Viral Reels Creator — Full Video Production Suite

A comprehensive ffmpeg-based skill for creating viral Instagram Reels, TikTok videos, and YouTube Shorts — including finding the best moments in raw footage, beat-synced editing, multi-platform export, and viral content strategy.

Quick Start — Read Before Doing Anything

1. Install dependencies if not available:

   apt-get update && apt-get install -y ffmpeg fonts-noto fonts-noto-color-emoji fontconfig
   # For scene detection + beat sync:
   pip install --break-system-packages opencv-python numpy librosa soundfile
   

2. Check uploaded files: Look in /mnt/user-data/uploads/ for user assets.
3. Copy assets to /home/claude/ before processing.
4. Output all final videos to /mnt/user-data/outputs/ and present to user.

Platform Specs (Primary: Instagram Reels)

Base output flags (always use):

-c:v libx264 -preset slow -crf 18 -pix_fmt yuv420p -r 30 -c:a aac -ar 44100 -ac 2 -movflags +faststart

For full platform-specific specs and export commands → references/platform-specs.md

Workflow Decision Tree

User Request
├── "Find best moments / highlights"?
│   └── Run scripts/scene-detect.py → get timestamps → trim clips
├── "Make it viral" / "viral hook"?
│   └── Load references/viral-hooks.md → apply hook formula
├── Single video input?
│   ├── Trim/Cut           → Section: TRIMMING
│   ├── Add captions       → load references/caption-templates.md
│   ├── Add music          → Section: AUDIO MIXING
│   ├── Sync cuts to beats → load references/beat-sync.md + scripts/beat-detect.py
│   ├── Color grade        → load references/color-grading.md
│   ├── Speed change       → Section: SPEED RAMP
│   ├── Add effects/anim   → load references/animations.md
│   └── Resize only        → Section: RESIZING
├── Multiple video inputs?
│   └── Merge with transitions → load references/transitions.md
├── Images input (slideshow)?
│   └── Ken Burns / Slideshow → Section: SLIDESHOW
├── Export for multiple platforms?
│   └── Run scripts/batch-export.sh → exports all platforms at once
├── Full reel from scratch?
│   └── Combine all: scene detect → hook → content → beat sync → captions + music
└── "Make it look good" / generic?
    └── Auto-resize + color grade (Golden Hour or Clarendon) + caption template + fade transitions

IMPORTANT: For captions, animations, color grading, transitions, viral strategy, beat sync, and platform specs — always read the corresponding reference file before generating ffmpeg commands.

FINDING THE BEST MOMENTS (Scene Detection)

When the user has raw footage and wants the best clips extracted:

# Find top 10 most visually interesting moments
python scripts/scene-detect.py raw_footage.mp4 --top 10 --thumbnail

# Extract a specific number of clips for a 30s reel (5s each = 6 clips)
python scripts/scene-detect.py raw_footage.mp4 --top 6 --clip-duration 5 --thumbnail

# For fast-paced content with many scene changes
python scripts/scene-detect.py video.mp4 --threshold 0.2 --top 15

# Output as JSON for further processing
python scripts/scene-detect.py video.mp4 --json > moments.json

Then trim each moment using the generated ffmpeg commands from the script output.

VIRAL STRATEGY — Quick Reference

Load references/viral-hooks.md for the full strategy. Core principles:

1. Hook in 3 seconds — pattern interrupt, bold claim, or curiosity gap in text overlay
2. Retention triggers at 7s, 12s, 20s, 25s to fight viewer drop-off
3. Beat-sync all cuts — biggest single factor in watch time and shares
4. End with a loop — last frame should nearly match first frame so viewers rewatch
5. On-screen text always — 85% of viewers watch without sound

The viral reel formula:

[0–3s]  HOOK     → Bold text claim or pattern interrupt
[3–7s]  SETUP    → Context, "here's what you'll see"
[7–20s] CONTENT  → Core value, fast cuts, visual variety
[20–25s] PAYOFF  → Reveal, transformation, punchline
[25–30s] CTA     → "Save this", "Comment X for part 2"

BEAT-SYNCED EDITING — Quick Reference

Load references/beat-sync.md for the full workflow.

# Step 1: Detect beats
python scripts/beat-detect.py music.mp3
# → outputs timestamps like: 0.469, 0.938, 1.407, 1.876...

