Trouble Maker Dance
15s · 2 photos
TWO-PHOTO HALLWAY ROUTINE
JISOO Addicted Dance AI: Two Photos, One Duet
The hallway clip with Rosé, cut to the fifteen seconds this model can take. It hard-cuts between the two of them: the first dancer opens, the second takes the middle, the first closes. One photo per dancer, and the cut does the rest.
2 photos · 15s · 9:16 · 720p · silent · no watermark
Not sure which photo goes where? →The two dancers
One photo each. The left seat opens and closes; the right one takes the middle.
Opens and closes
Shot 1 and shot 3 · 0.0–8.6s, 13.0–15.0s
Required
Takes the middle
Shot 2 · 8.6–13.0s
Required
No second person? Upload the same photo twice — it reads as two of you.
JPG, PNG or WebP · 240–8000px per side · up to 10MB
15s · 9:16 · 720p · silent
Confirm this before generating.
A failed render is refunded automatically.
My generationsPhotos are used only to render this video.
THE FIFTEEN SECONDS
Three shots, two dancers
The cut is not one continuous take. It hard-cuts at 8.6s and again at 13.0s, and those two seams decide which photo goes in which slot.
01/0.0–8.6s
Opens
One dancer, half-body, hands out toward the lens and then up. She carries the first half alone.
02/8.6–13.0s
Hard cut
Same room, same camera, a different dancer takes the middle. The cut lands on the beat.
03/13.0–15.0s
Cut back
The first dancer returns for the close, so one photo has to cover both ends of the video.
Why the seams matter: the model copies the edit, so each slot is tied to a shot, not to a side of the frame. Swap them and the two halves trade faces.
TWO POSITIONS
Which photo goes where
The two slots are not left and right. They are opens-and-closes and takes-the-middle. Get them backwards and the video still runs — the person who should open ends up in the middle instead.
Opens and closes
Shot 1 and shot 3 · 0.0–8.6s, 13.0–15.0s
Swapped, this person only appears in the middle four seconds.
Takes the middle
Shot 2 · 8.6–13.0s
Swapped, this person has to open and close a routine they only dance once.
Flips both photos and both prompts at once. No re-upload.
WHERE IT CAME FROM
Thirty-eight seconds, posted on 31 October
ADDICTED is the second pre-release single from JISOO's first full-length solo album. It came out on the morning of 30 October 2026, and the music video went up with it.
The next day she posted a hallway clip with Rosé: 38 seconds, one phone, no staging. It cuts between the two of them on the beat — one dances, the other answers, then it comes back. That clip is the one this template follows; the 15 seconds used here keep that shape.
Within four days there were versions from her choreographer, from fan accounts in a dozen countries, and from people who had never danced on camera before.
GenReel is an independent tool. It is not affiliated with JISOO, her label, or the creators of the original clip. The routine in your video is performed by an AI-generated stand-in, not by the artist.
- ADDICTED released, 8 a.m. KST
- The hallway clip with Rosé — "we so addicted.."
Fan versions: night-street solo, the swan duet, the Douyin outfit switch
- The version with choreographer Leejung — "double the addiction"
- The official dance performance video
HOW IT WORKS
Three steps, no filming
01
Pick two photos
One per dancer. The same photo twice is allowed — it reads as two of you.
02
Check the roles
Opens-and-closes on the left slot, takes-the-middle on the right. One button flips them.
03
Generate and download
Usually one to three minutes. Silent MP4, no watermark. Add the official sound in the app when you post.
PHOTOS
What actually works
Works
One person per photo. A group photo makes the model pick one face and paint the rest into the outfit.
Waist-up or full body, not a headshot. A headshot leaves the body to be invented, and the proportions shift between the two shots.
Eyes visible, facing the camera. Sunglasses or a hat breaks the head turn — the one move everyone recognises.
Similar light in both photos. One daylight photo and one at night reads as two different rooms.
Same photo twice is fine. No second person? Use the same one in both slots.
Breaks
A photo cut off at the neck
Heavy filters, stickers, beauty effects
A photo from behind, or a face lost in shadow
A photo of someone under 18, or of a public figure
THREE WAYS
Filming it, templating it, or this
Film it yourself
- What you do
- Learn 38 seconds of choreography, find a corridor, get both people in the same place, and shoot until one take works.
