You drop into a boxing stance and throw a jab, jab, slip, roll and uppercut at nobody at all. One photo of you, or of your grandma, is all it takes.
Updated : Page published.



Front-facing, well-lit selfies work best. One face per photo.
By generating you confirm you have permission to use the people in your photos. Terms
The shadow boxing AI trend, also written shadowboxing AI, is an 8-second clip in which the person from an ordinary photo suddenly drops into a boxing stance and throws a jab, jab, slip, roll and uppercut combination at thin air; videosi makes it from one photo of you.
Shadow Boxing 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.
Photorealistic live-action phone footage, 8 seconds, one continuous static shot, the whole body visible from head to shoes. Image 1 is the only source for the person…



Head to shoes in frame, standing, facing the camera in even light. The outfit in the photo is the outfit you box in.
Choose the bright living room, which is how most remakes look, or a moody riverside promenade at dusk, then 9:16 for TikTok, Reels and Shorts, 1:1 or 16:9.
Press Generate and the 8-second combo is rendered for you, a silent MP4 with no watermark.
Add a sound in the app you post from, add a caption like the trend's "Alright grandma relax", and label the clip as AI-generated.
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 shadow boxing AI trend (often written shadowboxing AI) takes one ordinary photo of someone, very often a grandparent standing in the living room, and turns it into a clip of that person dropping into a boxing stance and firing off a combination at the air: jabs, a slip of the head, a duck under an imaginary punch and an uppercut, all with a straight face. The joke is the contrast. Captions such as "Alright grandma relax" or "I told her I'm not hungry" sit on top of a grandmother boxing like a pro.
It spreads as a CapCut template called "Shadowboxing filter", an 11-second template that showed more than 226,000 uses on its CapCut page on 23 October 2026. The earliest remakes we found on TikTok were posted on 12 and 14 September 2026, almost all from template-sharing accounts that tag them "pioneer shadowboxing capcut template". Some captions tie the combo to a famous fighter's old training clip; we have not confirmed that link, and this page uses no footage of any real fighter. Some people search for the trend by the combo itself, "jab jab slip roll uppercut" or "jab slip uppercut".
It is not the same thing as the older "shadow boxing" game on TikTok, also called the Ish game, where two people point in four directions and try to make the other look the same way. That game needs two people and no AI.
This page does the whole recipe for you. The combo, the camera and the prompt are built in; you choose the photo, the scene and the format.
A light bounce on the toes, a quarter turn side-on and the fists come up to the chin.
A quick step in and a deep duck under a punch nobody threw, rolling back up on the other side.
Back in stance, hands high, eyes locked on an invisible opponent.
Two snapping lead jabs and a hard rear straight, each hand straight back to the chin.
The head slips to one side, rolls under, and the rear hand comes up in an uppercut.
One last jab, a step back and the high guard held to the end, ready to loop.
The combo uses the feet, knees and both arms. A photo from head to shoes gives the model your real legs and shoes; a face crop makes it invent them.
A standing photo with arms relaxed at the sides converts best. Seated or leaning photos often come out with odd hips and legs once the boxing starts.
No sunglasses, masks or wide hats. The head turns side-on and ducks twice, and a clear face is what lets the likeness come back afterwards.
Solid tops and trousers keep the punches readable on a phone. Very loose sleeves or long dresses can hide the arms and legs.
The shadow boxing AI trend, also called shadowboxing AI, is an 8-second clip, made on videosi from one photo, in which the person in an everyday picture, often a grandparent, drops into a boxing stance and throws a jab, slip, roll and uppercut combo at the air. It spreads as a CapCut template called "Shadowboxing filter"; the earliest remakes we found on TikTok date from 12 September 2026.
Upload one full-body photo to videosi, pick one of 2 scenes and 9:16, 1:1 or 16:9, and press Generate. The shadowboxing AI video is an 8-second silent MP4 and you write no prompt: the combo, the camera and the timing are built in. Rendering takes several minutes, and the result is saved in My creations if you close the page.
Yes. videosi (videosi.io) is a web tool that makes the shadow boxing AI video, also called the AI shadow boxing filter, from one photo. The motion, the scene and the prompt are built in, so you choose only the photo, the scene and the format. A standard 8-second video (Standard model, 720p) is 7 credits, with no watermark.
Many shadowboxing AI clips were made with a CapCut template titled "Shadowboxing filter", 11 seconds long, and templates like it change or disappear often, so check inside CapCut. videosi is independent and not affiliated with CapCut or TikTok. The videosi shadowboxing AI page is a web tool: upload a photo, download a silent 8-second MP4, and post it anywhere.
No. The shadow boxing trend most TikTok explainers talk about is a two-person pointing game, also called the Ish game, that went viral in late 2019. The shadowboxing AI trend is different: one photo becomes an 8-second clip of that person boxing the air. videosi makes the AI version only; the pointing game needs two real people.
There is no single confirmed song for the shadowboxing AI trend: the remakes we watched use different sounds, many from template-sharing accounts, and we have not verified one original audio. The videosi shadowboxing AI video is a silent 8-second MP4, so pick any sound from the TikTok or Instagram library when you post. The track "Shadowboxing" by other artists is unrelated.
Upload a photo and press Generate. 7 credits, refunded if it fails.
One email when new trends land on videosi, about once a week. Unsubscribe in one click.
Add videosi as a preferred source and our trend pages show up first in your Top Stories and AI answers on Google.
No subscription needed · Ready in about 3 minutes · Failed renders are free
Shadowboxing means throwing punches at the air with no opponent, a real boxing drill. In the shadowboxing AI trend on TikTok it means turning someone who never boxes, often a grandma or grandad, into a fighter for 8 seconds. videosi makes that clip from one photo, so nobody has to throw a real punch.
Dog and cat shadow boxing AI clips are part of the trend, but the videosi shadowboxing AI page is built and tuned for photos of people, with the motion of a human boxer on 2 legs. You can upload a pet photo, but a usable result is not guaranteed; a standing person photographed head to shoes gives the best 8-second combo.
A jab slip uppercut is 3 moves: a straight punch with the lead hand, a small dip of the head to the side to dodge a punch, then a rising punch from the rear hand, with the hands back at the chin after each one. The videosi shadowboxing AI video shows that combo from your photo; for real training, learn it from a boxing coach.
The videosi shadowboxing AI video is made with ByteDance's Seedance model (Seedance 2.0 Mini by default; Fast, Pro and Seedance 2.5 Ultra are one click away) from a five-part prompt plus a motion reference clip of a boxer. Doing it yourself usually takes several tries, because punches come out slow and floaty without the reference.
The shadowboxing AI video on videosi is not free: there is no free generation and no free trial. A standard 8-second video (Standard model, 720p) is 7 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.
The usual cause is the photo: a face crop instead of a full-body shot, sunglasses or a hat, or a seated pose. For the videosi shadowboxing AI video, use one standing photo from head to shoes in even light. Likeness is not guaranteed, and the 2 moments when the head turns side-on or ducks are where it drifts most.
Only an adult who has agreed to it. Grandparents are the most popular stars of the shadowboxing AI trend, so ask them first and show them the result. 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.
The videosi shadowboxing AI video is 8 seconds long, in 9:16, 1:1 or 16:9 at the resolution you pick (720p on the Standard model), as an MP4 with no watermark and no sound. Add a sound from TikTok's or Instagram's library when you post. Photos upload as JPG, PNG, WEBP or HEIC up to 10 MB.