Claude Skill

wedding-highlight

Wedding highlight editor for Claude: turns a day of wedding footage, photos, vows and speeches into a 3-5 min cinematic highlight film, a 60 s vertical teaser and a ceremony cut, cut on the music with the vows as the story spine. Western weddings, elopements, engagement and anniv

LLM Mart · 0 points · 0 views 0 listing impressions 0 install-command copies
Download viprasol-tech-claude-video-editing-skills-plugins_video-studio_skills_wedding-highlight-0293497.zip · 4 KB

Install

skills CLI npx skills add https://github.com/Viprasol-Tech/claude-video-editing-skills/tree/main/plugins/video-studio/skills/wedding-highlight
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install viprasol-tech-claude-video-editing-skills@llmmart
Git git clone https://github.com/Viprasol-Tech/claude-video-editing-skills.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole viprasol-tech/claude-video-editing-skills collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

Wedding Highlight

The couple's day as a 3-5 minute film that makes the family cry at the right moment, plus a teaser they can post this week.

  • Best for: Western weddings, elopements, civil ceremonies, engagement and anniversary films
  • Inputs: clips from the day (prep, ceremony, portraits, party), optional photos, the vows/speech audio (camera or lav), one licensed song
  • Default skin: cinematic-phone · alternatives soft-gradient, minimal-mono, frame-studio
  • Formats: 16:9 highlight (master), 9:16 teaser, 16:9 ceremony
  • Durations: teaser 30-60 s, highlight 3-5 min, ceremony as it happened (lightly cut)
  • Genre preset: --genre wedding-highlight (see python CORE/scripts/editor.py genre wedding-highlight)

Before you start

  • This skill needs video-studio-core in the same skills folder. CORE = that folder, SKILLS = the skills folder.
  • First run on a machine: python CORE/scripts/check_env.py and fix what it reports.
  • Ask before you cut: the couple's names (spelling), the date, the song (and its licence), which family members must appear, anything that must NOT appear (an ex, a fall, a guest who did not consent to filming).
  • Consent and privacy: wedding footage shows guests and often children. Deliver privately; post publicly only with the couple's OK, and blur or drop guests who asked not to be shown.
  • Copy the card dumps to a work area first. Never edit inside the camera folders or the skill folders.

How a pro cuts this

Structure (highlight, % of length)

beat % what goes here
prep / anticipation 15 details (rings, dress, invitations), getting ready, nerves, a letter being read
first look / arrival 15 the reveal, walk down the aisle, the partner's face
vows / ceremony 25 the vows as the spoken spine, ring exchange, the kiss, recessional
speeches 15 2-3 best lines (laugh + tear), cut as soundbites over reactions
celebration peak 20 entrance, first dance, party at the song's biggest section
send-off 10 sparklers / exit / last portrait, end on a resolved bar

Pacing. ASL 2-4 s; let emotional shots hold 4-6 s (a tear, a parent's face) and run party shots at 1-2 s. Visual change at least every 4 s. Vary shot length: never a metronome.

Cut rules.

  • Story comes from sound: lay the vows and speech lines first (radio edit), then cover with picture.
  • Cut on the beat in the party, on the phrase in the ceremony. Keep the breath before an emotional line.
  • No visible jump cuts in vows or speeches: cover every join with B-roll of the couple, guests or details.
  • J-cut the vows 0.5-1.5 s under the last music shot; L-cut a laugh line over the reaction it caused.
  • Never fake a reaction (a guest crying at a different line) and never splice words into vows the person did not say.

Transition grammar. Straight cuts on beats by default; one dissolve family for time passing (prep to ceremony), a dip to black before the send-off. Whip only in the party section. No more than two transition kinds per film.

Music and sound. The song leads (bed ~-6 dB), dips under vows and speech (duck: "speech-aware"). Cut the song to length on bar lines (fitTo: "beats") and end on a hit, not a fade. Keep natural sound (cheers, laughter, glasses) under the music in the party. SFX almost none (a soft whoosh on a whip at most).

Colour. wedding-pastel (soft, lifted blacks) or portra for warm skin; kodak-2383 for a cinematic feel. Intensity 0.6-0.8. Match cameras first (color.match), skin before look.

Captions. None in the highlight. In the teaser and the ceremony cut, vow subtitles with "preset": "clean". Couple names + date as a titlecard or text at the open or the end.

