Kling 3.0 Fashion & Footwear Video Prompts (Tested)

Add to home screen

Open from your home screen next time

  1. 1Open the browser menu ⋮
  2. 2Choose “Install app” or “Add to Home screen”
  3. 3Tap “Add” or “Install” to confirm

Menu labels vary by browser. If the option is missing, open this page in Safari or Chrome.
Blurbly
0
Pricing

Kling 3.0 · Video prompts

Kling 3.0 prompts for fashion and footwear video

If you sell clothes or shoes, you already have the photo: a lookbook frame, a packshot, a sneaker on a plinth. What you want from Kling 3.0 is that photo moving, with the garment keeping its colour, cut and trims. We tested four prompts for that job: a linen dress on a walk, a sneaker on a turntable, a wool coat caught by the wind and two hands lacing a boot.

Every clip is 5 seconds from Kling V3.0 Pro with cfg_scale 0.5 and sound off, shown next to the still it started from so you can check the hem, the laces and the sole yourself. The start frames are invented products on AI-generated adult models, made with Seedream 4.5. The sneaker took three attempts and the coat two, and both went wrong the same way: Kling had to show something the photo never showed, the inner side of the shoe and the front of the man’s face.

Tested 2026-10-19 via the Novita AI API (kling-v3.0-pro-i2v), the provider Blurbly uses for Kling. The products and brand are invented and the models are AI-generated adults. Clips are as Kling returned them, compressed for the web.

Kling 3.0 clip: the model walks toward a backward-tracking camera, linen skirt swinging, belt ends moving
Output of prompt 01Walking shot in a linen dress

Pick any prompt below and press Use this prompt; it lands in this box. Swap in your own product details and photo before you generate.

Prompt formula

Every prompt on this page follows the same order. Fill each slot for your product and keep the order.

SlotExampleWhy it matters
Continue the actionShe keeps walking toward the camera with a relaxed, natural stride.In image-to-video the photo already describes the subject. Spend the first sentence on what moves next. Re-describing the outfit gives Kling permission to redraw it.
Material in motionThe linen skirt swings and folds with each step, the tie belt ends sway.Fabric behaviour is what a still photo cannot show. Name the part that moves and the verb that fits the cloth: linen swings and creases, wool billows, leather creases at the flex point.
One camera moveThe camera tracks backward at her walking pace on a smooth gimbal.Tie the move to the subject’s speed so the garment stays the same size in frame. Two moves in five seconds (push-in plus orbit) usually blur the product.
Light that matches the photoGolden late-afternoon light. / Blue-hour light with a warm rim light from the left.Repeat the light already in the start frame. Asking for different light makes Kling relight the scene in the first second, which reads as a flicker.
Lock listHer face, the sage green dress, the belt and the white sandals stay identical to the first frame.List every attribute a customer would return the item over: colour name, trims, hardware, sole. Kling held every item we listed when the camera stayed on views the photo shows.
Negative promptface change, different person, extra limbs, deformed hands, color shift, text, logo, flickerKling 3.0 accepts negative_prompt. Use it for failure modes such as extra fingers. It did not override a movement Kling had already settled on (see the sneaker test).

4 tested Kling 3.0 video prompts

Each prompt is the exact text we sent to Kling 3.0. The video beside it is the unedited output, with the settings we used.

01

Walking shot in a linen dress

TikTok / Reels 9:16 opener · dress PDP video slot · Pinterest idea pin

9:16 (from start frame) · 1080×1920 · 5 s · image-to-video · Kling V3.0 Pro · cfg_scale 0.5 · sound off

Start frame: adult model in a sage green linen shirt-dress with a tie belt walking on a sunlit cobblestone street
Input
Output
Prompt 
She keeps walking toward the camera with a relaxed, natural stride. The linen skirt swings and folds with each step, the tie belt ends sway, her hair moves slightly. The camera tracks backward at her walking pace on a smooth gimbal. Golden late-afternoon light. Her face, the sage green dress, the belt and the white sandals stay identical to the first frame.
Negative prompt 
face change, different person, extra limbs, deformed hands, deformed feet, color shift, text, logo, flicker

Why it works

  • “Keeps walking” continues the pose already in the photo, so the first frame does not jump into a new action.
  • Naming the skirt, the belt ends and the hair gives Kling three separate secondary motions. That is what makes the fabric read as linen rather than a stiff print.
  • “Tracks backward at her walking pace” keeps her the same size in frame, so the dress stays readable for the full five seconds instead of filling the frame at the end.