# Step 2: Use beat timestamps as cut points
# (see beat-sync.md for full ffmpeg multi-clip assembly)

# Step 3: Add zoom pulse on every beat (120 BPM = 2Hz)
ffmpeg -i video.mp4 -filter_complex "
  zoompan=z='1+0.05*abs(sin(PI*t*2))':
    x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)':d=1:s=1080x1920:fps=30
" OUTPUT_FLAGS zoom_pulse.mp4

Core Operations

RESIZING — Fit Any Video to 9:16

# Scale + crop (fill frame — preferred for Reels)
ffmpeg -i input.mp4 -vf "scale=1080:1920:force_original_aspect_ratio=increase,crop=1080:1920" OUTPUT_FLAGS output.mp4

# Blurred background fill (cinematic — best for landscape source)
ffmpeg -i input.mp4 -filter_complex "
[0:v]scale=1080:1920:force_original_aspect_ratio=increase,crop=1080:1920,boxblur=20:5[bg];
[0:v]scale=1080:1920:force_original_aspect_ratio=decrease[fg];
[bg][fg]overlay=(W-w)/2:(H-h)/2
" OUTPUT_FLAGS output.mp4

TRIMMING — Cut Clips

# Fast trim (copy — no re-encode)
ffmpeg -ss 00:00:05 -i input.mp4 -t 00:00:30 -c copy trimmed.mp4

# Precise trim (re-encode for frame accuracy)
ffmpeg -i input.mp4 -ss 00:00:05 -to 00:00:35 OUTPUT_FLAGS trimmed.mp4

SPEED RAMP — Slow Motion & Fast Forward

# 0.5x slow motion
ffmpeg -i input.mp4 -filter_complex "[0:v]setpts=2.0*PTS[v];[0:a]atempo=0.5[a]" -map "[v]" -map "[a]" OUTPUT_FLAGS slow.mp4

# 2x speed
ffmpeg -i input.mp4 -filter_complex "[0:v]setpts=0.5*PTS[v];[0:a]atempo=2.0[a]" -map "[v]" -map "[a]" OUTPUT_FLAGS fast.mp4

# Speed ramp: normal → slow → normal
ffmpeg -i input.mp4 -filter_complex "
[0:v]trim=0:3,setpts=PTS-STARTPTS[v1];
[0:v]trim=3:6,setpts=2*(PTS-STARTPTS)[v2];
[0:v]trim=6:10,setpts=PTS-STARTPTS[v3];
[v1][v2][v3]concat=n=3:v=1:a=0[v]
" -map "[v]" OUTPUT_FLAGS ramp.mp4

AUDIO MIXING — Add Background Music

# Mix original audio + music (music at 30% volume)
ffmpeg -i video.mp4 -i music.mp3 -filter_complex "
[0:a]volume=1.0[voice];
[1:a]volume=0.3[music];
[voice][music]amix=inputs=2:duration=shortest
" -map 0:v OUTPUT_FLAGS output.mp4

# Fade music in/out
ffmpeg -i video.mp4 -i music.mp3 -filter_complex "
[1:a]afade=t=in:st=0:d=3,afade=t=out:st=27:d=3,volume=0.3[music];
[0:a]volume=1.0[voice];
[voice][music]amix=inputs=2:duration=shortest
" -map 0:v OUTPUT_FLAGS output.mp4

SLIDESHOW — Images to Video

# Ken Burns (zoom + pan)
ffmpeg -loop 1 -i image.jpg -vf "
scale=2160:3840,
zoompan=z='min(zoom+0.001,1.5)':x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)':d=150:s=1080x1920:fps=30
" -t 5 OUTPUT_FLAGS kenburns.mp4

Advanced Features — Reference Files

Always read the reference file before generating ffmpeg commands for these features.

