The Talk Dirty AI trend is a dance template, not a chatbot: an AI video in which one photo of you becomes the lead dancer on an arena stage, leading a crew in varsity jackets while pyro flames go off. This guide goes past the basics on the trend page: what the clip shows, what to pick for the song, and what decides whether the result looks like you.
Key takeaways
- "Talk Dirty AI" here means a 12-second AI dance clip set to Jason Derulo's 2013 song Talk Dirty, first seen on TikTok around 18 October 2026.
- CapCut and PixVerse templates spread the trend; videosi does the same job in the browser from one photo.
- The clip is silent, so you add the song and the AI label where you post.
- One 12-second video (Standard model, 720p) costs 10 credits, with packs from $6.99 for 10 credits. There is no free trial.
- A bright, sharp, full-body photo is the biggest factor in whether it looks like you.
In this guide
- What is the Talk Dirty AI trend?
- What song is used in the Talk Dirty dance videos?
- What happens in the 12 seconds?
- How do you make it from one photo?
- CapCut or PixVerse template, or videosi?
- Which AI model is behind it, and can I see the prompt?
- Is it free, and what does it cost?
- Does it work for parents and grandparents?
- Why doesn't it look like me?
- Is the trend still going?
What is the Talk Dirty AI trend?
It is a wave of AI videos that started on TikTok around 18 October 2026. Each clip looks like phone footage filmed from the audience at an arena concert. A lead dancer walks to centre stage, dances a solo in a single spotlight, then leads a crew in yellow-and-black and red-and-black varsity jackets through a groove while columns of flame go off at the front of the stage.
The twist is that the lead is whoever's photo went in. Creators shared templates with captions like "how to make the same dance video with one photo", and the clips show grandparents, office workers and friends taking the lead. The word "AI" in the name tells you the video is generated, and the word "dirty" comes only from the song title. The dance itself is a stage routine of swings, shuffles and a final pose, and the Talk Dirty dance AI page keeps it that way.
We do not know who posted the first clip, and the live performance the template copies has not been confirmed, so this guide names neither. We also do not quote view counts, because we have not verified any.
What song is used in the Talk Dirty dance videos?
The song is Talk Dirty (feat. 2 Chainz) by Jason Derulo, released in 2013 as a single from the album Tattoos. The AI clips use the chorus with the saxophone hook, and that hook is what most people recognise first.
The video you make on videosi is silent. Upload it to TikTok or Reels and pick the Talk Dirty sound from the platform's own library, where its licence covers the music, then line the flame burst up with the chorus. Do not bake a copy of the song into the file. videosi is independent and not affiliated with Jason Derulo, 2 Chainz or their labels.
What happens in the 12 seconds?
The routine has six beats, and the moves come from an AI-made clip of an ordinary dancer, not from concert footage.
- 0 to 1 s, walk to centre. A dark stage. You walk to the middle as the crew steps aside.
- 1 to 3 s, spotlight solo. One white spotlight, a wide stance, knees bent, hips grooving and arms swinging loosely in front.
- 3 to 5 s, shuffle. The lights come up, the crew lines up behind you and you shuffle with the forearms crossed low.
- 5 to 7 s, pyro and arm swing. A flame column bursts at the stage front and the whole line swings their arms forward on the beat.
- 7 to 9 s, chest groove. Warm orange light, the crew closes in and you groove upright with a hand tapping your chest.
- 9 to 12 s, final pose. The screen flares white, you step forward to the crowd and the crew drops into a crouch around you.
The light changes colour three times, from white to blue-white to red and orange. That matters later, when we get to why a face drifts.
How do you make the Talk Dirty dance from one photo?
Four steps on the Talk Dirty dance AI page, with no prompt to write:
- Upload one photo of yourself, full body, facing the camera, in even light. The outfit in the photo is what you wear on stage.
- Pick a format: 9:16 for TikTok and Reels, or 1:1 or 16:9. The arena, the crew and the pyro are fixed by the trend.
