Free Video Resizer

What resizing actually does to your video

Video resizing changes the pixel dimensions of every frame. Downscaling (going from 1080p to 720p, for example) discards pixels: a 1920x1080 frame has roughly 2 million pixels, a 1280x720 frame has roughly 920 thousand. The resampling algorithm decides which pixels of the original contribute to each pixel of the output and how much. Upscaling (going from 720p to 1080p) does the opposite: it invents new pixels by interpolating between existing ones. Upscaling does not add detail; it makes the existing detail visible at a larger size. AI upscaling (Topaz Video AI, ESRGAN-based tools) is a different technology that hallucinates plausible detail using a neural network. This tool does classical resampling, not AI.

The resampling algorithm matters. Nearest-neighbor picks the closest source pixel; fast but blocky. Bilinear averages a 2x2 neighborhood; slightly blurry. Bicubic uses a 4x4 weighted average with cubic interpolation; the workhorse for most downscales. Lanczos (typically Lanczos-3) uses a 6x6 sinc-based kernel and is the sharpest commonly available; it's FFmpeg's default for high-quality scaling. Area / box averaging is best for heavy downscales like 4K to 1080p because it preserves more detail than bicubic at large ratios. This tool uses Lanczos by default.

Aspect ratio is independent of resolution. 1920x1080, 1280x720, and 640x360 are all 16:9. 1080x1080 is 1:1 (square). 1080x1920 is 9:16 (vertical/portrait). Changing aspect ratio (16:9 to 9:16, for instance) means deciding what to do with the content that doesn't fit: letterbox (black bars), pillarbox (black bars on the sides), crop (lose part of the frame), or stretch (distort, always wrong). The aspect ratio lock in this tool prevents accidental stretching when you change one dimension.

How this tool works under the hood

Same engine as the video compressor and converter: FFmpeg compiled to WebAssembly via Emscripten. The browser-side binary is about 30 MB, runs entirely in the tab via SharedArrayBuffer multi-threading. When you drop a video, the file goes into the WebAssembly virtual filesystem via a streaming reader; even 1 GB files don't blow up browser memory at upload time.

The resize command FFmpeg runs looks like -i input.mp4 -vf scale=1280:720:flags=lanczos -c:v libx264 -preset medium -crf 23 -c:a copy output.mp4. The -vf scale filter does the resampling per-frame. The :flags=lanczos picks the Lanczos algorithm. Audio is copied through unchanged (resizing doesn't affect audio). For aspect-ratio-preserving resizes that round to even dimensions, the tool uses scale=1280:-2 (height auto-calculated, rounded to even because H.264 requires even dimensions).

When you change aspect ratio with the lock disabled, the tool optionally letterboxes by combining scale and pad filters: scale=W:H:force_original_aspect_ratio=decrease,pad=W:H:(ow-iw)/2:(oh-ih)/2. The result fits the target box with black bars where source content doesn't fill it. Progress messages stream from FFmpeg's stderr and update the on-screen progress bar in real time.

Brief history of video resolution and aspect ratio

• Standard-definition TV, 1950s to 2000s. Analog TV was 4:3 aspect ratio. NTSC (North America/Japan) ran at 720x480 effective resolution at 30 fps; PAL (Europe/much of Asia) ran at 720x576 at 25 fps. Films originally shot for 16:9 or 21:9 cinema were pan-and-scanned to fit 4:3 TVs, often cropping out half the original frame.
• HD broadcasting and widescreen TVs, 2000s. The 16:9 widescreen aspect ratio standardized: 1280x720 (720p) for early HD broadcasts, 1920x1080 (1080p) for full HD. The transition took roughly 2003 to 2012, with letterboxing the bridge format on legacy 4:3 displays.
• iPhone launches, 2007. Apple ships the first iPhone with a 480x320 portrait-oriented display. Vertical video as a concept begins; the idea that screens could be taller than wide reshapes consumer video over the next decade.
• Instagram launches with square video, 2010. Instagram debuts with a 640x640 square format that becomes a brand signature. The 1:1 aspect ratio remains uniquely tied to the platform throughout its first decade.
• Vertical video goes mainstream, 2013 to 2016. Snapchat (2013) popularizes 9:16 vertical video for short-form sharing. Musical.ly/TikTok (2016) makes vertical the default for short-form entertainment. By 2020 every major platform supports 9:16 as a first-class format.
• 4K mainstream and 8K niche, 2018 to 2026. 4K (3840x2160) becomes the typical output of modern phones and prosumer cameras. 8K (7680x4320) is largely a production format for downsampling rather than a delivery format. Browser-side resizing reaches practicality with ffmpeg.wasm 4.x around 2024.

How It Works

1. Upload your video: Select or drag a video file you want to resize.
2. Set target dimensions: Enter the width and height in pixels, or choose a preset (1080p, 720p, 480p, 360p, 240p). Enable "Maintain aspect ratio" to prevent distortion.
3. Download the resized video: Click Resize and download the output video at your chosen dimensions.

