Smash and Grab AI Video: See It, Break the Glass, Bolt
An object rests on a parked car's back seat. You approach the side window, glass explodes in a cinematic stunt effect, you snatch the object and sprint away. Upload a photo of yourself and one of the object. A fictional AI robbery, created with your own photo.
The smash and grab AI video is a fictional car-window robbery: a stationary camera positioned inside a parked car reveals an object on the rear seat, you approach, the window bursts like a film stunt, and you seize the object and flee. You provide a photo of yourself and one of the object, or your pet with its snacks.
Also called Smash and Grab effect · Car window heist
Example
What the Smash and Grab AI video looks like
This is the original reel that launched the trend. Your video is a fresh clip generated from your photo in the same style, not a duplicate of these frames. Every run produces unique results.
Original reel by @billionheirkc, 10 s
An early trend reel by @billionheirkc, featuring his own AI character: she notices a red handbag on the rear seat, breaks the window, snatches the bag and flees from two officers. Displayed as a vertical crop of his wide clip, with the bag's logo obscured. Your photo and your object replace them in a fresh scene.
How to create the Smash and Grab AI video on Smew AI
Everything runs in the browser. No editing app, no prompt to write, no watermark.
Upload the two photos
Press "Create your own" and upload yourself, full body if possible, then the object you snatch (a bag, a beverage, a box). Upload in this sequence, because sequence determines roles in the scene. Capture them with the camera or choose from your library.
Press Generate
The button displays the price. Seedance 2.5 creates the Smash and Grab scene directly from your photos, with no intermediate image, for 10 seconds with original sound. You write no prompt.
Download or share
The video is ready in a few minutes as a 9:16 MP4. Download it, share it from your phone, or retrieve it later in your Library.
Which photo works best?
Upload two photos: first yourself (or your pet, in the Pet version), then the object you snatch.
You: full body if possible, facing the camera, no sunglasses. You retain your face, clothing and shoes from the photo, and the escape shot displays you running.
One person per photo, sharp and well-lit.
The object: one item on a plain background, such as a bag, a beverage, a sneaker box or a camera. Items you can hold in one or two hands perform best.
Pet version: a clear photo of your pet's face and body, then a photo of its snacks or a toy.
Brand names and logos on the object appear plain or obscured intentionally.
Use your own photo, or one of an adult who consented. Never a celebrity, a stranger or a child.
Where did the Smash and Grab AI trend start?
Smash and Grab is a template format rather than a meme. Higgsfield introduced a "Smash and grab" AI effect in early September 2026: its example clips bear an 8 September date, and it accepts a character photo and a product photo. The structure remains constant. The camera sits inside a parked car with the product on the rear seat, the character breaks the side window and grabs it, and the clip transitions to an escape shot with police lights in the background. Higgsfield marketed it for "bold product reveals, playful action edits, and dynamic lifestyle ads".
AI creators experimented with it within days. On 11 September Los Angeles AI creator @billionheirkc shared a version featuring his own AI character and a red handbag, "created with Higgsfield's Smash and Grab template". A spec soda advertisement using "the Smash and Grab effect" appeared on 17 September. Additional AI video platforms introduced their own Smash and Grab templates: Kapwing on 8 September, Pollo AI by 16 September, imageat on 22 September and Media.io, whose trend page with copy-ready prompts launched on 6 October. It remains a niche format: on 14 October TikTok had no #smashandgrabai tag, and Google revealed no measurable US searches yet.
Higgsfield's "Smash and grab" effect displays its first example clips: a character photo and a product photo transform into a car-window break and an escape shot. Higgsfield
@billionheirkc shares a version featuring his own AI character and a red handbag, created with Higgsfield's template; it has 1,007 views and 24 likes by 14 October. TikTok
imageat introduces a Smash and Grab page: a 15-second clip from a photo of yourself and a photo of any object. imageat
Media.io releases Smash and Grab pages with copy-ready prompts for a handbag, a beverage, a sneaker box and a camera, plus pet and cartoon versions. Media.io
The format remains niche: TikTok has no #smashandgrabai or #smashandgrabtrend tag, and US Google searches for "smash and grab ai" are too scarce to measure over the past 90 days.