What the client judges. Emotion, that every important person is in it (parents, grandparents, wedding party), skin that looks like them, the song, and getting the teaser fast (within days).

Common mistakes. Opening on a logo; using every drone shot; music too loud over vows; uncovered jump cuts in speeches; a missing grandparent; party shots all the same length; a fade-out in the middle of a phrase.

Make the edit

The montage engine cuts on the song's beats. Clips are ordered by file modification time (capture order), so check the order first with --dry-run.

# 1. beat-cut assembly of the day (clips + photos), 16:9, ~4 min, music bakes into the assembled audio
python CORE/scripts/timeline.py auto day/ --work wh --mode montage --music song.wav --duration 240 --format 16:9 --dry-run
#    edit wh/timeline_spec.json: reorder clips into the beat structure, drop weak shots, set "transition"
python CORE/scripts/run.py - --timeline wh/timeline_spec.json --style SKILLS/cinematic-phone --genre wedding-highlight --format 16:9 --no-cut --model base --work wh --stop-after plan
# 2. look at it
python CORE/scripts/render.py wh/plan.json --frames 2,20,60,120,200
python CORE/scripts/render.py wh/plan.json
python CORE/scripts/finish.py wh/plan.json --platform youtube

Quick one-shot version (auto order, no spec editing): python CORE/scripts/run.py - --clips day/ --auto montage --music song.wav --duration 240 --format 16:9 --style SKILLS/cinematic-phone --genre wedding-highlight --model base --work wh --stop-after plan.

Vows-led cut (vows audio under picture). Transcribe the vows clip on its own, then build a spec where the vows clip keeps its audio and the B-roll clips are muted:

python CORE/scripts/transcribe.py --work vows --audio vows.mp4 --model base   # gives vows/transcript.json for "words"
python CORE/scripts/run.py - --timeline wh/vows_spec.json --style SKILLS/cinematic-phone --genre wedding-highlight --format 16:9 --music song.wav --model base --work whv --stop-after plan

In vows_spec.json give the vows clip "audio": "keep", "words": "../vows/transcript.json" and a "jcut": 1.0; give B-roll "audio": "mute". --music then goes into the plan, where finish.py ducks it under the vows.

Teaser (9:16, 45 s): repeat step 1 with --format 9:16 --duration 45 --work wt and pick only the best 15-20 shots.

Refine the plan

Edit wh/plan.json before the full render:

  • Remove creator captions from the highlight: delete the captions element (the skin adds one; the genre wants none). Also clear words when the audio is only music (Whisper may "hear" lyrics).
  • Montage audio is music, not dialogue: add "audio": {"platform": "youtube", "voice": false} so the dialogue chain does not compress the song. Vows-led cut: keep the chain, use "voice": {"denoise": "light", "eq": "warm"}.
  • Music (vows-led cut): "music": {"src": "song.wav", "under": 6, "duck": "speech-aware", "fitTo": "beats", "endButton": true}.
  • Colour: "color": {"look": "wedding-pastel", "intensity": 0.7} (or portra, kodak-2383); "finish": {"grain": 4, "vignette": 0.3, "letterbox": 0}. Mixed cameras: "color": {"match": "hero_frame.png"}.
  • Title: {"type": "titlecard", "kicker": "12.10.2026", "title": "Anna & Tom", "start": 0, "end": 3.5, "in": "blurIn", "out": "fade"} or an end title in the last 4 s. Check spelling with the couple. titlecard is an opaque full frame (it hides the footage); to put the names over a shot use {"type": "text", "text": "Anna & Tom", "y": 900, "in": "maskUp"}.
  • Vow subtitles (teaser/ceremony): {"type": "captions", "preset": "clean", "dodgeFace": true}; fix names in words.
  • Timeline tweaks (in the spec, then rebuild): "speedRamp" into the kiss or the confetti, "freezeEnd": 1.0 on the last portrait, "transition": {"type": "dissolve", "dur": 0.8} between parts of the day, "dip-black" before the exit.
  • Run python CORE/scripts/editor.py review wh/plan.json --genre wedding-highlight and read review.md.
  • Footage clean-up (video, before the spec): guest-cam handheld, dark reception and venue LED/tube light: python CORE/scripts/prep.py c07.mp4 --work fix/c07 --stabilize --denoise light --deflicker, then use fix/c07/source.mp4 in the spec. --stabilize = vidstab (crops ~6% at the edges), --denoise light|strong = video noise, --deflicker = mains-light flicker. Only on clips that need it.

