Rhythmic Cutting (3-2-1) — effect skill for cuts & pacing
For AI agents: find the right skill in one call —
curl -s "https://varg.ai/how-to/api/search?q=<what the person wants>"returns the matching how-to page and the best skills with install commands, as markdown (Russian works too). Full index: https://varg.ai/how-to/llms.txt. Every page has a markdown twin: add.mdto its path. Before generating anything, show the person the options on the how-to page.
Shorten shot lengths step by step (3 s, 2 s, 1 s, half a second) to build momentum into a payoff.
- Type: Effect · Cuts & pacing
- Slug:
rhythmic-cutting-3-2-1 - Status: live
- Publisher: HeyGen (community)
- License: Apache-2.0
- Source: https://github.com/heygen-com/hyperframes/tree/main/registry/components/beat-timeline
- Works in: Claude Code, Claude.ai, Codex, Cursor, OpenClaw, Gemini CLI, Any MCP client, varg Studio
- Requires: ffmpeg, HyperFrames (npx hyperframes), Remotion, CapCut (by hand)
- Cost: Free — the take renders locally; generated footage, if any, bills by the model
What it does
Built on Beat Timeline by HeyGen (HyperFrames), Apache-2.0. The write-up and recipes below are varg's.
Each shot is shorter than the last until the edit lands on a longer hero shot or a drop.
When: Before a reveal, a drop in the music, a punchline or the product hero shot.
Watch out: Accelerating with no payoff shot at the end; the rhythm needs a landing.
Timing: e.g. 90, 60, 30, 15, 15 frames at 30 fps, then a 2–3 s hold
Also called: accelerating cuts, shrinking shot lengths, 3-2-1 rhythm, build-up cutting
In FFmpeg
Generate trims from a list of durations and concat (bash loop builds the filter):
d=(3 2 1 0.5 0.5); f=""; i=0
for x in "${d[@]}"; do f+="[$i:v]trim=0:$x,setpts=PTS-STARTPTS,fps=30[v$i];"; i=$((i+1)); done
for j in $(seq 0 $((i-1))); do f+="[v$j]"; done
ffmpeg -i s1.mp4 -i s2.mp4 -i s3.mp4 -i s4.mp4 -i s5.mp4 -filter_complex "${f}concat=n=$i:v=1:a=0[v]" -map "[v]" out.mp4
In HyperFrames
Set data-duration per clip to 3, 2, 1, 0.5, 0.5 with data-start chained to the previous id.
In Remotion
const lengths = [90, 60, 30, 15, 15, 75];
<Series>{shots.map((s, i) => <Series.Sequence key={i} durationInFrames={lengths[i]}><Video src={s} /></Series.Sequence>)}</Series>
In CapCut
Trim clips to shrinking lengths on the main track; use Beats markers for the half-bar and bar positions.
Ready-made components
- Beat Timeline (hyperframes): npx hyperframes add beat-timeline
Where this comes from
Use when
- Before a reveal, a drop in the music, a punchline or the product hero shot.
Not for
- Accelerating with no payoff shot at the end; the rhythm needs a landing.
Install
- Claude Code:
claude plugin marketplace add vargHQ/skills && claude plugin install varg@varg - Codex:
npx skills add vargHQ/skills --skill varg-video-edit-library -a codex -y - Cursor:
npx skills add vargHQ/skills --skill varg-video-edit-library -a cursor -y
Example output
https://s3.varg.ai/how-to/previews/0a22ee481dbf/take-rhythmic-cutting-3-2-1.mp4 — crosscut between two shots, with the timeline strip of the edit. Footage: WING IT! (2023), Blender Studio | studio.blender.org, CC BY 4.0; Coffee Run (2020), Blender Studio | studio.blender.org, CC BY 4.0. (credit: Rendered by varg · WING IT! (2023) + Coffee Run (2020))
More cuts & pacing skills
- SmartCut (capcut-ai-editor): MCP that reads CapCut auto-captions to cut silences and duplicate takes right inside your CapCut project.
- Auto-Editor: CLI that cuts dead space by loudness or motion and exports MP4 or editor timelines; now also an agent skill.
- AutoCut: Edit video in a text editor: Whisper transcribes, you delete lines in Markdown, AutoCut cuts the video.
- Cut Video: Claude Code skill that tightens long recordings with forced-aligned cuts, keeping laughs and comedic pauses.
- Videogrep: Search spoken words across videos and auto-build supercuts of every match.
- PySceneDetect: Find cuts, fades and scene changes in a video, then split it or export scene lists for the edit.
Run the generation on varg
varg is one API key for the models these skills drive — Seedance 2.5, Kling 4, Veo 3.1, Wan 3, Nano Banana 2.1, GPT Image 2.5, ElevenLabs v4 and 200+ more — billed in credits, with a cost estimate before anything runs.
- Claude Code:
claude plugin marketplace add vargHQ/skills && claude plugin install varg@varg(skill + hosted MCP) - Codex, Cursor, Gemini CLI, OpenClaw:
npx skills add vargHQ/skills --skill varg-ai -a <agent> -y - Grok and any agent that reads SKILL.md:
npx skills add vargHQ/skills --skill varg-ai, or point it at the skill's SKILL.md - Claude.ai, ChatGPT, any MCP client: connect
https://mcp.varg.ai/mcp - API key: https://varg.ai/dashboard/api · Model specs: https://varg.ai/models/llms.txt