Why Use Video Resizer?

Platform requirements, bandwidth constraints, and embedding targets all demand specific video dimensions. Instagram requires square or portrait video, YouTube wants 16:9, Twitter has upload size limits, and embedded players need specific pixel dimensions to fill their containers. Resizing video in your browser avoids installing video editors or paying for cloud processing services, and keeps your footage private.

Features

• Custom dimensions: Enter any width and height in pixels.
• Aspect ratio lock: Maintain the original aspect ratio automatically when setting one dimension.
• Common presets: Quick presets for 4K, 1080p, 720p, 480p, and social media formats.
• No distortion: Optional letterboxing to avoid stretching the video.
• Browser-based processing: Your video files stay on your device throughout the entire process.

Real-world resize workflows

• Adapting horizontal video for TikTok/Reels. A 1920x1080 (16:9) clip uploaded to TikTok or Instagram Reels gets letterboxed inside their 9:16 vertical frame, looking small and unprofessional. Resizing to 1080x1920 with crop preserves the focus area at full screen. The tool can scale to vertical with optional letterboxing depending on whether you want bars or to lose the edges.
• Reducing 4K phone footage for email/chat. Modern phones default to 4K (3840x2160) recording, producing huge files. For a quick share via email, chat, or Slack, downscaling to 720p (1280x720) cuts file size to roughly a fifth without visible quality loss on phone-sized playback.
• Web embedding at consistent dimensions. A self-hosted hero video or product demo on a website usually targets a fixed container width. Resizing to match that container exactly (often 1280x720 or 960x540) keeps the file size predictable and avoids the browser doing extra scaling at render time.
• Standardizing archive resolution. A folder of family videos shot over a decade has resolutions ranging from 480p (old phones, screen recordings) to 4K (modern phones). Resizing everything to a single resolution (often 1080p) makes the archive predictable: every clip plays the same way in any viewer and storage cost is bounded.
• Generating preview-resolution variants. For an art portfolio or video library, you might want each clip available at both full resolution (for projection or large screens) and at thumbnail/preview resolution (360p or 480p) for quick browsing. Resize the master once and store both versions; loading the smaller variant is much faster on mobile.
• Fitting old footage into modern player aspect ratios. A 4:3 home video from 2005 plays in a 16:9 modern player with pillarboxing (black bars on the sides). Resizing to 1280x720 with letterbox/pillarbox lets you bake the bars into the file itself so any player presents it consistently, rather than relying on each player to do the right thing.

Common pitfalls and what they mean

• Stretching from aspect ratio mismatch. If you turn off the aspect ratio lock and enter dimensions that don't match the source, the output is stretched: people look thin or fat, circles become ovals. Always either keep the lock on, or pick dimensions that match the source aspect ratio, or accept letterbox/crop instead of stretch.
• Odd-number dimensions break H.264. H.264 (the universal video codec) requires even width and height. Entering 1281x721 produces an encoding error. The tool rounds odd values to the nearest even number to avoid this; if you see your dimensions silently adjusted by one pixel, that's why.
• Upscaling does not add detail. Taking a 720p source and resizing to 1080p does not give you a 1080p video; it gives you a 720p video stored in a 1080p container. AI upscaling (Topaz Video AI, ESRGAN-based tools) is a different technology that hallucinates plausible detail using a neural network. This tool does classical resampling, which can never invent detail the source didn't have.
• Chroma subsampling artifacts at extreme downscales. H.264 typically stores color at half the resolution of luminance (4:2:0 chroma subsampling). At extreme downscales (4K to 360p, for instance), text and sharp color edges can develop colored fringing because the chroma plane was already lower resolution before scaling. This is a codec limitation, not a tool bug.
• Variable framerate confuses scalers. Screen recordings and some phone cameras output variable framerate. Some scalers handle this badly, producing duplicate or skipped frames. The tool normalizes to constant framerate during resize to avoid this; output is constant fps even if the source was variable.
• Display aspect ratio overrides get lost. Some containers store a display aspect ratio override separate from pixel dimensions (anamorphic widescreen, for instance). Browser-side resize bakes the pixel dimensions into the output; if your source relied on a DAR flag to look correct, you may want to handle it explicitly in a desktop tool.

Privacy: your video never leaves your device

Cloud video-resize services (Clideo, OnlineConvertFree, FreeConvert, Kapwing for batch, dozens more) all upload your full video, run a scaler on their hardware, and send back the resized result. For a 200 MB phone video that's 200 MB up plus 50 to 200 MB down through their infrastructure. Video content commonly includes faces, locations, audio of conversations, GPS-tagged scenes, screen recordings of private interfaces. Most operators publish privacy policies committing to delete uploads within 1 to 24 hours and encrypt in transit, and major ones hold ISO/IEC 27001 certifications. They have strong business reasons to honor those policies. But "deleted within an hour" is not "never seen." During that window the file sits on operator infrastructure, accessible to any process or person with the right permissions, visible in logs and backups per the operator's retention policy.