Deliver

python CORE/scripts/deliver.py wh/plan.json --to youtube,prores-422hq --out-dir wh/deliver --burn no
python CORE/scripts/deliver.py wt/plan.json --to reels,tiktok --out-dir wt/deliver
python CORE/scripts/qc.py wh/deliver/final_youtube.mp4 --plan wh/plan.json --platform youtube --strict
python CORE/scripts/qc.py wt/deliver/final_reels.mp4 --plan wt/plan.json --platform reels --strict

Name files couple_highlight_v001_16x9.mp4, couple_teaser_v001_9x16.mp4. The full ceremony is a separate edit: run run.py ceremony.mp4 ... --no-cut (or multicam.py camA camB --out ceremony/multicam.mp4 for two angles).

Check before delivering

  • Contact sheet: skin looks natural, horizon level, no guest who asked not to be filmed, names spelled right.
  • Watch the vows section with headphones: music dips, every word clear, no click at joins, no uncovered jump cut.
  • Every key family member appears at least once (ask the couple for the list and tick it off).
  • The film ends on a downbeat, not mid-phrase; qc.py passes loudness (-14 LUFS, -1 dBTP) and has no flash frames.
  • Tell the couple what is in each file, the song credit, and how many revision rounds remain (two is common).

Example requests

  • "Here's our wedding footage and our song, make a 4-minute highlight with the vows in it."
  • "Make a 45-second vertical teaser for Instagram from the best moments."
  • "Cut the ceremony as its own video, both cameras, keep everything they said."
Files (claude-video-editing-skills)
  • SKILL.md 10.2 KB
    ---
    name: wedding-highlight
    description: "Wedding highlight editor for Claude: turns a day of wedding footage, photos, vows and speeches into a 3-5 min cinematic highlight film, a 60 s vertical teaser and a ceremony cut, cut on the music with the vows as the story spine. Western weddings, elopements, engagement and anniversary films. Use when the user says 'edit my wedding video', 'make a wedding highlight', 'wedding teaser for Instagram', 'cut our vows into a film', 'same-day edit', 'wedding trailer', or hands over wedding clips plus a song. Runs on video-studio-core (timeline, beats, music fit, colour, deliver, QC). For Indian multi-event weddings use wedding-india."
    license: MIT
    ---
    
    # Wedding Highlight
    
    The couple's day as a 3-5 minute film that makes the family cry at the right moment, plus a teaser they can post this week.
    
    - **Best for:** Western weddings, elopements, civil ceremonies, engagement and anniversary films
    - **Inputs:** clips from the day (prep, ceremony, portraits, party), optional photos, the vows/speech audio (camera or lav), one licensed song
    - **Default skin:** `cinematic-phone` · alternatives `soft-gradient`, `minimal-mono`, `frame-studio`
    - **Formats:** 16:9 highlight (master), 9:16 teaser, 16:9 ceremony
    - **Durations:** teaser 30-60 s, highlight 3-5 min, ceremony as it happened (lightly cut)
    - **Genre preset:** `--genre wedding-highlight` (see `python CORE/scripts/editor.py genre wedding-highlight`)
    
    ## Before you start
    
    - This skill needs **video-studio-core** in the same skills folder. `CORE` = that folder, `SKILLS` = the skills folder.
    - First run on a machine: `python CORE/scripts/check_env.py` and fix what it reports.
    - Ask before you cut: the couple's names (spelling), the date, the song (and its licence), which family members must
      appear, anything that must NOT appear (an ex, a fall, a guest who did not consent to filming).
    - **Consent and privacy:** wedding footage shows guests and often children. Deliver privately; post publicly only
      with the couple's OK, and blur or drop guests who asked not to be shown.
    - Copy the card dumps to a work area first. Never edit inside the camera folders or the skill folders.
    
