High Altitude Trampoline AI: The Extreme Jump Video, Starring You
You step off a giant dam, hit a tiny trampoline hundreds of metres below and bounce all the way back up to the top. One photo of you is all it takes.
Updated : Page published.
1Add your photo
Front-facing, well-lit selfies work best. One face per photo.
2Jump from
3Choose quality
By generating you confirm you have permission to use the people in your photos. Terms
No subscription needed · Ready in about 3 minutes · Failed renders are free
The high altitude trampoline AI trend is a 10-second clip in which a person jumps off a dam, cliff or skyscraper, lands on a tiny trampoline far below and bounces all the way back up to where they started; videosi makes it from one photo of you.
Page updates
- October 18, 2026 — Page published.
The High Altitude Trampoline AI prompt and model
High Altitude Trampoline AI is made with Seedance 2.0 / 2.5 (you pick the tier) by ByteDance. The prompt behind it has 5 parts, and the timing is what makes or breaks it.
- Model
- Seedance 2.0 / 2.5 (you pick the tier)
- Mode
- Photo + motion reference to video
- Length
- 10 seconds, 9:16 or 16:9
- Input
- 1 photo
- Identity lock Your photo is the only source for face, hair, build and outfit, held identical from the wide fall to the final close-up.
- Motion reference A silent 9-second dam jump supplies the leap, the top-down fall, the rebound and the close-up, and nothing else.
- Scene One drop (dam, sea cave, skyscraper atrium, canyon or helicopter) and one small blue-rimmed trampoline ringed by spectators far below.
- Timeline Five beats: the edge, the leap, the bull's-eye landing, the rebound with a backflip, back on top.
- Rules One jumper, one trampoline, real scale, a return to the same edge; no parachute, logos, captions or slow motion.
Photorealistic live-action footage that looks like a drone and phone stunt video, 10 seconds, natural daylight, one continuous sequence. Image 1 is the only source for the jumper…
Why it goes wrong when you do it yourself
- Without a motion reference the jumper usually falls into the water, lands beside the trampoline or never comes back up.
- Models shrink the drop to a few metres or make the trampoline huge, and the impossible scale that makes it funny is gone.
- The face is tiny for most of the fall, so the likeness drifts unless the identity lock comes first and the close-up is pinned at the end.
- Each try costs model credits and several minutes, and the landing and the rebound usually need several tries to line up.
High Altitude Trampoline AI examples
How to make your High Altitude Trampoline AI video
Upload one photo of yourself
Face to the camera, ideally head to knees, in even light. Your face, hair, build and clothes go straight onto the jumper.
Pick where you jump from
A giant dam, a sea cave cliff, a skyscraper atrium, a desert canyon or the door of a helicopter. Then choose 9:16 for TikTok, Reels and Shorts, or 16:9.
Generate and download
Press Generate and the 10-second jump and rebound is rendered for you as a silent MP4 with no watermark.
Post it with a sound
Add a trending sound in the app you post from and turn on the AI-generated label, because it looks real.
What is the High Altitude Trampoline AI trend?
Updated · videosi is an independent tool. It is not affiliated with TikTok, Instagram, YouTube, or with the owners of any trend, song, game or brand named on its pages.
The high altitude trampoline AI trend is a ten-second stunt that nobody could survive for real. Someone stands at the railing of a giant dam, a sea cliff or the top floor of a skyscraper atrium, with a crowd holding up phones beside them. They step over and drop. The camera looks straight down at a trampoline so small it is barely a dot, they hit the exact centre, and the mat throws them all the way back up to the railing they jumped from, where the clip ends on their out-of-breath grin.
On TikTok it goes by several names: high altitude trampoline, high-altitude trampoline and extreme jump AI effect, AI high altitude stunt, or simply trampoline AI. Clips captioned "High-altitude trampoline" were already around in January 2026, but the one-photo version took off in September 2026, when template accounts began posting it on 3 September and a wave of how-to videos followed from late September into October. It is still at an early stage; no original creator or song is confirmed.
Every one of these clips is AI-generated. They are made by a video model from a single photo of the person, a short reference clip that supplies the fall, the bounce and the camera angles, and a written prompt that describes the place.
This page does that whole recipe for you. The jump, the camera moves and the prompt are built in; you choose the photo, where you jump from (a dam, a sea cave, a skyscraper atrium, a desert canyon or a helicopter) and the shape of the video.
High Altitude Trampoline AI: shot by shot
- 0–1.5 s
At the edge
You stand at the railing above a huge drop, a crowd with phones beside you, look down once and step over.
- 1.5–3 s
The leap
You jump out into the air and the camera follows from above as you fall, arms out, getting smaller.
- 3–5 s
Bull's-eye
Top-down shot: a tiny trampoline far below grows until you land dead centre and the mat stretches deep.
- 5–7.5 s
The rebound
The trampoline fires you straight back up the whole height, with one tucked backflip on the way.
- 7.5–10 s
Back on top
You land on your feet where you started, grab the railing, and the camera closes in on your grin.
Tips for a better result
Show your whole body
You are seen head to toe for most of the clip. A full-length or head-to-knees photo gives the model far less to invent.
Sporty clothes read best
Leggings, a sports top or a tracksuit look right on a stunt jumper and keep their shape in the fall. Long coats and dresses flap and change.
