The IVE Looks Can Kill AI dance is a 15-second video of you, and up to two friends, doing the chorus of the Looks Can Kill dance challenge, made from one photo of you. You can start with a single photo, and the video will still show three dancers, with you in the centre.
IVE started the challenge on 12 October 2026, and per Kyunghyang Sports the challenge videos reached a combined 100 million views by 28 October. This guide answers the deeper questions that the trend page does not: why it is always a trio, how photos and places change the result, what the audio can and cannot do, and what to check before you post.
Key takeaways
- The result is always a trio, with you in the centre spot.
- Add two friends as Photo 2 and Photo 3, or get ordinary stand-ins on the sides.
- The original sound is on the MP4, but nobody lip-syncs and platforms may mute it.
- The effect has not been test-rendered yet, so a likeness is not guaranteed.
- A 15-second Standard 720p video costs 13 credits, and an account's first video costs 10.
In this guide
- Why is the result always a trio and not a solo?
- How do I add two friends to the line-up?
- What photos work best for the centre spot?
- What is the sniper dance move in the routine?
- What happens second by second in the 15 seconds?
- Which place should I pick for my version?
- Does the song play, and will a platform mute it?
- Why do three dancers sometimes share one face?
- What does one video cost in credits and dollars?
- What should I know before I post it?
Why is the result always a trio and not a solo?
The challenge is built around three performers, so the clip we use as the reference has three people in a row: one in the centre, one on the left and one on the right. You always take the centre spot. The left and right dancers become ordinary adult stand-ins in plain clothes when you upload only one photo.
We chose the trio on purpose. The moves depend on the line of three, with the centre dancer stepping forward for a few seconds while the others hold the beat, and then all three running to the lens together. Removing two of them would mean inventing a different routine. So there is no solo option, and the page says so up front. The stand-ins are not IVE members and not anyone you know, and nobody in the result keeps the face of anyone in the original clip.
How do I add two friends to the line-up?
Upload a second photo for the left spot and a third for the right spot. Each photo should show only one person, because a group photo makes the model guess who belongs where. Both friends should be adults who have agreed to appear.
You can add just one friend. In that case the other side keeps a stand-in, and the result is still a trio. The labels on the IVE Looks Can Kill AI dance page are Photo 1 for you in the centre, Photo 2 for a friend on the left and Photo 3 for a friend on the right, so it is easy to see which face goes where. If you want a post that feels like a real group, give each friend a clear, separate photo and avoid using the same person twice.
What photos work best for the centre spot?
The centre spot gets the most screen time, including the solo step forward and the final run toward the camera, so this photo carries the video. A few habits help a lot:
- Use a sharp, well-lit photo with no sunglasses, hat or hair across the face.
- Show yourself from head to at least the hips, so the model has your clothes for the full-body stage shots.
- Pick fitted tops and trousers that show the arms and legs. Long coats and big bags hide the hand moves.
- Keep only one person in the frame.
The instruction replaces the original centre dancer outright, including her hair and the vinyl outfit, so your own hair and clothes come from the photo. If the first result does not look like you, a sharper photo is the first thing to change.
What is the sniper dance move in the routine?
Korean outlets describe the key move as a hand gun shape, as if shooting. In its report on the 12 October release, Kyunghyang Sports says the three members form a gun shape with their hands against a hero academy backdrop, with long extended arm movements adding energy. By 28 October the same publication was calling the point move the "sniper dance", and it listed actors, idols and dancers who had tried it.
In our 15-second clip the pointing arrives late, in the arm sweeps and points between roughly 6.5 and 12 seconds. The song title explains the idea: a look strong enough to take someone out. We do not teach the choreography, since the video is made for you, but knowing where the point lands helps you judge whether a fast-moving AI result kept the shape of the move.
What happens second by second in the 15 seconds?
The clip follows the chorus in five beats:
- 0 to 1.5 s: you stand close to the camera with both arms thrown wide, then drop back into the line.
- 1.5 to 4.5 s: all three dance in a row with quick hand moves at chest height while the lights flash blue.
- 4.5 to 6.5 s: you step forward alone with a hip sway and a hair flip, and the side dancers keep the beat.
- 6.5 to 12 s: arm sweeps overhead, a point up and out, then a step back into formation.
- 12 to 15 s: all three run up to the lens and frame their faces with their hands, then wave.
The last beat is the hardest for any video model, because three faces fill the frame at once. That is where faces drift toward the original singers, and it is why we name each original outfit and hairstyle to remove. Fast hand moves near the face can also come out soft.
Which place should I pick for my version?
There are three choices. The original stage keeps the dark studio with its curved lighting truss, blue spotlights, glossy black floor and the big red light-bulb sign behind the dancers. A dance studio swaps that for a black back wall, a warm wooden floor and a row of small bright lights. A street at dusk puts the trio in front of a chain-link fence and a low concrete wall, with a grey apartment building behind.
If you want people to recognise the challenge at a glance, keep the stage. If you want your post to stand out from the many fan covers in practice rooms, offices and bedrooms, pick one of the new places. When you change the place, every frame moves there and the stage never appears. We have not test-rendered the studio or street yet, so they are the more experimental options.
Does the song play, and will a platform mute it?
Yes, the original sound is on the finished MP4. The site mixes the chorus in after the render, taken from the same section as the official trio clip, so the moves are timed to the music. Nobody sings or lip-syncs in the video, and the dancers keep relaxed mouths.
The song belongs to its rights holders, so a platform may mute or limit a repost, and videosi cannot predict that. If your post is muted, you can usually add the platform's own licensed version of the sound when you upload. Check the current rules of the app you post on before you count on the original track. The timeline matters here too: the single was pre-released on 19 October and the mini-album followed on 26 October, so licensing on each platform may keep changing.
Why do three dancers sometimes share one face?
This is the best-known weakness of multi-person AI video. With three dancers in a row, a plain prompt often copies one face onto two or three people. The shiny vinyl outfits are another trap, since unless they are named, the pink or red trousers can stay on your body.
Our page works around it in four ways. It replaces each original dancer outright, names her hair and outfit, asks for exactly three adults, and states that one face appears on one person only. That reduces the problem but does not remove it. Photos that look very alike make it worse, so give each spot its own clear face. Because the effect has not been tested with real renders yet, we cannot promise a perfect likeness.
What does one video cost in credits and dollars?
This is not a free effect. A 15-second video on Standard at 720p costs 13 credits, and an account's first video costs 10. Better models and higher resolutions cost more. The Single pack is $6.99 for 10 credits, the Trio Pack is $17.99 for 30 credits, and credits never expire. There is also a Trend Pass at $9.99 a month with 30 credits.
A failed render returns its credits automatically. Any pack is a one-off purchase, and plans renew until cancelled. For a single post with friends, the Single pack covers a first video, and the pack you pick later depends on how many versions you want to try.
What should I know before I post it?
Four things are worth saying plainly. First, the result is always a trio, and you are in the middle. Second, the song is on the clip but nobody lip-syncs, and fast hands near the face can look soft. Third, we have not test-rendered this effect, including the studio and street places, so treat the first result as a draft and try a sharper photo if the face is off.
Fourth, videosi is independent and is not connected with IVE or Starship Entertainment. Label the post as AI when you share it, and only use photos of adults who agreed. For how large the challenge became, see Kyunghyang Sports on the 100 million combined views reported on 28 October 2026, then try the IVE Looks Can Kill AI dance page.