    ## How a pro cuts this
    
    **Structure (highlight, % of length)**
    
    | beat | % | what goes here |
    | --- | --- | --- |
    | prep / anticipation | 15 | details (rings, dress, invitations), getting ready, nerves, a letter being read |
    | first look / arrival | 15 | the reveal, walk down the aisle, the partner's face |
    | vows / ceremony | 25 | the vows as the spoken spine, ring exchange, the kiss, recessional |
    | speeches | 15 | 2-3 best lines (laugh + tear), cut as soundbites over reactions |
    | celebration peak | 20 | entrance, first dance, party at the song's biggest section |
    | send-off | 10 | sparklers / exit / last portrait, end on a resolved bar |
    
    **Pacing.** ASL 2-4 s; let emotional shots hold 4-6 s (a tear, a parent's face) and run party shots at 1-2 s.
    Visual change at least every 4 s. Vary shot length: never a metronome.
    
    **Cut rules.**
    - Story comes from sound: lay the vows and speech lines first (radio edit), then cover with picture.
    - Cut on the beat in the party, on the phrase in the ceremony. Keep the breath before an emotional line.
    - No visible jump cuts in vows or speeches: cover every join with B-roll of the couple, guests or details.
    - J-cut the vows 0.5-1.5 s under the last music shot; L-cut a laugh line over the reaction it caused.
    - Never fake a reaction (a guest crying at a different line) and never splice words into vows the person did not say.
    
    **Transition grammar.** Straight cuts on beats by default; one dissolve family for time passing (prep to ceremony),
    a dip to black before the send-off. Whip only in the party section. No more than two transition kinds per film.
    
    **Music and sound.** The song leads (bed ~-6 dB), dips under vows and speech (`duck: "speech-aware"`). Cut the song to
    length on bar lines (`fitTo: "beats"`) and end on a hit, not a fade. Keep natural sound (cheers, laughter, glasses) under
    the music in the party. SFX almost none (a soft whoosh on a whip at most).
    
    **Colour.** `wedding-pastel` (soft, lifted blacks) or `portra` for warm skin; `kodak-2383` for a cinematic feel.
    Intensity 0.6-0.8. Match cameras first (`color.match`), skin before look.
    
    **Captions.** None in the highlight. In the teaser and the ceremony cut, vow subtitles with `"preset": "clean"`.
    Couple names + date as a `titlecard` or `text` at the open or the end.
    
    **What the client judges.** Emotion, that every important person is in it (parents, grandparents, wedding party),
    skin that looks like them, the song, and getting the teaser fast (within days).
    
    **Common mistakes.** Opening on a logo; using every drone shot; music too loud over vows; uncovered jump cuts in
    speeches; a missing grandparent; party shots all the same length; a fade-out in the middle of a phrase.
    
    ## Make the edit
    
    The montage engine cuts on the song's beats. Clips are ordered by file modification time (capture order), so check
    the order first with `--dry-run`.
    
    ```bash
    # 1. beat-cut assembly of the day (clips + photos), 16:9, ~4 min, music bakes into the assembled audio
    python CORE/scripts/timeline.py auto day/ --work wh --mode montage --music song.wav --duration 240 --format 16:9 --dry-run
    #    edit wh/timeline_spec.json: reorder clips into the beat structure, drop weak shots, set "transition"
    python CORE/scripts/run.py - --timeline wh/timeline_spec.json --style SKILLS/cinematic-phone --genre wedding-highlight --format 16:9 --no-cut --model base --work wh --stop-after plan
    # 2. look at it
    python CORE/scripts/render.py wh/plan.json --frames 2,20,60,120,200
    python CORE/scripts/render.py wh/plan.json
    python CORE/scripts/finish.py wh/plan.json --platform youtube
    ```
    
    Quick one-shot version (auto order, no spec editing): `python CORE/scripts/run.py - --clips day/ --auto montage --music song.wav --duration 240 --format 16:9 --style SKILLS/cinematic-phone --genre wedding-highlight --model base --work wh --stop-after plan`.
    
    **Vows-led cut (vows audio under picture).** Transcribe the vows clip on its own, then build a spec where the vows clip
    keeps its audio and the B-roll clips are muted:
    
    ```bash
    python CORE/scripts/transcribe.py --work vows --audio vows.mp4 --model base   # gives vows/transcript.json for "words"
    python CORE/scripts/run.py - --timeline wh/vows_spec.json --style SKILLS/cinematic-phone --genre wedding-highlight --format 16:9 --music song.wav --model base --work whv --stop-after plan
    ```
    
    In `vows_spec.json` give the vows clip `"audio": "keep"`, `"words": "../vows/transcript.json"` and a `"jcut": 1.0`;
    give B-roll `"audio": "mute"`. `--music` then goes into the plan, where `finish.py` ducks it under the vows.
    
