Reel guide
Above the clouds — yourself instead of the person in the trend
- Mode
- Genjutsu Lite → Motion Transfer
- What it cost me
- 40 credits per run · 200 for me, failed tries included
How it works
A man by a tent in the mountains turns around, and after the cut he's standing above the clouds, arms spread wide. The scene, the light and the movement come from the source. The AI replaced the man with me, from my real photos.
What you'll need
No code needed if the person in the source has empty hands.
Step by step
Done: 0 of 5
Step 1. Source
Where: Instagram
Find the trending reel and download it (any reel downloader or a screen recording). Look at the person's hands: if he's holding something, the AI will carry the object over along with the gesture. Pick a source with empty hands.
Step 2. Your photos
Where: Phone
You need four real photos of your face: straight on, three-quarter and profile. Even light, no filters. The fifth image is the outfit: a person from behind in the clothes you want. The AI takes only the clothes from it.
Step 3. Generation
Where: Higgsfield
In Higgsfield choose Genjutsu Lite → Motion Transfer → upload the source as the video and five images in this order: four face photos first, then the outfit → paste the prompt → run. Check the price in credits before you run it.
Prompt · brackets: the outfitORIGINAL VIDEO: @[Video 1](video_1) IDENTITY: @[Image 1](image_1) @[Image 2](image_2) @[Image 3](image_3) @[Image 4](image_4) OUTFIT: @[Image 5](image_5) Use the attached video as the exact timing, camera, framing, body motion, scene and editing reference. Keep the scene, light and cut unchanged. Replace the man in the video with the real person in images 1, 2, 3 and 4. These are real photos of him from the front, three-quarter and profile: copy his head and face exactly from them at every angle. It must be the same person. Do not invent, slim, lengthen, age or idealise any feature; if in doubt, follow image 1. Image 5 is only for clothing: [a plain black baseball cap, a black hooded jacket with the hood down, black trousers]. Do not take the face or head shape from image 5. His hands are empty. Photorealistic phone footage, natural grain, true colours. One person only. No text, no icons, no captions.
The prompt doesn't describe the face, it only says “copy it from the photos”. Describe a face in words and the AI starts improving it.
Step 4. Check
Where: Phone
Scrub through the result frame by frame. On the close shots compare the face with the photos, and look at the hands: mine still has slight smears near the fist.
Step 5. Edit
Where: CapCut
My result came out square. I put it on a black 9:16 background and added text and sound.
Where it breaks
- Motion Transfer copies an object in the hand along with the gesture. Forbidding it in the prompt doesn't work;
- trying to remove the object with Object Swap turned it into a black pole: 40 credits wasted. In the end I erased the object in the source before generating, with code;
- a sheet with a generated person gives someone else's features. Take the face only from real photos.
Follow — a new trend almost every week
Open Instagram