Copy and paste
The Smash and Grab AI prompts
You do not need this: the one-tap flow executes it for you. [Image 1] and [Image 2] are your photos, in the sequence you upload them: the video model uses them as a reference and creates the scene directly from them, with no photo step first.
Human
2 photos: you, full body if you can + the item you grab (a bag, a drink, a box)Seedance 2.5 Video Edit10 s · 720p · 9:16126 credits
Video prompt
Edit [Video 1]. Replace only the explicitly mapped source subjects in place:
[Image 1] replaces the person running toward the vehicle and taking the item; photo slot: You, full body if you can, in every shot where that subject appears.
[Image 2] replaces the item being taken from the car seat; photo slot: The item you grab (a bag, a drink, a box), in every shot where that subject appears.
[Image 1]: Copy the uploaded person's face, facial structure, eyes, skin tone, age, hair and facial hair or absence of facial hair. Remove source facial hair or hairstyles that the upload does not have. Keep identity consistent across every shot. Keep the source outfit unless the selected variant explicitly changes it or requires a body/species change.
[Image 2]: Use the uploaded object's actual shape, colors, materials and distinguishing details for the mapped object, following its source position, physical interactions and any requested transformation stages. Never put a human face on the object.
Selected variant: Human / main. Apply only the requested cast, wardrobe, setting or effect changes for this variant; preserve the source action and shot timing instead of inventing a new scene sequence.
Use the complete preview as the source for the Smash and Grab AI / Human variant. Preserve its actual action, effect progression, camera moves, shot sequence, physical interactions and mouth/jaw timing. Photo-role labels determine the intended identity assignment; never assign uploaded faces to unrelated background people or props.
Track each source subject through front, profile, back and occluded views. Keep all non-target subjects unchanged. Never duplicate an uploaded identity, merge subjects, swap their roles or transfer an identity to a prop or background subject.
Preserve the complete source video from its first frame to its last: source duration, frame order, every shot boundary, playback speed, camera movement, framing, lighting and original performance; keep background/props unless the selected variant explicitly changes them. Preserve each subject's original head, gaze, body, hand/paw gestures, expression, mouth and jaw opening/closing and their exact source timing. Do not create a new dance, scene, shot list, dialogue or lyrics. Do not add, remove, reorder, shorten, extend or retime shots or actions.
Produce silent video. The matching unmodified original soundtrack is restored separately after this edit. Do not generate speech, singing or music. Do not add new captions, subtitles, logos or a watermark.
2 photos: your pet, face and body clear + its treats or a toySeedance 2.5 Video Edit10 s · 720p · 9:16152 credits
Video prompt
Edit [Video 1]. Replace only the explicitly mapped source subjects in place:
[Image 1] replaces the person running toward the vehicle and taking the item; photo slot: Your pet, face and body clear, in every shot where that subject appears.
[Image 2] replaces the item being taken from the car seat; photo slot: Its treats or a toy, in every shot where that subject appears.
[Image 1]: Replace the ENTIRE mapped performer in EVERY source shot with the uploaded pet's actual species, face, muzzle, eyes, ears, fur, markings and coherent animal body. Replace human skin, hands, fingers, torso, sleeves and clothing belonging to this target with the pet's own fur and natural paws. Foreground hands belonging to this performer are included. Paws follow source gesture timing and limb trajectories without becoming five-fingered human hands. Never put a pet head on the source human body. Other separately mapped human roles stay human.
[Image 2]: Use the uploaded object's actual shape, colors, materials and distinguishing details for the mapped object, following its source position, physical interactions and any requested transformation stages. Never put a human face on the object.
Selected variant: Pet / pet. Apply only the requested cast, wardrobe, setting or effect changes for this variant; preserve the source action and shot timing instead of inventing a new scene sequence.
Use the complete preview as the source for the Smash and Grab AI / Pet variant. Preserve its actual action, effect progression, camera moves, shot sequence, physical interactions and mouth/jaw timing. Photo-role labels determine the intended identity assignment; never assign uploaded faces to unrelated background people or props.