- What you get
- Your own performance, filmed by you.
- How long
- An afternoon, at least.
A CapCut template
- What you do
- You still need footage of both of you dancing. The template only edits it.
- What you get
- A cut in someone else's style.
- How long
- Minutes — after the footage exists.
Two photos here
- What you do
- Pick one photo per dancer. No filming, no choreography, no second person in the room.
- What you get
- 15 seconds, 9:16, both dancers cast by you.
- How long
- Usually one to three minutes.
If you already have footage of both of you dancing, a CapCut template will look more like you than this does. This is for the case where you don't — different city, different schedule, or nobody who wants to be on camera.
WHAT YOU GET
One video, 800 credits
- Length
- 15 seconds
- Aspect
- 9:16 vertical
- Resolution
- 720p
- Audio
- Silent MP4 — add the official sound in the app
No watermark
800
credits per video
If a render fails, the credits come back automatically.
Why silent
The clip carries the song; the export drops it. The recording belongs to the people who made it, so a silent export means the platform won't mute your post for copyright — you add the official sound in one tap when you publish, where it's licensed.
Where it breaks
- The seams: shots 1 and 3 are the same person, and the model sometimes drifts to a third face when it cuts back.
- Two photos with very different lighting read as two different rooms, even though the reference keeps one hallway.
- Hands crossing the face — the first shot is dense with hand movement and loses detail there.
- The middle shot starts in profile. A photo where the face is mostly hidden makes that angle drift.
QUESTIONS
JISOO Addicted Dance AI FAQ
What is the JISOO Addicted dance?
A short two-person routine built on JISOO's single ADDICTED, released on 30 October 2026. She posted a 38-second hallway version with Rosé the next day, and the head turn on the beat became the part everyone copies.
How many photos do I need?
Two — one per dancer. If you don't have a second person, use the same photo in both slots; it reads as two of you.
Why is the video 15 seconds when the original is longer?
The model takes 2–15 seconds of reference footage. The full hallway clip runs 38 seconds; the 15-second cut used here is the part where the two of them alternate. Those two seams are why the edit is 15 seconds and not 38.
Which photo goes in which slot?
The left slot is the dancer in the first shot and the closing shot. The right slot is the one who takes the middle. There's a swap button if you get them backwards — no re-upload.
Will I look like JISOO in the video?
No. Both dancers come from the photos you upload, and the routine is performed by an AI-generated stand-in. The clip's original performers are replaced entirely — your hair, clothes and face come from your photos.
Does the video come with the song?
No — it exports silent. Add the official ADDICTED sound in TikTok, Reels or Shorts when you publish, where the recording is licensed. That also stops the platform muting your post.
Can I post it on TikTok, Reels or Shorts?
Yes. It's a 9:16 MP4 with no watermark. Turn on the platform's AI-generated label when you post — all three apps ask for it.
The two people in my photos are very different heights. Will that break it?
Not the way it would in a two-shot — they never share a frame, so there's no height to match. What matters is framing distance: one full-body photo and one close head shot make the two shots feel like different rooms.
Can I use a photo of my pet?
You can try, but this template is built around two people taking turns in the same hallway. A pet photo usually comes back with the animal standing in a human pose.
How long does it take?
Usually one to three minutes. You can close the tab — the render keeps going and the result stays in your account.
What does it cost, and what if it fails?
800 credits per video. A failed render returns its credits automatically, so a retry doesn't cost you twice.
Is this an official JISOO product?
No. GenReel is an independent tool with no affiliation to JISOO, her label, or the creators of the original clip.
Do I have to say it's AI-generated?
Yes. TikTok, Instagram and YouTube all require the label for realistic AI video, and the platform may add it for you. Leave it on.
Whose photos am I allowed to use?
Your own, and photos of adults who agreed to appear in an AI-generated video and to it being shared. Not a public figure, not a stranger, and not someone under 18.
MORE ROUTINES
Other dances with fixed positions
Two photos. Three shots.
You don't need to learn it, and you don't need the other person in the room.
Make the duet15s · 9:16 · 720p · silent · 800 credits