, Stretch is the right choice. If you want the output to look the same as the source, just smaller, or to fit inside a target shape such as 1080 × 1920 without distorting a 16:9 clip, Fit is the right choice. Picking the wrong one is a common cause of awkward, stretched thumbnails and oddly squished reels.

Output Format, Encoding, and Privacy

Two more comparisons matter before you commit: the output format and where the work happens. The result is always a WebM file, created by a real-time pass through the browser's canvas captureStream and MediaRecorder APIs using a supported VP9 or VP8 WebM configuration. WebM is chosen because it uses the codecs already shipped with the browser, so no large media library is bundled and no server-side encoder is needed. If you need MP4 output, frame-accurate delivery, or exact bitrate control, a desktop encoder is the correct tool, not this one.

The privacy comparison is just as concrete. Decode, canvas resizing, recording, and download generation all happen in your current browser tab, so the source file is never uploaded to a server. The finite duration of the output WebM is written by a shared audited Matroska duration writer. For a deeper look at what does and does not leave your device, see the explanation of local resizing.

One more comparison belongs in this section: what the re-encode can and cannot preserve. Quality, bitrate, frame timing, audio layout, and color metadata can all differ from the source because the encoder is making a new file from a canvas of decoded frames, not editing the original stream. File size can differ from the source and depends on the codec the browser picks and the complexity of the scene.

How to Resize a Video Locally Step by Step

After the comparisons, the actual job is short. Open the Video Resizer and follow these steps:

  1. Choose one supported local video (MP4, WebM, MOV, M4V, or Ogg, up to 500 MiB and five minutes) and wait for the browser to read its metadata.
  2. Enter the maximum width and the maximum height you want, then choose Fit to preserve the source aspect ratio inside a bounding box, or Stretch to force the exact width and height you entered.
  3. Select Resize video, keep the browser tab open during the real-time re-encode, then download the resulting WebM when the recorder finishes.

Keep the tab open until the download link appears, and treat the resulting file as a real re-encode rather than a metadata tweak.

Worked Example: 4K Downscaled With Fit Mode

Comparing the math explicitly makes the decision easier. Take a 3840 × 2160 source (a 4K clip) and ask for a 1920 × 1080 target box in Fit mode.

  • Width scale: 1920 ÷ 3840 = 0.5
  • Height scale: 1080 ÷ 2160 = 0.5
  • Smaller of the two scales: 0.5
  • Width output: 3840 × 0.5 = 1920
  • Height output: 2160 × 0.5 = 1080
  • Both values are already even, so the codec-safe rounding step is a no-op.
  • Final output: 1920 × 1080 WebM, aspect ratio unchanged at 16:9.

Change the box to 1280 × 1920 and the smaller scale becomes 1280 ÷ 3840 = 0.333, giving output dimensions of 1280 × 720 inside the taller box. The bounding box is taller than the source ratio requires, so the picture sits within the box rather than being stretched to fill it. The decision between the two outputs is purely about whether you want full HD landscape or a 720p clip framed inside a vertical canvas.

When a Browser Resizer Is Not the Right Tool

The last comparison is honesty about what the Video Resizer cannot do. It will not upscale detail — pixel information that is not in the source stays absent in the output. It will not remove black bars, crop a specific subject, or preserve subtitles or multiple audio tracks beyond what the browser exposes. It is also a real-time re-encode running at up to 30 frames per second, so processing normally takes about as long as the video itself. For frame-accurate delivery, footage longer than five minutes, required MP4 output, color-managed workflows, or exact bitrate control, the right next step is a dedicated desktop encoder. You should also resize only media you own or are licensed to edit; nothing in this tool bypasses DRM or promises professional mastering. For more on the kinds of traps that show up at this stage, the Video Resizer tips and common mistakes guide covers them in detail.