Scripts

Combining Multiple Effects

Stack effects in a single -filter_complex. Order matters:

1. Scale/crop to 1080x1920
2. Apply color grading
3. Apply speed changes
4. Apply animations/effects
5. Overlay captions/text
6. Add transitions (for multi-clip)

Example: Full viral reel pipeline

ffmpeg -i clip1.mp4 -i music.mp3 -filter_complex "
[0:v]scale=1080:1920:force_original_aspect_ratio=increase,crop=1080:1920,
eq=brightness=0.06:saturation=1.4:contrast=1.05,
vignette=PI/5,
drawtext=text='Wait for it...':fontfile=/usr/share/fonts/truetype/noto/NotoSans-Bold.ttf:
  fontsize=72:fontcolor=white:borderw=4:bordercolor=black:
  x=(w-tw)/2:y=(h-th)/2:enable='between(t,0,3)',
drawtext=text='Your Caption Here':fontfile=/usr/share/fonts/truetype/noto/NotoSans-Bold.ttf:
  fontsize=56:fontcolor=white:borderw=3:bordercolor=black:
  x=(w-tw)/2:y=h-th-250:enable='between(t,3,28)',
drawtext=text='Save this!':fontfile=/usr/share/fonts/truetype/noto/NotoSans-Bold.ttf:
  fontsize=40:fontcolor=yellow:borderw=3:bordercolor=black:
  x=(w-tw)/2:y=h-th-120:enable='between(t,25,30)',
fade=t=in:st=0:d=1,fade=t=out:st=28:d=2[v];
[1:a]afade=t=in:st=0:d=2,afade=t=out:st=27:d=3,volume=0.4[a]
" -map "[v]" -map "[a]" -t 30 \
-c:v libx264 -preset slow -crf 18 -pix_fmt yuv420p -r 30 \
-c:a aac -ar 44100 -ac 2 -movflags +faststart \
/mnt/user-data/outputs/reel_final.mp4

Multi-Platform Export

After finishing the reel, export for all platforms at once:

bash scripts/batch-export.sh /mnt/user-data/outputs/reel_final.mp4 my_reel all
# Creates: my_reel_instagram.mp4, my_reel_tiktok.mp4, my_reel_youtube_shorts.mp4,
#          my_reel_facebook.mp4, my_reel_snapchat.mp4, my_reel_pinterest.mp4

Or export to specific platforms:

bash scripts/batch-export.sh reel.mp4 cooking_video instagram,tiktok,youtube

Watermark / Logo Overlay

# Animated logo (fade in, semi-transparent)
ffmpeg -i video.mp4 -i logo.png -filter_complex "
[1:v]scale=120:-1,format=rgba,colorchannelmixer=aa=0.7[logo];
[0:v][logo]overlay=W-w-30:H-h-30
" OUTPUT_FLAGS output.mp4

Auto-Captions from Audio (Whisper)

pip install --break-system-packages openai-whisper pysrt

import whisper

model = whisper.load_model("base")
result = model.transcribe("audio.mp3", word_timestamps=True)

with open("captions.srt", "w") as f:
    for i, seg in enumerate(result["segments"], 1):
        start = seg["start"]
        end = seg["end"]
        text = seg["text"].strip()
        f.write(f"{i}\n")
        f.write(f"{int(start//3600):02}:{int(start%3600//60):02}:{start%60:06.3f} --> ")
        f.write(f"{int(end//3600):02}:{int(end%3600//60):02}:{end%60:06.3f}\n")
        f.write(f"{text}\n\n")

Then burn captions (see references/caption-templates.md for styled options):

ffmpeg -i video.mp4 -vf "subtitles=captions.srt:force_style='FontSize=22,PrimaryColour=&Hffffff,OutlineColour=&H000000,Outline=2,Alignment=2'" OUTPUT_FLAGS output.mp4

Error Handling & Tips

• Check for audio: ffprobe -i input.mp4 -show_streams -select_streams a
• Silent video? Add -an or generate silent audio: -f lavfi -i anullsrc
• Fonts missing? apt-get install -y fonts-noto fonts-liberation
• Preview fast: Add -t 5 -preset ultrafast to render only 5 seconds
• Emoji in text? Use fontfile=/usr/share/fonts/truetype/noto/NotoColorEmoji.ttf
• Always use -movflags +faststart for web/mobile playback
• Duration limits: 90s Instagram/Facebook · 60s TikTok/YouTube Shorts/Snapchat

Output Checklist

Before delivering the final video:

• Resolution is 1080×1920
• Duration within platform limit (90s Instagram, 60s TikTok/Shorts)
• FPS is 30 (or 60 for YouTube Shorts)
• Has audio track (even if silent)
• Uses H.264 + AAC codecs
• Hook text visible in first 3 seconds
• CTA text visible in last 5 seconds
• File is in /mnt/user-data/outputs/
• Platform variants exported via batch-export.sh if needed