The best prompts for Kling AI product videos follow a five-part structure: subject description, surface material, lighting setup, camera motion, and environment context. Kling 3.0 responds well to physically grounded descriptions and breaks down when prompts stack too many abstract concepts or emotional descriptors. After producing hundreds of Kling product shots for DTC ad campaigns, these are the prompt patterns that consistently render clean, usable footage.
What prompt structure works best for Kling 3.0 product shots?
Kling 3.0 parses prompts better when you front-load the subject and its physical properties before describing motion or environment. The model handles material descriptions with surprising accuracy, so specifying "matte glass bottle with condensation droplets" outperforms "beautiful premium bottle" every time.
Here is the five-part formula we use in production:
Subject and material — Start with the product and its physical surface. "A 200ml amber glass serum bottle with a brushed gold dropper cap" gives Kling enough geometry to render accurately. Avoid brand names in the prompt since the model ignores them and they waste token space.
Surface interaction — Describe what the product touches or sits on. "Resting on a wet marble slab with shallow water pooling around the base" gives the model a contact surface to ground the physics. Without this, products tend to float or cast incorrect shadows.
Lighting direction — Kling 3.0 responds to directional lighting cues. "Single key light from upper left, warm 3200K tone, soft shadow falling right" produces dramatically more consistent results than "cinematic lighting" or "studio lit." Specifying color temperature and shadow direction gives the model two constraints that prevent the flat, overlit look common in AI product videos.
Camera motion — Keep it to one move per generation. "Slow dolly-in from medium shot to close-up over 5 seconds" works. "Orbit while pushing in and tilting up" fails because Kling tries to interpolate all three and creates warped geometry around frame 60-80. For product videos, the most reliable motions are dolly-in, slow orbit (under 90 degrees), and static with a rack focus shift.
Environment and mood — Place this last. "Dark background with bokeh highlights suggesting a bathroom vanity" tells the model the setting without overriding the product description. Keep it to one sentence.
Example prompts that render clean product footage
These are actual prompts we have used in production, edited to remove client-specific details:
Skincare bottle hero shot: "A frosted white glass moisturizer jar with a silver twist cap, sitting on a limestone shelf. A single soft light source from upper right creates a gentle gradient across the jar surface. Slow push-in from waist-level angle to close-up. Minimal depth of field, neutral beige background."
Supplement pouch with powder: "A matte black stand-up pouch on a wooden cutting board surrounded by scattered turmeric powder and fresh ginger root. Natural window light from camera left, warm afternoon tone. Static camera, 35mm focal length, shallow depth of field with the background softly blurred."
Beverage can with motion: "An aluminum can with water droplets on its surface, placed on a dark reflective surface. A splash of orange liquid rises behind the can in slow motion. Backlit with a cool white rim light. Slow orbit from front to three-quarter view. Clean dark studio background."
What to avoid in Kling product video prompts
A few patterns consistently produce bad output:
- Stacking adjectives without physics — "Luxurious premium elegant bottle" tells the model nothing about geometry or material. Every adjective should describe something the model can render: reflectivity, texture, shape, color.
- Multiple camera moves — Compound motions (orbit plus zoom plus tilt) cause warping, especially on reflective or transparent surfaces. Pick one.
- Specifying exact text on packaging — Kling 3.0 still struggles with legible text. If your workflow requires readable labels, generate the video without text and composite it in post, or use image-to-video with a product photo that already has the correct label.
- Prompting duration expectations — Kling 3.0 in Master mode generates up to 10-second clips. Requesting "30 second cinematic ad" does nothing productive. Prompt for a single shot and edit your timeline from multiple generations.
Which Kling 3.0 mode should you use for product videos?
For product ads, use Master mode when the shot requires accurate material rendering on reflective or translucent surfaces. Pro mode works for matte, opaque products where surface detail is less critical and you need faster turnaround. Standard mode produces noticeably softer textures and is best reserved for early concept tests, not final renders.
When using image-to-video, upload a product photo shot on a clean background with even lighting. Kling 3.0's image-to-video mode preserves the source geometry better when the input image has high contrast between the product and background.