- Press Generate. The 12-second clip is rendered as a silent MP4 with no watermark.
- Download it, upload it to TikTok or Reels, add the Talk Dirty sound there and switch on the AI label.
It runs in the browser on a phone or a computer, so there is no app to install and no template link to copy.
Should I use a CapCut or PixVerse template, or videosi?
The clips on TikTok were made with shared CapCut and PixVerse templates, not with a live TikTok camera filter. Templates are quick when you find a working one, but they change or disappear often, and each one is somebody else's setup.
videosi does the same thing as a page you can return to: one photo in, a 12-second stage clip out. The difference you will notice is control. You choose the photo and the format, you see the price before you press Generate, and a failed, timed-out or blocked render returns its credits automatically. Use a template if you already have a link that works; use the generator if you want it to work the next time too.
Which AI model is behind it, and can I see the prompt?
videosi makes the clip with Seedance, the video model from ByteDance. Seedance 2.0 Mini is the default, and Fast, Pro and Ultra (which is Seedance 2.5) are one click away. It takes your photo and a 12-second motion reference, and writes nothing on the screen.
The full prompt is not published, but its structure is no secret. It has six blocks: identity, motion, scene, timeline, rules and audio. The identity block says your photo is the only source for face, hair, build and outfit. The motion block says the reference supplies the moves and nothing else. The rules block keeps you the brightest person in frame and the crew's faces generic. If you try the dance in a general video model with a text prompt, the usual failure is an invented nightclub or a generic routine, and a backup dancer who ends up with your face.
Is Talk Dirty Dance AI free, and what does it cost?
It is not free. There is no free generation and no free trial. One 12-second video on the Standard model at 720p costs 10 credits. Better models and higher resolutions cost more, and the generator shows the price before you press Generate.
Credit packs start at $6.99 for 10 credits and never expire, and plans start at $29.90 a month if you plan to make many. A failed render returns its credits. Since a try is cheap and the photo is the main variable, change the photo before you change anything else if the first result is off.
Does it work for parents and grandparents?
Yes, with their permission. Grandparents leading the crew are among the most shared Talk Dirty AI clips, and the examples on the trend page include a grandfather in a flat cap at the front.
- Ask first. Only use your own photo or one from someone who agreed.
- Get the whole body. A clear photo from head to shoes keeps the outfit right on the big stage.
- Skip celebrities and children. Do not upload photos of either.
- Label the post. Switch on the AI label when you share it, so nobody thinks their relative really took the stage.
Why doesn't my Talk Dirty dance video look like me?
Most misses come from the photo. A small, dark or sideways face gives the model too little to hold through three lighting changes in 12 seconds: the white spotlight, the blue-white wash and the red pyro light. Without enough detail, the face drifts with the light.
- Use a bright, sharp, front-facing photo without sunglasses or a cap pulled low.
- Show your legs and shoes. The routine is a wide stance, a shuffle and a crouch around you.
- Wear something that stands out. The crew wears yellow, red and black, so a light or bold outfit makes you pop as the lead.
- Avoid busy backgrounds in the photo, which compete with the stage.
Even then, likeness is not guaranteed, and a hand or a backup dancer may blur for a few frames during the fast arm swings. The crew are generic AI dancers, and you will not be given a singer's face or styling.
Is the Talk Dirty dance AI trend still going?
As of 25 October 2026 it is still spreading. The first AI stage clips appeared on TikTok around 18 October, template tutorials followed from 19 October, and new clips were still being posted on 24 and 25 October. Searches for the dance rose over the same week. That puts it in its early weeks, not past its peak.
If you want something ready to post, the Talk Dirty AI generator takes about the time of a coffee. For more of the same kind, the Gangnam Style AI page turns another classic song into a one-photo dance, AI Dance lists the dance trends that work from one photo of you, and ATEEZ BAD AI does the same for a K-pop routine.