Track each source subject through front, profile, back and occluded views. Keep all non-target subjects unchanged. Never duplicate an uploaded identity, merge subjects, swap their roles or transfer an identity to a prop or background subject.
Preserve the complete source video from its first frame to its last: source duration, frame order, every shot boundary, playback speed, camera movement, framing, lighting and original performance; keep background/props unless the selected variant explicitly changes them. Preserve each subject's original head, gaze, body, hand/paw gestures, expression, mouth and jaw opening/closing and their exact source timing. Do not create a new dance, scene, shot list, dialogue or lyrics. Do not add, remove, reorder, shorten, extend or retime shots or actions.
Produce silent video. The matching unmodified original soundtrack is restored separately after this edit. Do not generate speech, singing or music. Do not add new captions, subtitles, logos or a watermark.
You pay in credits. The Generate button shows this price before you spend anything. These are the options on this page.
Option
Model
What you get
Credits
Human
Seedance 2.5 Video Edit
10 s · 720p · 9:16
126
Pet
Seedance 2.5 Video Edit
10 s · 720p · 9:16
152
One-time packs and plans are on the pricing page. A failed video gives its credits back.
What happens in the smash and grab AI video, second by second?
Your clip follows the trend's structure, in two shots:
Time
What you see
0-2 s
A stationary camera inside a parked car. Your object rests on the rear seat, and you approach the closed side window.
2-4 s
You notice it and lean close to the glass, eyes wide.
4-6 s
The window explodes in one stylized cinematic stunt moment. Your hand extends through the opening, grabs the object and withdraws it.
6-10 s
Cut outside: a tracking shot of you sprinting down a warm street with the same object, police lights distant behind you.
The sound is generated with the clip: the glass, your footsteps, distant sirens and an original tense score, with no lyrics. The Pet version is a playful 12-second cartoon: your pet, as a small 3D animated version of itself, notices its snacks in a toy car, the window bursts into soap bubbles, and it scurries off with a wind-up toy patrol car slowly behind.
Why does it work? The opening frame reveals the prize before any action, so the break delivers as a payoff. That is also why brands adopt it: the object is the hero of the clip. Higgsfield promoted its effect for product reveals and lifestyle ads from the beginning.
How does Smew AI create it, and what does it cost?
Seedance 2.5 creates the scene directly from your two photos, with no AI image in between, so your face and your object remain yours from the opening frame to the last.
Human
Pet
Photos
You, then the object
Your pet, then its snacks or a toy
Look
Live-action film scene
Playful 3D cartoon
Length
10 s
12 s
Frame
Vertical 9:16, 720p
Vertical 9:16, 720p
Sound
Glass, footsteps, sirens, original score
Bubbles, chimes, paw steps, original tune
The exact price displays on the Generate button before you spend anything. You write no prompt. The prompts begin from Media.io's copy-ready Smash and Grab prompts ("Classic Handbag Getaway" and "Pet Treat Heist", published on 6 October 2026), preserved in their wording and sequence, with our photo instructions and rules that maintain your face, your outfit and your object consistent in every shot. Higgsfield's template does not reveal its prompt, and the creators' posts share none.
Is it a duplicate of a real reel? No. The trend has no single original reel: every version originates from a template. Your clip is a fresh take of the same scene, so the car, the street and the actions vary from run to run. A run requires a few minutes.
What goes wrong, and how do you fix it?
Problem
Fix
Your face looks less like you
Use a sharp photo with your face toward the camera and no sunglasses, then run it again.
The object changes size or shape
Use a photo of the object alone on a plain background, the entire object visible. Small objects that fit in one or two hands perform best.
The hand passes through the glass before it breaks
The prompt sequences the break first, then the reach. Run it again.
Two copies of the object, or it jumps into the hand
The prompt prohibits both. Run it again.
The logo on your object is missing
That is intentional: the prompt requests no brand names, logos or readable labels. If it is your own product, insert your logo in an editor.
Letters or a badge appear on an officer or the car
The prompt prohibits them, but AI can still generate garbled letters. Run it again, and never post a take that resembles a real police force.
