Koldflux Media Lab, browser tool check, October 27, 2026
Tool: https://koldflux.com/tools/video/speed  ("Change Video Speed Online Free: 0.5x to 2x With Sound")
Runner: headless Chromium on the Koldflux VPS, Playwright, viewport 1440 x 1000
Input: the same 3 second excerpt used in test-run.sh, re-encoded to H.264 for the upload
       ffprobe: width=1920 height=1080 r_frame_rate=30/1 duration=3.000000 size=852634
Script: tool-check.cjs (this folder)

What was done:
  1. Opened the live tool page.
  2. Set the file input to the 3 second clip above.
  3. Read the speed controls the page offers.
  4. Clicked "Set speed to 1.5x", the page default, and downloaded the result.
  5. Read the downloaded file back with ffprobe and framemd5.

Raw output:
  PAGE: Change Video Speed Online Free: 0.5x to 2x With Sound · Koldflux
  RADIOS: []
  BUTTONS: ["Set speed to 1.5x","Share","Reject analytics","Accept analytics","Analytics preferences"]
  LABELS: ["Speed\n0.5x (slow motion)\n0.75x (slow motion)\n1.5x\n2x","Replace file\nVideo files · 100 MB per file\nctrl_1x.mp4\nctrl_1x.mp4"]
  CLICKED_SET
  DOWNLOADED: ctrl_1x-out.mp4
  codec_name=h264
  width=1920
  height=1080
  r_frame_rate=30/1
  duration=2.066667
  size=792669
  frames 62 identical 0 0.0
  PAGE_ERRORS: []

What this shows:
  The tool accepted a real 1920x1080 30 fps clip, processed it in the browser with no
  page errors, and produced a downloadable H.264 MP4 that kept 1920x1080 and 30 fps.
  The 3.000 s input came out at 2.066667 s, which is the 1.5x setting, and the 62 output
  frames hold no repeated frame, which is what a speed up should look like.

Limits of this check:
  The 0.5x and 0.75x slow motion settings were not exercised. The page offers them, but the
  script did not manage to select them, so nothing here says what those settings produce.
  The output was not compared frame by frame with our own ffmpeg encode.


Koldflux Media Lab, second browser check: 0.5x slow motion, October 27, 2026 (editor review)
Script: tool-check-05.cjs (this folder). Same input clip as above. Speed chosen first, then the file.

Raw output:
SELECT VALUE 0.5
SELECTS [["0.5=0.5× (slow motion)","0.75=0.75× (slow motion)","1.5=1.5×","2=2×"]]
BUTTONS ["Set speed to 0.5×","Share","Reject analytics","Accept analytics","Analytics preferences"]
BUTTON Set speed to 0.5×
codec_name=h264
width=1920
height=1080
r_frame_rate=30/1
nb_frames=179
duration=5.966667
size=1099174

frames 179 identical-to-previous 0
audio: none
PAGE_ERRORS []

What this shows:
  At 0.5x the 3.000 s clip came out at 5.966667 s with 179 frames at 1920x1080 and 30 fps in H.264,
  which is the 2x longer slow motion the setting promises. The tool has no audio in this output because
  the input excerpt has no audio track. A frame hash cannot count repeats here, because H.264 re-encodes
  every held frame slightly differently; the lossless FFV1 test in test-results.txt is the repeat measurement.