How to recreate a trending video with yourself in it

The workflow

1. 1

   Pick a trend and tap Recreate

   Open Recast. The Motion Library sits beside the composer: Community holds the trends, and Made holds Made’s own clips and Recast examples. Tap a card to watch the template (a Community template names its creator), then tap Recreate. Its whole recipe loads: the clip, the template’s cast images under Attachments as @Image1, @Image2 and so on, and its direction as the prompt. A trend with one person who stays in frame, face visible, is the easiest first test.

2. 2

   Put your photo in place of @Image1

   Tap Use your photo and choose a clear photo of yourself: JPG, PNG or WebP under 10 MB, face visible, and the whole outfit if you want it carried over. It takes @Image1’s place, and while the prompt is still the template’s, Recast rewrites it without the template cast’s description so your photo sets the look. In a two-person template, choose two photos at once to replace @Image1 and @Image2. A template that loads only its clip takes your photo under Attachments. Use photos of yourself or of adults who agreed.

3. 3

   Or start from a clip you filmed

   To use your own footage, tap “Add a reference video to edit” and choose an MP4 or MOV: 4–30 seconds for Recast Pro, 4–15 for Recast Lite. A longer clip is trimmed to the model’s limit before upload, and footage outside its size or frame-rate range is converted for you; a clip under 4 seconds, or one far narrower or wider than 9:16 or 16:9, is refused. Film one person fully in frame, in steady light, then add your photo under Attachments.

4. 4

   Choose Pro or Lite, check the price, generate

   Pick Recast Pro or Recast Lite in the menu at the top of the composer. Under Options, Resolution starts at 480p, with 720p and 1080p, and Prompt is optional. Generate shows the price; Recast checks it again before starting and stops if it came out higher. The video goes to the top of Assets, and you’re told when it’s ready. Watch the turns, the hands and the last second, not only the first frame, and check that the cuts still land with the sound.

Recorded Made demonstration

One crosswalk trend, recast from one photo

This is a real Recast Pro run from October 27, 2026. Crosswalk Street-Style Walk, an 8.7-second Community trend from the Motion Library with its own sound, was recreated, and a photo of Kofi, a man Made generated for its AI Influencer examples, went in as @Image1. The request is exactly what the tool builds for Recreate, Use your photo and Recast Pro at 480p, sent through Made’s API from a script rather than clicked through the browser. He replaces the blonde performer in every shot, with the same poses, cuts, camera and street, and the original soundtrack is kept unchanged. The template’s brown tote, logo included, stays in his hand because the photo shows no bag. By the end the cuts land about a quarter second early, so the picture runs slightly ahead of the sound, and at 480p the closest shots look soft. It’s one run, not a measure of how often a recast holds.

Recast Pro (Seedance 2.5 Edit) · Generated 2026-10-27 · View the example record and price basis (JSON)

Prompts you can inspect and adapt

Recast the person in @Video1 as the person in @Image1: their face, hair, skin tone, body shape and outfit. Keep every move, gesture, expression and mouth movement at the same moments, and keep the place, lighting, framing and camera movement exactly as in @Video1. Anyone else in @Video1 stays as they are.

Made Recast: video.
Edit @Video1 into a new photorealistic version. Preserve its exact duration, shot order, hard cuts, camera motion, poses, blocking, gestures and timing. Generate visuals only; the original soundtrack will be restored after processing.
Use these replacement references consistently throughout every shot:
@Image1: @Image1 replaces the single central foreground woman in @Video1: the cap-wearing person who walks toward camera carrying a tote and briefly sits with a phone. Supplies her new face, hair, full walking outfit and bag; background pedestrians are not targets.
Replace the matching subjects, clothing, objects and environment with the appearances in the replacement references.
Deliver only the unobstructed live-action scene from the very first frame through the last. Exclude all overlaid text, interface controls, borders and graphic marks from every frame. Seamlessly reconstruct any scene details obscured by source overlays, including when those overlays move.
Use the new fictional identities and outfits from the replacement references for the recast people. Preserve their performances and positions, not the original people's likenesses. Apply the references to the source choreography; do not freeze the scene in a reference pose or copy a character sheet's studio background. Keep natural hands, contact, shadows and perspective. No added foreground people, captions, logos, split screens or extra cuts.

Preserve the source shot duration, camera movement, timing, and all elements not explicitly selected for editing.

How a recast is priced

Add your clip, or recreate a template, to see its exact price on Generate. Before the run starts, Recast checks the price again; if it came out higher than the button showed, it stops and asks you to try again instead of charging more.

What to review before you use the result

Work with Claude and Made

The Motion Library, a template’s recipe and Use your photo live in Recast itself. With the Made connector, Claude can look up the video-editing model in the catalog and quote an edit of a clip and photo you supply, if the connected client exposes that model and an upload path.

Use Made to put me into my attached dance clip in place of the dancer, using my attached photo. Check the current catalog for the Seedance 2.5 edit model and its video and image inputs. Keep every move, the timing, the camera and the original sound. Show me the price and the prompt before generating one result at 480p.

Questions about this workflow

Keep creating