Kling 3.0 Motion Control: AI Motion Transfer, Step by Step

Say you've filmed a 12-second routine for a trend and want your AI character, or your brand's mascot, to perform it instead. That's motion transfer: Kling 3.0 Motion Control reads the movement in your clip (steps, gestures, expressions, timing) and renders it with the character from a single image.

It lives in Spicy Studio's Character Swap; this link opens it with Kling 3.0 Motion Control selected. In the Video Lab you describe motion in words. Here it comes from footage, so your clip matters more than any prompt.

Character image (required): the person or design that appears in the result, up to 10 MB, clear and evenly lit.

Motion video (required): the driving clip, 3 to 30 seconds long.

Keep Original Audio (on by default): leaves the driving clip's soundtrack on the result; switch it off if you'll add your own track or the audio isn't yours to reuse.

Prompt and negative prompt (optional): a note on light or mood (“golden-hour light, warm tones”) and what to keep out (“extra people, blurred face”).

Billing follows the driving video: its length is rounded up to the next whole second, with a 3-second minimum.

Standard: 20 credits a second: 60 for 3 seconds, 200 for 10, 600 for 30.

Pro: 27 credits a second: 81 for 3 seconds, 270 for 10, 810 for 30.

A 7.4-second clip is billed as 8 seconds (160 or 216 credits), so trim dead air from both ends. The 25 free credits on a new account don't reach the 60-credit minimum; a €2 Spark pack (90 credits, no expiry) covers a 3- or 4-second Standard test. See pricing.

Check your footage before uploading:

One performer. Kling's own Motion Control page says it supports one primary subject and favors the dominant one when several faces appear.

Visible from start to finish. Keep the face and moving limbs in frame. If the performer turns away or covers their face, the Playbook suggests the lip-sync route instead.

One continuous take, trimmed tight. No cuts or jump edits. You pay for every uploaded second.

A steady camera and a plain background, such as a phone on a tripod in front of a wall.

Motion transfer works best when your character starts the way the performer starts. The Playbook's method: screenshot the driving video's first frame, then recreate it with your character in the Image Lab, keeping the pose, framing and camera angle. Give the model the screenshot plus a reference image of your character; Nano Banana 2.1 does it for 12 credits at 2K.

That one step prevents the usual mismatches: a head-and-shoulders portrait against a full-body dance, a character facing left against a performer facing the camera, a square image against a vertical clip. The Playbook's motion control workflow walks through it step by step, free with an account.

Pro costs 35% more per second. Test the same 3 seconds on both (60 and 81 credits) and pay for Pro only if your shot shows the difference.

Character Swap also offers Spicy Motion Control, built on the older Kling 2.6. It takes the same inputs and switches but costs less and lets you choose the resolution: 12 credits a second at 720p (Standard) or 20 at 1080p (Pro). Ten seconds cost 120 credits at 720p, against 200 on Kling 3.0 Standard.

Kling says 3.0 keeps faces far more consistent than 2.6 through complex motion, changing angles and longer clips, so 2.6 is enough when faces matter less: choreography drafts, or full-body shots where the face is small. Spicy Studio also runs the two differently: Kling 3.0 has the character follow the performer's orientation in the video, while 2.6 keeps the orientation of your image.

Driving videos: use footage you filmed, footage of a performer who agreed to it, or footage you've licensed. Don't re-run another creator's TikTok or Reel; film your own version of the trend.

Character images: use a character you created or a fictional person you generated. A real, identifiable person needs their documented consent, and no likeness may be used to deceive, impersonate or harass.

Publishing: label realistic AI video where the platform asks for it. Under Spicy Studio's Terms, you're responsible for the rights and consents behind everything you upload and post.

No footage, only an idea: describe the motion in the Video Lab with Kling 3.0 or Kling Omni O3.

Complex camera moves, or motion and sound from a reference clip: the Playbook's complex motion control workflow uses Seedance's reference video instead; it needs Spicy Pro or higher.

Specific spoken words: Lip Sync turns a face photo and your MP3 (up to 60 seconds) into a talking clip (lip sync guide).

A person placed into a still image: AI character replacement.

Picking a Video Lab model: the Seedance 2.0, Kling Omni O3 and Kling 3.0 comparison lays out inputs, length and credits.