The video model rejects a photo
Some photos get rejected. The credits return; attempt another photo.
Every run is a fresh take, so results vary.
How should you post it: the AI label, captions and hashtags
Identify it as AI and fiction. The clip resembles a film scene of a real crime. Activate TikTok's AI-generated label, include "AI" in your caption, and keep it an obvious joke about yourself or an advertisement for your own product. Never send it to someone as genuine, never create one of a person who did not consent, and never attempt it for real: breaking into a car is a crime.
Captions: "Some things are worth it", "It was on the back seat", "Saw it. Wanted it. Took it." Or, for a product: "So good, people break into cars for it".
Hashtags: #smashandgrab is an older tag that also contains real crime clips and memes, so include #aivideo and #aitrend. The clip includes its own sound; you can disable it on the result screen and select a sound when you post.
It is an AI video format from September 2026. A stationary camera inside a parked car reveals an object on the rear seat; a character approaches, the side window explodes like a film stunt, they seize the object and flee, and the clip transitions to an escape shot with police lights in the background. You upload a photo of yourself and a photo of the object, and AI creates the scene.
Where did the Smash and Grab trend come from?+
From Higgsfield's "Smash and grab" AI effect, with example clips dated 8 September 2026. It uses one character photo and one product photo. AI creators shared their takes starting 11 September, and between 8 September and 6 October, Kapwing, Pollo AI, imageat and Media.io launched their own Smash and Grab templates.
Is this the Higgsfield smash and grab?+
It is the same concept: the smash and grab AI trend originated as Higgsfield's "Smash and grab" effect. Smew AI is a different tool, so you need no smash and grab AI prompt, just your photo and the object to create the video.
Is the smash and grab video real?+
No. Everything in it is AI-generated: the car, the glass, the street and the officers. Only your face, clothing and object originate from your photos. Keep the AI label active when you post it, never represent it as genuine, and never attempt it for real.
Is the smash and grab meme big on TikTok?+
Not yet. It is more a template than a meme, and on 14 October 2026 TikTok had no #smashandgrabai tag. You need no smash and grab app to begin one: this smash and grab AI video generator creates it from two photos.
How do I make a smash and grab AI video?+
Select Human or Pet, press "Create your own" and upload two photos: first yourself (or your pet), then the object. Press Generate; the button displays the price first. You receive a vertical clip with sound in a few minutes, and it waits in your Library when ready. You write no prompt.
Can I use it as an ad for my own product?+
Yes, that is what the format was designed for: the stolen product is the hero of this AI heist video, much like an AI product reveal video. Use a clean photo of the product on a plain background. Brand names, logos and labels appear plain or obscured intentionally, so insert your logo in an editor afterward.
Can I make it with my dog or cat?+
Yes. The Pet version is a pet treat heist: AI transforms your pet into a small 3D cartoon version of itself that notices its snacks in a toy car. The window bursts into soap bubbles, it embraces the snacks and scurries off with a wind-up toy patrol car behind. Upload a clear photo of your pet, then a photo of its snacks or a toy.
Which prompt does this template use?+
Higgsfield's template does not reveal its prompt, and the creators shared none. Our prompts begin from Media.io's copy-ready Smash and Grab prompts ("Classic Handbag Getaway" and "Pet Treat Heist", 6 October 2026), word for word and in sequence, with our photo instructions and rules that maintain your face and your object consistent in every shot and keep logos absent.
How long is the video, and does it have sound?+
Human is 10 seconds and Pet is 12 seconds, both vertical 9:16 MP4s at 720p with no watermark. The sound is generated with the clip: glass, footsteps and sirens with an original score, or bubbles and a playful tune for Pet. You can disable it on the result screen.
Can I put a friend or a celebrity in it?+
A friend only if they consent, and only as an obvious joke. Never a celebrity, a stranger or a child: a clip of someone breaking into a car can embarrass them or be interpreted as genuine.
Does the smash and grab AI video cost credits?+
Yes. Each video costs credits, and the create page displays the exact price on the Generate button before you spend anything. If a video fails, the credits return.
