Spotting a viral TikTok with 5M+ views in your niche and asking "how do I make one of those for my product?" used to mean booking a creator and hoping they could replicate the format. With Kling 2.6 Motion Control inside UGC Copilot, you can clone the actual motion onto your own character in under 15 minutes. This is a step-by-step walkthrough of the clone-video workflow with Motion Control enabled.
Borrow the Structure, Not the Content
Before any walkthrough, the framing matters. Cloning a viral video does not mean copying the footage, the audio, the script, or the brand. It means studying what made the video work — the hook timing, the pacing, the gesture choreography, the camera path — and rebuilding that structure with your own product, your own character, and your own script.
This distinction is both ethical and legal. Structural patterns aren't copyrightable. Performances and footage are. Motion Control transfers motion, not visual content — your output looks nothing like the original reference because your character, scene, and outfit are different. Done right, the reference becomes invisible in the final clip.
Step 1: Pick a Reference That Will Actually Clone Well
Not every viral video is a good reference. The wrong choice will burn 10+ credits with poor output before you figure out it was the source clip's fault, not the model.
What makes a viral video clone-friendly:
- Single creator on screen. Multi-person trends (duets, partner dances) confuse motion transfer.
- Well-lit, vertically framed. Match your output orientation. 9:16 in, 9:16 out.
- Motion-forward. The more movement in the reference, the more there is to transfer.
- Under 30 seconds of usable motion. Motion Control caps at 30s in video orientation. If the viral video is 60s, trim to the most representative segment first.
- Recognizable choreography. The "structural DNA" — the gesture pattern that made it spread — needs to be visible enough to extract.
Save the reference clip locally. You'll upload it in the next step.
Step 2: Start a Clone Video Project
In UGC Copilot, create a new project and select Clone Video as the mode. This unlocks the reference-driven analysis pipeline and the Motion Control toggle on the render queue.
Step 3: Upload Your Reference
Drop the reference clip into the Video Analysis step. UGC Copilot validates the format (MP4/WebM/MOV) and stores it on Firebase Storage. The in-app tip will remind you that Motion Control caps each scene at 30 seconds — if your reference is longer, trim to lead with the most representative motion before uploading.
Step 4: Run Deep Video Analysis (4 credits)
The analysis step uses Gemini Flash with high-thinking mode plus FFmpeg keyframe extraction to break down the reference into a structural blueprint:
- Hook structure and timing
- Pacing and beat changes
- Dialogue rhythm and density
- Scene composition and camera work
- Visual style and color palette
- CTA placement and treatment
This costs 4 credits (~$0.50 on the entry $25 / 200-credit PAYG pack) and runs in about 60–90 seconds. The output becomes the script and structural plan you'll render against.
Step 5: Choose Your Character
Two options:
- AI Twin — Use your existing brand persona. The clone now has your face but the reference's choreography. Best for brand-building campaigns where the persona compounds equity over time.
- Fresh AI Persona — Generate a new persona matched to the reference's demographic. Best when you want to A/B test different "creator looks" against the same proven motion.
Most clone-video advertisers default to AI Twin. The Twin compounds; a fresh persona is one-off.
Step 6: Enable Motion Control on Render
In the render queue, select Kling as your engine. The Motion Control toggle appears with a live cost preview that scales with your selected duration and quality tier — at Standard quality with video orientation it typically reads in the 44–164 cr/scene range; at Pro it reads 88–328 cr/scene. Enable it.
Pick the orientation:
- Video (recommended for full-body trends) — up to 30s, transfers full body motion and camera path
- Image (recommended for hero shots) — up to 10s, preserves character framing while transferring gesture
For most viral TikTok trends, video orientation is the right call. The whole point is to capture the motion as it appeared in the original.
Step 7: Render
Hit render. A typical 8-second scene on Standard quality completes in 4–6 minutes. Pro/HQ takes 8–12 minutes. Three scenes can render in parallel.
Real Numbers: Cost & Time for One Cloned Ad
For a complete 24-second cloned UGC ad split into three 8-second scenes (the standard format):
- Deep video analysis: 4 credits
- 3 scenes × ~44 credits (Standard) = 132 credits
- Optional text overlay: 1 credit per scene = 3 credits
- Total: ~139 credits, roughly $17 on PAYG ($25 / 200 cr pack)
Total wall-clock time: 60 seconds for analysis + 4–6 minutes for parallel renders + 1 minute for export. Roughly 8 minutes from upload to finished ad.
Variation Tactics: 1 Reference × 5 Personas × 2 Angles
The real leverage in Motion Control isn't producing one cloned ad — it's producing 10 of them off one reference. Three multipliers compound:
Multiplier 1: Persona variation
Generate the same cloned motion with five different AI personas (different faces, ages, styles). Same proven choreography, demographic-matched character. Costs 5× the render but gives you a cohort to test against.
Multiplier 2: Hook angle variation
The reference's structure includes a hook beat. Swap two or three different opening lines (problem, identity, contrarian) over the same motion. Each variation is a fresh ad in the algorithm's eyes.
Multiplier 3: Background and product variation
Even with motion locked, the AI Twin's outfit, environment, and held product can shift. This is the cheapest variation lever — same render cost, fresh creative for the algorithm.
Stacked: 1 reference × 5 personas × 2 hook angles × 2 background variations = 20 variations off one source clip. At ~140 credits per variation, that's ~2,800 credits, or ~$350 on PAYG, for an entire month's worth of ad creative built off one viral trend.
When the Clone Misses: Common Fixes
Output looks "off" / character drift
Most often a reference issue. Trim the reference shorter and lead with the cleanest motion segment. Avoid reference clips with multi-person scenes or chaotic cuts.
Lip-sync looks broken
Motion Control transfers the reference's lip movement. If your script timing differs significantly from the reference, the character's mouth will visibly drift from your words. Either match script timing to reference, use cutaways and B-roll to hide the drift, or fall back to standard Kling 3.0 for dialogue-heavy scenes.
Duration is wrong
Motion Control doesn't accept a duration parameter — output length is reference-driven. To get a shorter or longer output, trim or extend the reference clip before uploading.
Wrong orientation produced wrong framing
Switch between 'video' and 'image' character orientation. Video transfers full motion + camera. Image preserves the framing of your character image and only transfers gesture.
Conclusion
Cloning a viral TikTok used to be a shoot day and a creator brief. With Motion Control, it's an upload, a 4-credit analysis, and a render queue. The discipline isn't technical anymore — it's editorial: picking the right reference, picking the right character, and treating "borrow the structure, not the content" as the only ethical framing. Get those right, and one viral source clip becomes a month's worth of ad creative your algorithm has never seen before.
For the full technical breakdown of Kling 2.6 Motion Control, see our Complete Guide to Kling 2.6 Motion Control. For the broader video cloning workflow, see The Ultimate Guide to Video Cloning.