Tour a 3-section sports montage and find each section start
Proof frames
Tool calls
overview_chunks
Arguments
{
"source": "$CLIP",
"chunk_sec": 60,
"samples_per_chunk": 8
}
Result
{
"ok": true,
"data": {
"chunks": [
{
"file_path": "/private<cache>/133c3feaf47520809b97/contact_sheets/sheet_8samples_4cols_720d6f.jpg",
"sample_count": 8,
"rows": 2,
"cols": 4,
"start_sec": 1.36,
"end_sec": 20.400000000000002,
"timestamps": [
1.36,
4.08,
6.800000000000001,
9.520000000000001,
12.24,
14.96,
17.68,
20.400000000000002
],
"frame_width": 768,
"frame_height": 432,
"index": 1,
"range_start_sec": 0,
"range_end_sec": 21.76,
"tool_file": {
"_tag": "ToolFile",
"name": "sheet_8samples_4cols_720d6f.jpg",
"mimeType": "image/jpeg",
"encoding": "base64",
"data": "<701276 base64 chars dropped>",
"byteLength": 525956
}
}
],
"total": 1,
"start_sec": 0,
"end_sec": 21.76,
"chunk_sec": 60
}
}
detect_cuts
Arguments
{
"source": "$CLIP"
}
Result
{
"ok": true,
"data": {
"cuts": [
{
"timestamp_sec": 7.12,
"score": 0.638708,
"kind": "cut"
},
{
"timestamp_sec": 15.68,
"score": 0.872432,
"kind": "cut"
}
],
"total": 2
}
}
Question and answer
- Question
- Give me a quick tour of this video. Where do the sections start?
- Answer
- detect_cuts lists 2 changes, at 7.12, 15.68 seconds.
How it was checked
- Truth from
- constructed
- Fixed at
Planted by this recipe before any tool ran: ffmpeg joined 3 sections with hard cuts and no sound, in this order: beach-volleyball-slow-motion, ulaanbaatar-marathon-2026, kettlebell-juggling-beach. The sections start at 26.28 s, 21.96 s, 24.84 s in their sources and are 7.08, 8.52, 6.16 s long (no scene change above 0.1 inside any section; sources skipped for having no such section: princeton-yale-football-1903). The truth is the planted join times 7.08 s, 15.6 s, the start of each section after the first, with tolerance 0.5 s. The calls are overview_chunks over the whole montage as one sheet, then detect_cuts with its default threshold, both fixed before the run.
A script joined 3 sections, each from a different clip of the same kind, with hard cuts and no sound, so the answer is the planted joins and not a real film's structure. The first section's start (0 s) is not counted.
Try it yourself
Using agentic-video, give me a quick tour of my video. Run overview_chunks on the whole video to see its sections, then run detect_cuts on the whole video and tell me the time of each section start, in seconds. Use my own file.
- Get the .mcpb file and open it in Claude Desktop.
- Or:
pip install agentic-video
Clip and licence
- Licence
- CC-BY-4.0
- Author
- Alexis Jazz
- Source
- commons.wikimedia.org
This file is licensed under the Creative Commons Attribution 4.0 International license.
Also uses (insert 1)
- Licence
- CC-BY-SA-4.0
- Author
- Orgio89
- Source
- commons.wikimedia.org
This file is licensed under the Creative Commons Attribution-Share Alike 4.0 International license.
Also uses (insert 2)
- Licence
- CC-BY-SA-4.0
- Author
- Taco fleur
- Source
- commons.wikimedia.org
This file is licensed under the Creative Commons Attribution-Share Alike 4.0 International license.