How to use Kling AI for ads: start frames, motion and sound

What is Kling, and how do creators use it for ads?

Kling is an AI video model from Kling AI. It turns text, or an image you upload, into video with sound. In the ad tutorials we reviewed, it is almost always the second step: another model makes a realistic still, and Kling animates it.

Facts from Kling's own pages, checked 12 Oct 2026:

• VIDEO 3.0 (user guide) makes 3 to 15 seconds at 1080p or 720p. It supports image-to-video and start-and-end-frame video.
• Elements keep a character consistent. You build one from 2 to 4 reference images or a short video, and it keeps its look and voice across camera moves.
• Multi-Shot lets the model plan cuts. Custom Multi-Shot lets you set each shot's length and content yourself.
• Text in the input image (signs, captions, logos) is preserved, which Kling notes helps e-commerce work.
• Kling 4.0 (Kling blog) began early access on 28 Sep 2026, with the full model due in October 2026. It adds native generations of 3 to 30 seconds, up to 10 keyframes, and up to 15 reference assets per generation.

Which tools do creators pair with Kling?

Nano Banana is the most common partner, because the start frame decides most of the result. Based on 13 public tutorials posted between 21 Jan 2026 and 8 Oct 2026, reviewed by Anglr.

How do you make an ad with Kling, step by step?

The tutorials agree on start frame first, short motion prompt second. They differ on sound: Tyler Reed uses Kling's native audio, while Will Connell adds a separate voiceover.

1. Make the start frame in an image model. Tyler Reed, @maxxmalist, Hamza Malik, Will Connell and @oggii_0 build the person, product and setting as a still first. Tyler Reed asks for natural light, a selfie angle and the product in hand, in a vertical frame.
2. Describe the movement, not the scene. Kling's image-to-video guide says the image already sets the scene, so the prompt should name the subject and how it moves. @maxxmalist uses a two-word motion instruction for each half of a before-and-after ad. Tyler Reed lists a few simple actions in order (look at camera, pick up the mug, sip, smile), and Will Connell asks for small natural movements.
3. Write one prompt per scene and join them. @maxxmalist prompts each scene from its own frame, and Hamza Malik joins the scenes in CapCut with speed changes. On Kling 3.0, Custom Multi-Shot can do part of this inside one generation.
4. Pick the model by format. Tyler Reed, George Stock and @jacobrodri_ all treat Kling as one option among several. George Stock picks Kling for hand and product motion, and @jacobrodri_ names it as a lower-cost choice for a quick reaction clip.
5. Decide where the voice comes from. Tyler Reed uses Kling's own audio in a single generation. Will Connell records the voiceover separately in ElevenLabs and lays it over the clip.

Two techniques come from single authors. Aswad Tahir builds a front, three-quarter and profile sheet of one face to keep a spokesperson consistent. @Strength04_X makes a hero product shot and an exploded version, then animates between them. Kling's start-and-end-frame mode is built for that second idea.

How should you write a Kling prompt?

Write it like shot direction in Kling's order: subject, action, setting, camera, lighting, pacing, sound. Use simple words and sentences, and leave duration and aspect ratio to the settings.

• With a start frame, keep to the subject and its movement. Kling warns that a description far from the image can cause a cut.
• Physics: describe movements that are physically plausible. Kling's own guide says complex physics, such as a bouncing ball, is still hard.
• Dialogue: write the speaker's name, the tone or accent in parentheses, then the line. If a character element already has a bound voice, don't describe the voice again.
• References: give each one a job (character, product, scene, style or motion). Kling 4.0's advice for product videos is to give the product reference a specific job, covering its shape, colour and logo.

Some creators put timestamps in the prompt; @OrhanGhazi65942 writes time-coded beats for a 30-second explainer. Kling's documented way to control shot lengths is Custom Multi-Shot in the settings, so try that first. Others, like @Hope_Ai01, write the whole film as one prose prompt.

Templates for seven ad formats are on Kling prompts, and the AI ad prompt builder writes a single shot in Kling's order. To estimate the cost of a batch of takes, use the AI video cost calculator and check current plans on Kling pricing.

Common problems and fixes

What creators report

These are the authors' own claims, not tested by Anglr.

• Tyler Reed, comparing models on the same start frame: "Kling 3.0 outperformed on product shots and anything requiring VFX or cinematic polish."
• Tyler Reed also reports a 10-second UGC clip with audio in about five minutes.
• @maxxmalist reports a full UGC ad in five minutes from a first frame and per-scene prompts.
• Hamza Malik reports animated ads built this way in under two hours each.

Kling makes the clip; the script and the edit make the ad. If your Kling ads already perform, Anglr pays AI creators on performance to remake winning scripts in every format: see what we look for. For the full workflow by ad style, read AI UGC ads or AI 3D and animated ads, or browse AI ad guides.