Swap for your product

  • Fabric verbs: satin “slides and catches the light”, denim “barely moves”, a pleated skirt “fans open with each step”.
  • Lock list: replace “sage green dress, the belt and the white sandals” with your colour name and every trim (buttons, buckle, contrast stitching).
  • Location and light must match your start frame. For a studio photo, write “plain grey paper backdrop, soft key light” in place of the street.

Attempts: 1. This is the first generation.

02

Sneaker quarter-turn on a turntable

Shopify / PDP looping video · Amazon product video · 1:1 feed ad

1:1 (from start frame) · 1440×1440 · 5 s · image-to-video · Kling V3.0 Pro · cfg_scale 0.5 · sound off

Start frame: plain white leather low-top sneaker with a gum sole and tan suede heel tab on a grey turntable
Input
Output
Prompt 
The turntable rotates slowly and smoothly about a quarter turn (90 degrees) so the toe of the sneaker swings toward the camera, then the turntable stops and the shoe holds still. Only the outer side and the toe are visible during the move. The camera is locked off; the studio lighting and soft shadow stay the same. The sneaker keeps exactly the same shape, plain white leather, white laces, gum sole and tan suede heel tab throughout; no logos, no perforations.
Negative prompt 
perforations, holes, logo, text, brand mark, morphing shoe, shape change, extra laces, warped sole, camera movement, flicker

Why it works

  • “About a quarter turn (90 degrees)” is a move Kling can finish in five seconds, so the clip ends on a deliberate toe-forward hold instead of stopping mid-spin.
  • “Only the outer side and the toe are visible” keeps the shoe on faces the start frame actually shows. Our two earlier attempts turned further and invented the hidden side.
  • The lock list names the parts that drift most in rotation: sole colour, heel tab and laces.

Swap for your product

  • Turn angle: about 90° for a 5 s clip. Ask for more only if your start frame shows the extra sides, or animate a second photo of the other side.
  • Materials: “plain white leather, gum sole, tan suede heel tab” → your upper, sole and any contrast panel.
  • Direction: “the toe swings toward the camera” suits sneakers; for heels or boots where the back sells, write “the heel swings toward the camera”.
  • Keep the plinth plain. A patterned turntable makes small speed changes visible.

Attempts: 3, failed tries included. This is the one we kept.

03

Wool overcoat in a gust of wind

Outerwear campaign Reels · YouTube Shorts · paid social 9:16

9:16 (from start frame) · 1080×1920 · 5 s · image-to-video · Kling V3.0 Pro · cfg_scale 0.5 · sound off

Start frame: adult man in an open camel wool overcoat on a city rooftop at blue hour, face in three-quarter profile
Input
Output
Prompt 
A strong gust of wind comes from the left: the open camel coat billows and flaps behind him, the hem swings back, then settles as the gust passes. He keeps his head turned to the left, looking off into the distance, and pulls the left lapel closed with his right hand. Slow cinematic push-in. Blue-hour light with a warm rim light from the left. His face, the camel color and the cut of the coat stay exactly the same as the first frame.
Negative prompt 
face change, different person, extra limbs, deformed hands, deformed feet, color shift, text, logo, flicker

Why it works

  • The wind has a direction and a life cycle (it comes from the left, billows, then settles), so the coat moves in one believable arc instead of flapping at random for five seconds.
  • “Keeps his head turned to the left” holds the face in the angle the photo shows. In our first attempt he turned to camera and his face changed.
  • Giving the hand a job (pull the lapel closed) shows how the coat hangs when it is worn closed, which a buyer wants to see.

Swap for your product

  • Wind strength by fabric: “a strong gust” for wool and trench coats, “a light breeze” for silk and chiffon, which otherwise flies out of frame.
  • The closing gesture: “pulls the zip up to the chin” for a parka, “ties the belt” for a trench.
  • Colour name: “camel” → your exact shade name (“oatmeal”, “charcoal melange”). The lock list only protects what it names, and “brown” covers too many shades.

Attempts: 2, failed tries included. This is the one we kept.

04

Hands lacing a leather boot

PDP detail clip · Reels / TikTok product close-up · Pinterest video pin

9:16 (from start frame) · 1080×1920 · 5 s · image-to-video · Kling V3.0 Pro · cfg_scale 0.5 · sound off

Start frame: two adult hands holding the laces of a brown leather lace-up ankle boot on a wooden step
Input
Output
Prompt 
Both hands pull the laces tight, cross them and tie a neat bow at the top of the boot, then give the bow a short final tug. The leather creases naturally at the ankle. Static macro camera, shallow depth of field, warm morning light. Exactly two hands with five fingers each; the boot, brass eyelets and brown laces stay the same as the first frame.
Negative prompt 
extra fingers, fused fingers, extra hands, deformed hands, laces passing through leather, morphing boot, text, logo, flicker

Why it works

  • The action is split into four beats: pull tight, cross, tie, tug. That gives the hands something specific to do in every second instead of fidgeting.
  • “Exactly two hands with five fingers each” plus the negative prompt kept the fingers intact through the bow, and the laces stayed threaded through the same eyelets.
  • The start frame already shows the fingers holding the laces, so Kling never has to invent the grip, which is where hand clips usually break.

