The WYW Sweetie AI video, also written WYW My Sweetie, is a 15-second clip of two people running the same jersey club dance side by side, filmed from a phone on the ground, with the trend sound on it. On videosi you make it from two photos, or one photo of you both, and you write no prompt.
This guide answers the questions that sit around the trend and that the page FAQ does not: why the low camera works, what to wear, how to pick photos for two people, whether the two names are the same thing and what you can do with the sound. Everything here comes from the page and its notes, and anything we could not confirm is marked unconfirmed.
Key takeaways
- The video always has two dancers. There is no solo mode, so you need two photos or one photo showing both people.
- The low fixed camera is the look of the trend, so use photos that show the body down to at least the knees.
- Come back and back is the caption for the routine, and www my sweetie is how people search for the sound. We treat them as the same trend.
- The finished MP4 carries the trend sound but nobody lip-syncs. Do not use it in ads without clearing the music.
- A standard 15-second video (Standard model, 720p) is 13 credits, and packs start at $6.99 for 10 credits.
In this guide
- Why does the low camera angle make the dance look so good?
- Are "come back and back" and "www my sweetie" the same trend?
- What makes a jersey club routine look the way it does?
- What should the two people wear in their photos?
- Should I upload two photos or one photo of both of us?
- Which of the three places suits which photos?
- What happens in the 15 seconds, beat by beat?
- Why is the sound on the video if nobody sings?
- Can I use the original song in a business post or ad?
- What should I check before posting the result?
- Which model tier should I try first?
Why does the low camera angle make the dance look so good?
The original clips are filmed with the phone resting on the ground and tilted up. That one choice does most of the work. Legs fill the frame, baggy trousers swing with every step, and the kick into the low lunge seems to rush toward the lens. A camera at eye level would show the same moves as a small, flat routine.
This is also why the WYW Sweetie AI page keeps the camera fixed and low and does not let it move. The motion reference carries the angle, the full-body framing and the timing, so the result looks like the trend clips and not like a generic dance video. The trade-off is that a low lens stretches legs and shrinks faces, which is why the photos matter so much (see below).
Are "come back and back" and "www my sweetie" the same trend?
Yes, for practical purposes. The sound is listed on TikTok as "Wyw my sweetie #JerseyClub - Slowed" by nissi and 25Hzz. "Come back and back" is the caption many posts use for the routine, including the most-viewed version we found, and "www my sweetie" is a common way people type the sound name into search.
What we could not confirm: what the letters WYW stand for. As of October 2026 we found no reliable explanation, so we do not offer one. We also could not confirm whether the song was made with AI music tools, even though some posts ask. If you write a caption, use the sound title as TikTok lists it and leave the guessing out.
What makes a jersey club routine look the way it does?
The tag on the sound says Jersey Club, and the dance follows that genre. According to the Wikipedia article on Jersey club, the style began in Newark, New Jersey, runs at roughly 130 to 140 beats per minute, and is built on swung, bouncy kick patterns with chopped samples. That bounce is why the routine opens with soft knee bounces and hand rolls rather than big jumps.
For an AI video this matters in one practical way: the moves are small, fast and repeated, so the two dancers need to stay in sync. That is the first thing to go wrong when people try this with a plain text prompt, because without the original clip as a reference the dancers drift apart within a few seconds. On videosi the motion reference is built in.
What should the two people wear in their photos?
The dance is done in baggy trousers, and the baggy swing is part of the look, but your photos decide what the AI dancers wear. Each person keeps the complete outfit from their own photo, shoes included. So choose a photo in an outfit you would actually want to see dancing: loose trousers or a skirt that moves, and shoes you can see.
Two simple rules help. Pick clothes that look different from each other, for example a dark top next to a light one, so the two dancers stay easy to tell apart through the fast footwork. And avoid sunglasses, a low hat or a dark photo, because those are the usual reasons the face comes out wrong. The people from the original clip are not kept, so their hats, caps and printed trousers will not appear unless your photo has them.
Should I upload two photos or one photo of both of us?
Both work. With two photos, photo 1 becomes the left dancer and photo 2 becomes the right dancer, and each keeps their own face, hair and build. This is the safer choice when the two people look quite different, because the model never has to guess who is who.
One photo of both people is quicker and is good when you are standing side by side already, with you on the left and your friend on the right. Make sure both faces are clear and neither person is cut off. If the photo has the two of you hugging or overlapping, use two separate photos instead.
There is no one-person mode. Solo posts exist on TikTok, but the videosi page always makes exactly two adult dancers. Only upload photos of adults who agreed to it. Celebrities, public figures and minors are not accepted.
Which of the three places suits which photos?
You can pick the night plaza, a white dance studio or a car park at dusk. The night plaza is the original setting, a tiled city square in front of a tall lit building, and it suits photos taken in warm or evening light. The white studio is the cleanest background and works well when your photos are bright and the outfits are colourful, because nothing competes with them.
The dusk car park has painted bay lines on the asphalt and a deep blue sky. It is a good match for streetwear and dark outfits, which stand out against the blue. None of the places show readable signs or logos.
What happens in the 15 seconds, beat by beat?
The routine has five parts. From 0 to 4 seconds both dancers do soft knee bounces and hand rolls at the chest. From 4 to 8 seconds come side steps with wide arm swings. At 8 to 10 seconds a quick front kick drops into a low, wide lunge that fills the bottom of the frame. From 10 to 13 seconds the footwork stays low and wide with arms crossing the chest, and in the last two seconds both straighten up and bring their hands toward the head.
Knowing the beats helps you judge the result. The kick and lunge at 8 to 10 seconds is the hardest moment, so watch it first. If hands or feet blur for a moment there, that is normal for fast footwork and not a sign the whole video failed.
Why is the sound on the video if nobody sings?
The finished MP4 has the trend sound from the original clip muxed onto it, so it plays with the music when you post it. The dancers do not lip-sync, because the song is the soundtrack rather than something the people mouth. Their mouths stay closed, relaxed or smiling.
Can I use the original song in a business post or ad?
Not without clearing it first. The sound belongs to its rights holders, and videosi gives you the video with the sound on it for personal posting, not a licence for the music. If you plan a brand post, a paid promotion or an ad, clear the music yourself or replace the audio with a track you are allowed to use.
As of October 2026 we could not confirm the licensing terms of this particular sound, so treat it as not cleared for commercial use. For a personal post to your own account, using the sound the platform offers is the normal route.
What should I check before posting the result?
First, check that the two faces look like the two people, in every frame, with no swapping. Second, look at the kick and lunge for blurred hands or feet. Third, make sure the mouths are closed or smiling and nobody seems to be singing. Fourth, confirm there is no stray text or logo in the background.
Likeness is not guaranteed, and a failed, timed-out or blocked render returns its credits. If one dancer looks off, try again with a clearer, brighter photo that shows more of the body. Label the post as AI-generated using your platform's own option, and only post photos of people who agreed.
Which model tier should I try first?
videosi makes the WYW Sweetie video with ByteDance's Seedance family. The Seedance 2.0 page from ByteDance Seed describes it as a video model that takes text, image, audio and video inputs, and that is what lets a photo and a motion reference work together.
On the trend page the default is Standard (Seedance 2.0 Mini), and Fast, Pro and Ultra are one click away. A standard 15-second video (Standard model, 720p) is 13 credits, better models and higher resolutions cost more, and there is no free generation. Credit packs start at $6.99 for 10 credits and never expire, and plans start at $29.90 a month.
