fofr/smart-ffmpeg 🖼️📝🔢 → 🖼️

▶️ 2.5K runs 📅 Aug 2025 ⚙️ Cog 0.14.0

About

Quickly edit, trim, re-encode, concatenate videos using ffmpeg and gpt-5 generated commands

Example Output

Prompt:

"Concatenate these two videos"

Output

Performance Metrics

39.52s Prediction Time

40.11s Total Time

All Input Parameters

{
  "files": [
    "https://replicate.delivery/pbxt/NYZZGpR5tZINdJQLLp8pUziwdNeouXe1HQt8VkGWEzWNRVqw/vpz04dbqv9rme0crnxg9r1x9j8.mp4",
    "https://replicate.delivery/pbxt/NYZZH2dFu1qBUt2PpTKd2ZadLMGT9Z0E9vLzg7hidlLG9NWX/hc7hehnhw5rm80crnxf8j703cm.mp4"
  ],
  "prompt": "Concatenate these two videos",
  "max_attempts": 3
}

Input Parameters

files (required) Type: array: List of input files (images or videos)
prompt (required) Type: string: High-level editing request, e.g., 'combine with crossfade and add watermark'
max_attempts Type: integerDefault: 3Range: 1 - 5: Max ffmpeg retries with model assistance

Output Schema

Output

Type: array • Items Type: string • Items Format: uri

Example Execution Logs

🔎 Detected 2 video(s), 0 image(s)
🎯 Task: Concatenate these two videos
🧠 Attempt 1/3: planning command with GPT-5…
🛠️ Proposed command (attempt 1):

ffmpeg -y -i /tmp/sff_inputs_1755519909196/cog-input-3749997689.mp4 -i /tmp/sff_inputs_1755519909196/cog-input-696354944.mp4 -filter_complex '[0:v:0][1:v:0]concat=n=2:v=1:a=0[outv]' -map '[outv]' -c:v libx264 -pix_fmt yuv420p -r 24 /tmp/sff_outputs_1755519909196/concatenated.mp4

Version Details

Version ID: 28a1f8b717055209a4606b5668e121d5b439578c4be8edb8776117a69297c360
Version Created: August 18, 2025

Run on Replicate →