Swap for your product

  • Change the action to your closure: “pull the zip up to the collar”, “fasten the buckle and tuck the strap”, “press the hook-and-loop strap down”.
  • Hardware: “brass eyelets and brown laces” → your eyelets, hooks or speed-lace loops.
  • Keep the foot on a step or bench in the start frame. Standing poses put the hands at the edge of the frame and they get cropped mid-action.

Attempts: 1. This is the first generation.

What went wrong in testing

We asked for a full 360° turntable spin in 5 seconds and got roughly half a turn.
Kling kept the turntable at a believable speed instead of spinning fast. Ask for a rotation that fits the length: about 90° in 5 seconds finished cleanly and ended on a hold.
The hidden side of the sneaker was invented: three rows of perforations appeared on the inner side, a detail our shoe does not have and one that resembles a well-known trainer.
Rotate only onto sides your start frame shows. If you need the other side, animate a separate photo of that side and cut the two clips together.
Adding “no perforations” to the prompt and “perforations, holes” to the negative prompt changed nothing. The second attempt was almost frame-for-frame the same as the first.
When Kling repeats a motion from the same start frame, negatives will not steer it. Change the requested action (we changed the turn angle) or change the start frame.
In the first coat attempt the model turned to face the camera, and the face Kling invented for the front view looked like a different, older man.
The start frame only showed his face in three-quarter profile. We told him to keep looking off to the left; the second attempt kept the same face throughout.
There is no aspect-ratio setting in Kling image-to-video. A square start frame came back 1440×1440 and a 9:16 frame came back 1080×1920.
Crop the start frame to the placement first: 9:16 for Reels and TikTok, 1:1 for feed and product pages. Cropping the finished clip instead tends to cut off the hem or the shoe tip.

When to use another model

Seedance 2.5

When you need one continuous take longer than Kling’s 15-second maximum, or want to combine a garment photo, a model photo and a location in reference-to-video mode.

Wan 3.0

For longer, lower-cost try-on or haul clips where a presenter talks and moves for 15–30 seconds; we tested it on electronics, not apparel.

Seedance 2.5 UGC ad prompts

If the clip needs a person talking about the product with their own voice rather than a silent fashion shot.

Do it in Blurbly without writing prompts

  1. Add one garment photo with + Product (a lookbook frame, a flat lay or a packshot) and keep Video selected.

  2. In the text box, say what should move and what must not change, for example “9:16 walking shot for TikTok, keep the dress colour exact”.

  3. Set length and ratio on the Video specs card and confirm. The plan that follows has a preview frame per shot and a Shot list with each action and camera move; check the trims and the sole in every frame.

  4. Type any fix in the box under the plan, then tap Unlock to render the video.

Get a video like this from one product photo

Blurbly plans the shots, writes the prompts and picks the model. You see the plan before you pay.

Make a fashion video from your product photo

Frequently asked questions

Can Kling 3.0 keep my exact garment from a product photo?

In our image-to-video tests the colour, trims and silhouette held for the full 5 seconds whenever the camera and the subject stayed on views the photo shows. Problems started when the clip revealed something new: the inner side of a sneaker, or a face turning toward the camera. Plan the motion around what your photo shows.

What settings work best for Kling 3.0 clothing videos?

Image-to-video from a clean, well-lit photo already cropped to the placement, 5 seconds, one action and one camera move. We used the Pro tier with cfg_scale 0.5 and sound off. We did not compare the Standard tier on this page; the same prompts run on it unchanged.

How long does Kling 3.0 take to make a 5-second clip?

Our 5-second Pro clips came back in about 2 to 5 minutes each through the API (113 to 296 seconds across seven runs). Queue times vary with load.

Can Kling 3.0 do a full 360° product spin?

Not reliably in 5 seconds from one photo. It rotated about half a turn and invented the side the photo did not show. For a full spin, shoot or generate each side and animate them separately, or use a longer clip with a start frame that shows more of the product.

Can I use AI-generated models in Kling videos for my store?

Our models are AI-generated adults made for this test. If you animate photos of real people, you need their permission. For commercial-use terms of Kling output, check the Kling 3.0 model page and the provider’s current terms, and follow your ad platform’s rules on AI-generated content.

Sources