reelkit

Finish short-form video for TikTok, Instagram Reels and YouTube Shorts in one command.

Raw clips from phones, editors and AI video tools come out in the wrong shape, at random loudness, and full of metadata. reelkit turns them into post-ready 1080×1920 MP4s:

Step	What it does
vertical	Scales and crops any clip to 1080×1920 (9:16), with an optional punch-in zoom
loud	Two-pass EBU R128 loudness normalisation to -14 LUFS / -1.5 dBTP (the level the platforms play back at)
strip	Removes all container and stream metadata and chapters, without re-encoding
faststart	Moves the moov atom to the front so the video starts playing instantly
finish	Runs vertical, then loud, then strip in one go
thumb	Saves a full-quality cover frame as a JPG at any timestamp
info	Checks a clip against the spec and says whether it's ready to post

Example

A landscape export straight out of an editor, checked before and after:

$ reelkit info IMG_4471.mp4
  [!!] resolution 1920x1080 (want 1080x1920)
  [ok] duration 8.0s (want <= 180s)
  [!!] metadata tags present: encoder, title
  [!!] loudness -21.85 LUFS (want -14.0)
run `reelkit finish` to fix

$ reelkit finish IMG_4471.mp4 --zoom 1.05
ready to post -> IMG_4471.final.mp4

$ reelkit info IMG_4471.final.mp4
  [ok] resolution 1080x1920 (want 1080x1920)
  [ok] duration 8.1s (want <= 180s)
  [ok] metadata clean
  [ok] loudness -13.99 LUFS (want -14.0)
ready to post

Why I built this

I post a lot of short-form product videos, and every clip went through the same manual routine: reframe to 9:16, fix the loudness so it didn't sound quiet next to everything else in the feed, strip the editor and camera metadata, and make sure it starts playing instantly. I kept a text file of ffmpeg commands and pasted them in one by one. reelkit is that text file turned into one command.

Install

You need Python 3.9+ and ffmpeg on your PATH.

# Windows
winget install Gyan.FFmpeg
# macOS
brew install ffmpeg
# Linux
sudo apt install ffmpeg

Then install reelkit directly from GitHub:

pip install git+https://github.com/reelsmith/reelkit.git

Or clone it and install locally:

git clone https://github.com/reelsmith/reelkit.git
cd reelkit
pip install .

Usage

reelkit finish raw.mp4                 # -> raw.final.mp4, ready to upload
reelkit finish raw.mp4 --zoom 1.05     # add a subtle 5% punch-in
reelkit vertical landscape.mov -o out.mp4
reelkit loud voiceover_mix.mp4
reelkit strip export.mp4
reelkit faststart export.mp4
reelkit finish raw_clips/             # every video in the folder
reelkit thumb raw.final.mp4 --at 1.2   # cover image from 1.2s in
reelkit info final.mp4                 # pass/fail checklist, exit code 0 = ready
reelkit --dry-run finish raw.mp4       # print the ffmpeg commands without running them

If you don't pass -o, the output is written next to the input with a suffix, for example clip.clean.mp4 or clip.9x16.mp4.

finish accepts a folder too. It processes every .mp4, .mov, .m4v, .mkv and .webm inside it and skips files it has already finished.

How it works

raw clip ──► vertical ──► loud (measure) ──► loud (correct) ──► strip ──► clip.final.mp4
             1080x1920    LUFS + true peak   linear gain, -14    no metadata, faststart

• Reframe: scales to cover 1080×1920, then center-crops, so there are never black bars. --zoom 1.05 scales 5% larger before cropping for a subtle punch-in.
• Encode: H.264 -preset slow -crf 18, yuv420p for maximum compatibility, AAC 192 kbps at 48 kHz.
• Strip: a stream copy with -map_metadata -1 -map_chapters -1 and bit-exact flags, so it doesn't lose any quality.
• Temporary files are cleaned up even if a step fails.

You can also run it without installing: python -m reelkit finish raw.mp4.

Why -14 LUFS?

TikTok, Instagram and YouTube all turn loud uploads down and leave quiet ones quiet. If you master to about -14 LUFS integrated with a -1.5 dBTP ceiling, your audio plays back as you mixed it, without getting squashed or sounding thin next to other videos in the feed. reelkit runs ffmpeg's loudnorm twice. The first pass measures the audio and the second applies a linear correction, so the dynamics aren't pumped.

FAQ

Does strip re-encode my video? No. It only rewrites the container, so it's instant and doesn't lose any quality.

Will this change how my video looks? vertical and finish re-encode at CRF 18, which is visually lossless for social media. strip, faststart and thumb never touch the pixels.

Why exit code 2 from info? So you can use it in scripts: 0 means ready, 2 means it needs fixing, 1 means an error.

Roadmap

• Burned-in captions from an .srt file, in TikTok-style fonts
• Safe-zone aware caption placement (see safezone)
• Per-platform presets (--for tiktok, --for shorts)
• Install from PyPI with pip install reelkit

Development

pip install -e ".[dev]"
pytest

The tests check the ffmpeg commands reelkit builds, so they don't need any video files.

License

MIT