How to Clone Any UGC Video with AI in 18 Minutes
Swap creators and regenerate any UGC video with a brand new AI character — no filming, no actors, no expensive gear.
Original Tutorial by @mikefutia
This guide is based on the viral X thread demonstrating the 6-step UGC cloning workflow.
Watch Original Video🛠️ Tools You'll Need
Seedance 2.0
AI video generator for shot-by-shot UGC recreation
ElevenLabs
Voice cloning & consistent AI voice across clips
CapCut
Free video editor for cutting and final assembly
Midjourney / DALL-E
Generate your new AI character reference image
TikTok / Meta Ads Library
Find high-performing UGC videos to clone
Descript (Optional)
Transcribe audio and adjust timing easily
📋 The 6-Step Process (Timestamps Included)
Find the Video to Clone
Start by sourcing a high-performing UGC video from TikTok, Instagram Reels, or the Meta Ads Library. Look for videos with proven engagement — these are your winning templates.
💡 Pro Tip: Choose videos under 60 seconds with a single creator talking directly to camera. These clone the cleanest.
Generate Your New Character
Use Midjourney, DALL-E 3, or Google ImageFX to create a reference image of your new AI character. Match the general vibe (age, setting, style) of the original for best results.
📸 Prompt example: "A friendly 28-year-old woman holding a skincare bottle in a bright bathroom, natural lighting, smartphone selfie style, photorealistic, 4K"
Save this image — you'll use it as the character reference for every clip to maintain consistency.
Cut the Original into Clips
Import the original video into CapCut (or any editor) and slice it into 3-5 second segments. Each segment should represent one distinct shot or action.
- Cut on scene changes or natural pauses
- Export each clip individually
- Note the audio/script for each segment
Generate Clip 1 in Seedance
Head to Seedance 2.0 and upload your character reference image. Write a detailed prompt describing the exact action, expression, and camera angle from the original clip 1.
Prompt structure: [Character from ref image] + [specific action] + [camera angle] + [setting] + [emotion] + [duration]
Generate and iterate until the first clip perfectly matches the pacing and motion of the original.
Lock the Voice Across All Clips
This is where most people mess up — voice consistency. Use ElevenLabs to:
- Clone a voice (or select a stock voice
Comments