This tool never uploads anything. The full pipeline (file selection, decoding via browser-native readers, scaling via ffmpeg.wasm WebAssembly, download via the browser's blob API) runs inside your browser tab. No uploads, no network requests carrying video data, no log entries. You can verify by opening browser dev tools on the Network tab before resizing: no requests leave with video content. Only the initial page load and the one-time roughly 30 MB ffmpeg.wasm download (cached for subsequent visits) touch the network. Put the browser in airplane mode after page load and the resizer still works on local files.

When another tool is the right pick

• Files over 2 GB. Browser memory limits become a wall above roughly 2 GB. Use desktop HandBrake or FFmpeg CLI, which can stream from disk and use all available system RAM.
• AI upscaling for resolution recovery. If you genuinely need to make a 480p source look usable at 1080p or 4K, you need a neural-network upscaler like Topaz Video AI, Real-ESRGAN, or DLSS-based tools. These hallucinate plausible detail from learned patterns. Classical resampling (what this tool does) cannot create detail that wasn't in the source.
• Frame-accurate cropping and pan-and-scan paths. For pan-and-scan workflows where the crop area moves over time, or for animated reframes, use desktop editors like DaVinci Resolve, Premiere, or Final Cut Pro. They expose keyframe-based crop parameters that browser tools don't.
• Batch resize of hundreds of files. A shell script with FFmpeg CLI (for f in *.mp4; do ffmpeg -i "$f" -vf scale=1280:720 "out_$f"; done) is dramatically faster than running a browser tool 100 times. Can also use GPU scalers (NVIDIA NPP, Intel QuickSync) for additional speedup.

Frequently Asked Questions

What is the maximum input video resolution?

The tool can handle videos up to the memory and compute limits of your browser. 4K (3840×2160) videos work on modern devices, though processing time increases significantly with resolution.

Will resizing reduce quality?

Downscaling (reducing resolution) involves some quality loss inherent in the compression. The tool uses high-quality resampling. Upscaling beyond the original resolution will not add detail, it just makes pixels bigger.

How do I resize video for Instagram?

For Instagram Feed posts: 1080×1080 (square) or 1080×1350 (portrait). For Stories and Reels: 1080×1920. Select the preset or enter these dimensions manually with aspect ratio lock enabled.

Other frequently asked questions

What's the difference between resizing and compressing?

Resizing changes the pixel dimensions of every frame (1920x1080 to 1280x720, for example). Compression changes the bitrate at which those pixels are stored (higher CRF, smaller file). Both reduce file size but in different ways: resizing trades resolution for size, compression trades quality at the same resolution for size. For maximum file size reduction, combine both: resize to a smaller resolution and apply compression. For just smaller dimensions without quality compromise, resize alone.

Should I use letterbox/pillarbox or crop when changing aspect ratio?

Letterbox/pillarbox (black bars) preserves all the original content at the cost of unused screen space and a slightly amateur look on social platforms. Crop loses content at the edges but fills the frame and looks professional, ideal when the important action is centered. For TikTok/Reels uploads of horizontal footage, crop usually wins if the focus is centered; letterbox wins if cropping would lose key content.

Why does my resized video have a slightly different file size than expected?

Resizing a 1080p file to 720p doesn't always cut the file size by 4x even though the pixel count drops by 4x. The video re-encoding step uses CRF (Constant Rate Factor), which targets visual quality rather than file size. The smaller resolution needs fewer bits to encode at the same quality, but content complexity matters more than the pixel reduction ratio. Typical file size reduction for 1080p to 720p ranges from 40% to 60%.

Why is resizing slower than I expected?

Resizing video is not just a scale operation; the encoder also has to re-encode every frame at the new dimensions. Encoding is the slow part, not scaling. A 5-minute 1080p clip resizing to 720p might take 3 to 10 minutes depending on your CPU. The -preset flag controls the speed-vs-efficiency tradeoff; the tool uses medium by default. Hardware-accelerated scaling/encoding is 10 to 20x faster on desktop tools but unavailable in the browser.

Is there a desktop or command-line equivalent?

Yes. FFmpeg CLI: ffmpeg -i input.mp4 -vf scale=1280:720:flags=lanczos -c:v libx264 -crf 23 -c:a copy output.mp4. HandBrake (GUI, free, cross-platform). Shutter Encoder (friendlier FFmpeg GUI). DaVinci Resolve and Premiere offer it inside their export dialogs. All produce essentially identical output to this browser tool because they share the same underlying scaler.

Does resizing affect audio?

No. Audio is independent of video resolution. The tool copies the audio stream unchanged from input to output, so audio quality and file size for the audio track are identical before and after resize. Only the video stream changes dimensions.