Software Signal: target-size compression fixture
Observed 3 October 2026

Purpose
Compare a bitrate estimate with actual HandBrakeCLI outputs under two byte
budgets. This is not a HandBrake versus Shutter Encoder quality or speed test.

Ownership
The source is an owned generated test pattern with a synthetic 440 Hz tone.
It contains no personal data, camera footage, speech or third-party music.
FFmpeg generated the source on 1 October. HandBrake generated the two outputs
on 3 October. FFmpeg 6.1.1 was used for output decoding and frame extraction.
Readers may use these synthetic files to reproduce this compression exercise.

Environment
Windows 11 Home, build 26200; Intel Core Ultra 7 155H.
HandBrakeCLI 1.11.2, official Windows x86_64 release.
Official release URL:
https://github.com/HandBrake/HandBrake/releases/download/1.11.2/HandBrakeCLI-1.11.2-win-x86_64.zip
Official archive SHA-256, matched before extraction:
80bfe8d5f5d11cc3ef76b834add3ed4e82dee6523ffeb435c283f88b1a21f09d
Published checksum source:
https://handbrake.fr/rotation.php?file=HandBrakeCLI-1.11.2-win-x86_64.zip

Input
source-30s.mp4: 13,971,909 bytes
SHA-256: 39d9880d57e6dc4803039b5e56e5832efc4520927a0864b0c1317915bb70c337
30 seconds; 720 x 1280; H.264 yuv420p; 30 fps; 900 video frames;
AAC mono at 48 kHz. No subtitle track was added in this compression test.

Planning method
Decimal MB = 1,000,000 bytes. Decimal kb/s = 1,000 bits per second.
video_kbps = floor(target_MB * 8000 * (1 - reserve_percent/100) / seconds - audio_kbps)
5 MB, 30 seconds, 128 kb/s audio and 5% reserve => 1,138 kb/s video.
2 MB, 30 seconds, 128 kb/s audio and 5% reserve => 378 kb/s video.
The reserve is an editorial allowance, not measured overhead or a guarantee.

Reproduction: both jobs
HandBrakeCLI -i source-30s.mp4 -o OUTPUT.mp4 -f av_mp4 -e x264
  --encoder-preset medium -b VIDEO_KBPS --multi-pass --no-turbo
  -a 1 -E av_aac -B 128 -6 mono -R 48 -r 30 --cfr --crop-mode none
  -w 720 -l 1280 --non-anamorphic --no-markers --no-optimize
Replace VIDEO_KBPS with 1138 or 378 and choose a distinct output filename.
The wrapped command above is one invocation, not a platform-specific script.

Measured results
handbrake-5mb.mp4: 4,800,904 bytes = 4.800904 decimal MB
SHA-256: 0aa8abeacd5331d9fa2e9ecba33b64bf52deb55166465fda6e8dbeaaad4b2e86
handbrake-2mb.mp4: 1,929,999 bytes = 1.929999 decimal MB
SHA-256: 0abb69ed944fbb067827e00ff9eeea839cdf326a28291bd0ef2225c871a3976d

Both encodes exited successfully. Each output retained 720 x 1280 square
pixels, 30 fps and 900 video frames. AAC output was mono at 48 kHz.
The MP4 container duration was approximately 30.02 seconds in each case.
Both passed a complete decode-to-null check with exit code 0 and 900 frames.
The media metadata identified HandBrake 1.11.2 2026060700 as the encoder.

Visual inspection
The source frame and two output frames at 14 seconds were inspected.
The lower-budget output showed more damage around small timer text and
diagonal edges. No full subjective playback study or listening test was run.
No VMAF, speed ranking or claim about all sources follows from this result.

Limits
Only one synthetic input was used. The two exports differ in video bitrate.
No Shutter Encoder export exists in this test. No recipient upload service
was tested. Passing these byte budgets does not guarantee that another
source will fit or that its resulting text and motion will remain useful.