Leave the sunglasses off
The close-up at the end is the payoff; covered eyes make the likeness drift right at that moment.
Tie long hair back
Hair is the part that moves most in the fall and the bounce. A ponytail in your photo stays more consistent than loose long hair.
Good to know before you make a High Altitude Trampoline AI video
- The five locations are fixed; you cannot upload your own place or move the trampoline.
- Your face is small during the fall and only large in the last two seconds, so likeness is judged mostly on the close-up.
- Hands on the railing and the backflip are where invented detail shows first.
- The video is silent, and likeness is not guaranteed. A failed, timed-out or blocked render returns its credits automatically.
High Altitude Trampoline AI: questions
What is the high altitude trampoline AI trend?
The high altitude trampoline AI trend is a 10-second clip in which a person jumps off a dam, cliff or skyscraper, lands on a tiny trampoline far below and bounces all the way back up; videosi makes it from one photo. The one-photo version spread on TikTok from September 2026 under names like extreme jump AI effect and AI high altitude stunt. No original creator is confirmed.
How do I make the high altitude trampoline video from one photo?
Upload one photo of yourself to videosi, pick where you jump from (5 options: dam, sea cave, skyscraper atrium, canyon or helicopter), choose 9:16 or 16:9 and press Generate. The high altitude trampoline AI video is a 10-second silent MP4. Rendering takes several minutes, and you can close the page because the result is saved in My creations. You write no prompt.
Is there a website or generator for the high altitude trampoline AI effect?
Yes. videosi (videosi.io) is a web tool that makes the high altitude trampoline AI video from 1 photo. The jump, the camera angles and the prompt are built in, so you choose only the photo, the location and the format. A standard 10-second video (Standard model, 720p) is 9 credits, with no watermark.
Is there a high altitude trampoline CapCut template or TikTok filter?
Most TikTok tutorials for the high altitude trampoline effect point to a CapCut or Dreamina template, and those templates change often, so check inside the app. videosi is independent and not affiliated with TikTok, CapCut or Dreamina. The videosi high altitude trampoline AI video is a web tool: you upload 1 photo and download a 10-second silent MP4.
What AI model or prompt makes the high altitude trampoline video?
The high altitude trampoline AI video on videosi is made with ByteDance's Seedance video model (Seedance 2.0 Mini by default; Fast, Pro and Ultra, which is Seedance 2.5, are one click away), from a five-part prompt plus a 9-second motion reference of a dam jump. By hand the usual failures are a jumper who lands in the water or never bounces back. On videosi the prompt and reference are built in.
How much does the high altitude trampoline AI video cost, and is it free?
The high altitude trampoline AI video on videosi is not free: there is no free generation and no free trial. A standard 10-second video (Standard model, 720p) is 9 credits; better models and higher resolutions cost more. Credit packs start at $6.99 for 10 credits and never expire, and plans start at $29.90 a month. A failed render returns its credits automatically.
Why does my trampoline jump video not look like me?
In the high altitude trampoline AI video your face is small for about 7 of the 10 seconds, so a weak photo shows quickly. Use a front-facing photo that shows your body as well as your face, in even light, with no sunglasses or hat, and a ponytail if your hair is long. Likeness is not guaranteed. A failed, timed-out or blocked render on videosi returns its credits automatically.
Can I use a photo of my friend, my mum or my partner?
Yes, if they are an adult and have agreed to it. The videosi high altitude trampoline AI video works from 1 photo of any consenting adult, and watching a friend bounce back up a dam is most of the fun. videosi does not accept photos of celebrities, public figures or minors, and users must be 18 or older. Photos and results are deleted within 30 days.
Where can I jump from in the high altitude trampoline AI video?
The videosi high altitude trampoline AI video has 5 starting points: a giant concrete dam, the rim of a sea cave, the top floor of a skyscraper atrium, a skywalk over a desert canyon, or the open door of a helicopter. The trampoline far below and the rebound stay the same in all of them, and your face, hair and outfit always come from your photo.
Is the high altitude trampoline video real, and is it safe to try?
No, it is not real and must never be tried. Every high altitude trampoline clip, including the 10-second videosi version, is fully AI-generated: no trampoline can catch a fall of that height. TikTok, Instagram and YouTube ask creators to label realistic AI content, so turn on the AI-generated label when you post to keep it a joke and lower the risk of a takedown.
Can I make the high altitude trampoline video with ChatGPT, Sora, Veo or Kling?
You can try, but without a motion reference general video models rarely keep the real scale, the bull's-eye landing and the full rebound in one 10-second shot. videosi sends Seedance a silent reference clip of the trend together with your photo, so the high altitude trampoline AI video follows the same 5 beats every time, and you write no prompt.
How long is the high altitude trampoline AI video, and does it have sound?
The videosi high altitude trampoline AI video is 10 seconds long, in 9:16 or 16:9 at the resolution you pick (720p on the Standard model), saved as an MP4 with no watermark and no sound. Add a trending sound in the app where you post. Photos upload as JPG, PNG, WEBP or HEIC up to 10 MB.
Trending this week
All trends this weekMore effects like this
All viral effectsMake your High Altitude Trampoline AI video
Upload a photo and press Generate. 9 credits, refunded if it fails.
Get the next trend first
One email when new trends land on videosi, about once a week. Unsubscribe in one click.