    **Teaser (9:16, 45 s):** repeat step 1 with `--format 9:16 --duration 45 --work wt` and pick only the best 15-20 shots.
    
    ## Refine the plan
    
    Edit `wh/plan.json` before the full render:
    
    - **Remove creator captions** from the highlight: delete the `captions` element (the skin adds one; the genre wants none).
      Also clear `words` when the audio is only music (Whisper may "hear" lyrics).
    - **Montage audio is music, not dialogue:** add `"audio": {"platform": "youtube", "voice": false}` so the dialogue chain
      does not compress the song. Vows-led cut: keep the chain, use `"voice": {"denoise": "light", "eq": "warm"}`.
    - **Music (vows-led cut):** `"music": {"src": "song.wav", "under": 6, "duck": "speech-aware", "fitTo": "beats", "endButton": true}`.
    - **Colour:** `"color": {"look": "wedding-pastel", "intensity": 0.7}` (or `portra`, `kodak-2383`);
      `"finish": {"grain": 4, "vignette": 0.3, "letterbox": 0}`. Mixed cameras: `"color": {"match": "hero_frame.png"}`.
    - **Title:** `{"type": "titlecard", "kicker": "12.10.2026", "title": "Anna & Tom", "start": 0, "end": 3.5, "in": "blurIn", "out": "fade"}`
      or an end title in the last 4 s. Check spelling with the couple. `titlecard` is an opaque full frame (it hides the
      footage); to put the names over a shot use `{"type": "text", "text": "Anna & Tom", "y": 900, "in": "maskUp"}`.
    - **Vow subtitles (teaser/ceremony):** `{"type": "captions", "preset": "clean", "dodgeFace": true}`; fix names in `words`.
    - **Timeline tweaks** (in the spec, then rebuild): `"speedRamp"` into the kiss or the confetti, `"freezeEnd": 1.0` on
      the last portrait, `"transition": {"type": "dissolve", "dur": 0.8}` between parts of the day, `"dip-black"` before the exit.
    - Run `python CORE/scripts/editor.py review wh/plan.json --genre wedding-highlight` and read `review.md`.
    - **Footage clean-up (video, before the spec):** guest-cam handheld, dark reception and venue LED/tube light: `python CORE/scripts/prep.py c07.mp4 --work fix/c07 --stabilize --denoise light --deflicker`, then use `fix/c07/source.mp4` in the spec. `--stabilize` = vidstab (crops ~6% at the edges), `--denoise light|strong` = video noise, `--deflicker` = mains-light flicker. Only on clips that need it.
    
    ## Deliver
    
    ```bash
    python CORE/scripts/deliver.py wh/plan.json --to youtube,prores-422hq --out-dir wh/deliver --burn no
    python CORE/scripts/deliver.py wt/plan.json --to reels,tiktok --out-dir wt/deliver
    python CORE/scripts/qc.py wh/deliver/final_youtube.mp4 --plan wh/plan.json --platform youtube --strict
    python CORE/scripts/qc.py wt/deliver/final_reels.mp4 --plan wt/plan.json --platform reels --strict
    ```
    
    Name files `couple_highlight_v001_16x9.mp4`, `couple_teaser_v001_9x16.mp4`. The full ceremony is a separate edit:
    run `run.py ceremony.mp4 ... --no-cut` (or `multicam.py camA camB --out ceremony/multicam.mp4` for two angles).
    
    ## Check before delivering
    
    - Contact sheet: skin looks natural, horizon level, no guest who asked not to be filmed, names spelled right.
    - Watch the vows section with headphones: music dips, every word clear, no click at joins, no uncovered jump cut.
    - Every key family member appears at least once (ask the couple for the list and tick it off).
    - The film ends on a downbeat, not mid-phrase; `qc.py` passes loudness (-14 LUFS, -1 dBTP) and has no flash frames.
    - Tell the couple what is in each file, the song credit, and how many revision rounds remain (two is common).
    
    ## Example requests
    
    - "Here's our wedding footage and our song, make a 4-minute highlight with the vows in it."
    - "Make a 45-second vertical teaser for Instagram from the best moments."
    - "Cut the ceremony as its own video, both cameras, keep everything they said."
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.