AI Video Post-Processing for TikTok: Make Generated Content Perform

Raw AI video gets buried on TikTok

You generated a great clip with Runway or Kling. The content is compelling. You upload it to TikTok. It gets 200 views and dies.

The problem isn't the content. It's the format.

TikTok's algorithm evaluates videos in the first 0.5 seconds. Wrong aspect ratio, sterile visuals, missing audio normalization, AI metadata flags. Any of these reduce initial distribution. Combined, they kill your reach.

Post-processing fixes all of them. One FFmpeg pipeline turns AI output into content that performs like native TikTok video.

Why AI video underperforms on TikTok

Five specific issues:

Wrong aspect ratio: Most AI generators output 16:9 or 1:1. TikTok is 9:16 (1080x1920). Uploading the wrong ratio means black bars or auto-cropping that cuts important content.
No grain or noise: Phone cameras produce sensor noise. AI video is perfectly clean. TikTok users subconsciously register this as "not real."
AI metadata: Generation tools embed model info, tool identifiers, and creation parameters. Platforms can read this.
Audio issues: AI video often has no audio, or audio at inconsistent levels. TikTok penalizes silent videos and rewards consistent audio.
Too-high quality: Counterintuitive, but TikTok content that's too clean looks like an ad. Users scroll past ads.

The complete post-processing pipeline

Here's the full FFmpeg filter chain that fixes every issue in one pass.

Step 1: Resize to 9:16

ffmpeg -i input.mp4 \
  -vf "scale=1080:1920:force_original_aspect_ratio=decrease,pad=1080:1920:(ow-iw)/2:(oh-ih)/2:black" \
  -c:v libx264 -crf 20 output.mp4

This scales to fit within 1080x1920, then pads with black bars if the aspect ratio doesn't match exactly. For a better look, use a blurred background instead:

ffmpeg -i input.mp4 \
  -filter_complex "[0:v]scale=1080:1920:force_original_aspect_ratio=increase,crop=1080:1920,boxblur=20[bg];[0:v]scale=1080:1920:force_original_aspect_ratio=decrease[fg];[bg][fg]overlay=(W-w)/2:(H-h)/2" \
  -c:v libx264 -crf 20 output.mp4

Step 2: Add film grain

ffmpeg -i input.mp4 \
  -vf "noise=alls=18:allf=t" \
  -c:v libx264 -crf 20 output.mp4

alls=18 is the sweet spot for TikTok. Phone cameras in indoor lighting produce roughly this level of grain.

Step 3: Strip AI metadata

ffmpeg -i input.mp4 -map_metadata -1 -c:v copy -c:a copy clean.mp4

Removes all metadata including AI generation parameters, tool identifiers, and model versions.

Step 4: Normalize audio

ffmpeg -i input.mp4 \
  -af "loudnorm=I=-14:TP=-2:LRA=7" \
  -c:v copy output.mp4

TikTok targets -14 LUFS for audio loudness. The loudnorm filter brings any audio to this standard.

Step 5: Add a captions placeholder track

If you're adding captions later (and you should, 80% of TikTok is watched on mute):

ffmpeg -i input.mp4 \
  -vf "drawtext=text='':fontsize=1:fontcolor=white:x=0:y=0" \
  -c:v libx264 -crf 20 output.mp4

This is a no-op text overlay that ensures the video stream is re-encoded with proper subtitle track compatibility.

Combined pipeline

All steps in one command:

ffmpeg -i input.mp4 \
  -map_metadata -1 \
  -filter_complex "[0:v]scale=1080:1920:force_original_aspect_ratio=decrease,pad=1080:1920:(ow-iw)/2:(oh-ih)/2:black,noise=alls=18:allf=t,eq=brightness=0.01:contrast=1.03:saturation=0.95[v]" \
  -map "[v]" -map 0:a? \
  -af "loudnorm=I=-14:TP=-2:LRA=7" \
  -c:v libx264 -crf 22 -preset medium \
  -c:a aac -b:a 128k \
  -movflags +faststart \
  output.mp4

The -movflags +faststart moves the moov atom to the beginning of the file for faster playback start on mobile.

Automate with RenderIO

Processing one video locally takes 30-60 seconds. Processing 50 videos per day means an hour of compute time on your machine.

Send it to the API instead:

curl -X POST https://renderio.dev/api/v1/run-ffmpeg-command \
  -H "X-API-KEY: your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "ffmpeg_command": "-i {{in_video}} -map_metadata -1 -filter_complex \"[0:v]scale=1080:1920:force_original_aspect_ratio=decrease,pad=1080:1920:(ow-iw)/2:(oh-ih)/2:black,noise=alls=18:allf=t,eq=brightness=0.01:contrast=1.03:saturation=0.95[v]\" -map \"[v]\" -map 0:a? -af \"loudnorm=I=-14:TP=-2:LRA=7\" -c:v libx264 -crf 22 -preset medium -c:a aac -b:a 128k -movflags +faststart {{out_video}}",
    "input_files": { "in_video": "https://example.com/runway-output.mp4" },
    "output_files": { "out_video": "tiktok-ready.mp4" }
  }'

Response:

{
  "command_id": "cmd_abc123",
  "status": "processing"
}

Poll for completion:

curl https://renderio.dev/api/v1/commands/cmd_abc123 \
  -H "X-API-KEY: your_api_key"

Batch processing with JavaScript

const aiVideos = [
  { url: "https://storage.example.com/gen-1.mp4", name: "tiktok-1" },
  { url: "https://storage.example.com/gen-2.mp4", name: "tiktok-2" },
  { url: "https://storage.example.com/gen-3.mp4", name: "tiktok-3" },
];

const FFMPEG_TIKTOK_PIPELINE = `-i {{in_video}} -map_metadata -1 -filter_complex "[0:v]scale=1080:1920:force_original_aspect_ratio=decrease,pad=1080:1920:(ow-iw)/2:(oh-ih)/2:black,noise=alls=18:allf=t,eq=brightness=0.01:contrast=1.03:saturation=0.95[v]" -map "[v]" -map 0:a? -af "loudnorm=I=-14:TP=-2:LRA=7" -c:v libx264 -crf 22 -c:a aac -b:a 128k -movflags +faststart {{out_video}}`;

const jobs = aiVideos.map(video =>
  fetch("https://renderio.dev/api/v1/run-ffmpeg-command", {
    method: "POST",
    headers: {
      "X-API-KEY": "your_api_key",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      ffmpeg_command: FFMPEG_TIKTOK_PIPELINE,
      input_files: { in_video: video.url },
      output_files: { out_video: `${video.name}.mp4` },
    }),
  })
);

const results = await Promise.all(jobs);

All three videos process in parallel on Cloudflare's edge. Total time: roughly the same as processing one.

Results you can expect

After implementing this pipeline, AI video creators report:

3-5x more views in the first hour compared to raw AI uploads
Higher completion rates because the video "feels" native
No AI content flags from metadata stripping
Consistent audio that doesn't make users reach for the volume

The entire pipeline costs pennies per video on RenderIO. Plans start at $9/mo (Starter, 500 commands), with the Pro plan at$ 49/mo covering 5,000 commands and the Business plan at $99/mo covering 20,000 commands.

The Growth plan at $29/mo covers 1,000 commands. Scale to Pro ($ 49/mo) for higher volume.