Trending Reels with your AI model: turn a video you have the rights to into her version

Most of what goes viral on Instagram is a format: the same dance, the same walk to camera, the same "get ready with me", done by thousands of accounts in their own way. For an AI influencer that's good news. You don't need an idea a day; you need to join the right formats quickly, in her version, before they fade.

One rule first: start only from a video you have the rights to. Your own clip, one you shot or paid for, or one whose creator gave you permission. fanlygen asks you to confirm it for every source video, and the result is a new AI-generated video with your model, not a copy of somebody else's footage.

This is how we do it with our own AI models (Carima and Jessica, whose numbers we shared in this case study), and what we learned building the trend tool in fanlygen.

First, know what kind of trend it is

Not all trends are the same problem, and the tool that nails one fails another.

1. One continuous shot

One person, one camera, one take: a dance, a walk, a turn. This is the easiest case. The model copies the motion of the original onto your character, frame by frame, and keeps the music.

What to check: the start. The clip begins from a photo of your character in the pose of the first frame, so that photo decides everything. Pick it before you animate.

2. A trend with cuts

Many Reels are edited: front, back, close-up, outfit change, four or five shots in fifteen seconds. Copying this as one continuous video can't work, because the original isn't continuous.

What works: splitting it. Find the cuts, make one photo of your character per shot, check each one, animate each piece with the motion of its shot, then put them back together on the original audio.

3. A trend where someone talks

"Things I'd never do as an AI", a story to camera, a reaction. Here the motion matters less than the mouth: the words have to match her lips, and the voice has to be hers, not the original creator's.

What works: transcribing what's said, regenerating it with her cloned voice (in the same language or translated), and syncing the lips to the new audio.

How it works in fanlygen

1. Upload the video, or paste its link, and confirm you have the right to use it as a reference. Pick your model.
2. Choose the mode: faithful replica (one shot, up to 15 seconds), cut-by-cut (edited trends, each shot handled separately) or talking (her voice, in the original language or another one).
3. Pick the starting photo. You get candidate photos of her in the pose of the opening frame. You can keep the original setting and outfit or change them. For cut-by-cut you see one photo per shot and can redo any single one.
4. Animate. The motion comes from the original video, the face from your character. You can leave the page; the result lands in your library and you get a notification.
5. Publish to Instagram or Fanvue from the same place, or schedule it (on TikTok it arrives as a draft you post from the app).

The faithful replica costs 25 credits per second of video. Plans start at €11.90 for 500 credits a month, and new accounts get 300 free credits.

The limits nobody mentions

• Moderation filters. Some motion models refuse clothing they consider too revealing, even a normal bikini. When that happens we fall back to a different engine automatically, with no extra charge, and tell you the result will be "similar" rather than identical.
• Outfit changes inside one shot (the classic "transition") are still hard for every motion model we've tested: they keep the outfit of the starting photo. Split the trend at the change instead.
• Watermarks. Uploading a screen recording brings the platform's logo and the creator's handle with it. Paste the link instead.
• Rights. Use only videos you own or have permission to use, and add trending sounds through the platform's own library. Re-uploading someone else's footage with only the face swapped is copying their content, not joining a trend.
• Label it. It's AI-generated content of a synthetic person: use the platform's AI label (Instagram's "AI info").

Other tools that do this

To be fair, it's a whole category now. Inflyu clones a Reel from its link up to 15 seconds and keeps the music; The Influencer AI remakes a reel from a URL up to 10 seconds; Higgsfield is strong on motion. If you only need trends, they're worth a look. fanlygen's difference is that the trend is one step of a workflow that also creates the character, plans the week, publishes and answers fans. Here's our honest comparison.

FAQ

Can my AI influencer do a dance trend with her own face?

Yes, starting from a dance video you have the rights to (your own is the simplest): pick a starting photo of her in the opening pose, and the motion follows the source video.

How long can the video be?

Up to 15 seconds for a faithful replica; edited trends are handled shot by shot.

Does it keep the original music?

Yes, the audio of the trend is put back on the finished video.

Can she redo a video where someone talks, in another language?

Yes, with the talking mode: her cloned voice, translated if you want, with the lips synced to the new audio.