Before attempting a remix
Last reviewed: October 11, 2026.
If you want to make a Bulin 47 video with your photo, define what you want to change first. Replacing a person in an existing recording, animating a photo with reference movement and changing a voice are different workflows. This guide helps you plan a test. The separate homepage generator accepts a permitted adult photo and a 3–10 second reference clip to guide movement. It does not provide a licensed original template or guarantee exact recasting. No paid end-to-end output study or measured success rate is established here.
Start with the original and AI versions explained and our sources and their limits.
1. Choose a specific goal
Character replacement: changes a person's appearance while following the source action. Evaluate this approach if you want to keep the interaction with Kevin Wai, the speaker and the setting. Replacing a face does not necessarily change the body, hair or clothing.
Motion transfer: uses a reference video to guide the movement of a person represented in a photo. It may rebuild the setting. Similar shoulder movements do not prove that Kevin, the hands or the original camera shot were preserved.
Audio and lip synchronization: reusing permitted audio, adjusting mouth movement and creating another voice are separate tasks. Adding a soundtrack after editing does not automatically fix mismatched lips. This exercise does not require imitating the artist's voice.
2. Gather materials you can use
Prepare a clear photo of an adult who has agreed to the intended use and a video segment you have permission to edit. A half-body or full-body image can help describe the intended appearance. Avoid blurred images, covered faces and tiny subjects. Check the service's current input requirements before uploading anything.
A public link is not permission to download, edit or redistribute a video. Photo permission does not also cover the recording, performance or music. We do not have commercial permission to offer the original as a template. If you cannot obtain the required permission, use your own street freestyle scene with consenting participants. Describe it as inspired footage without promising the original scene or sound.
3. Identify the target and the parts to keep
In a scene with several people, identify the subject by position, clothing and action. When studying the original, the intended target would be the performer at the back in the pink shirt. Kevin is at the front holding the phone and speaker. Check whether the tool allows that subject to be selected. A text box alone does not guarantee control over every person.
For permitted footage, adapt an instruction such as:
Replace only the performer at the back with the person in my photo. Keep the shoulder rhythm, camera position and interaction. Leave the person in the foreground, phone, speaker and crowd unchanged. Do not add a new voice.
This states an intention, not a preservation guarantee. If the service does not expose subject selection, do not assume it will choose correctly.
4. Make a separate audio decision
Choose silence, permitted original audio or your own recording. A sound being available inside Instagram or TikTok does not establish permission to export it with another video or use it commercially. Write down which track you can use, where you will publish and what uses are permitted. Watch and listen to the entire result. Compare the beginning, ending and mouth movement. A restored soundtrack can still be out of sync.
5. Test five seconds and inspect failures
Start with about five seconds that meet the chosen service's requirements. Pick footage with a visible subject and, if possible, a head turn or brief foreground obstruction. Before trying again, check whether Kevin was replaced by mistake, whether identity recovers after the obstruction, whether hands merge with the speaker, whether the crowd changes and whether sound matches. Record failed attempts too. If the wrong person is repeatedly selected and you cannot control selection, stop using that route.
Public workflows and your next step
Charlie Sebastian describes a Genjutsu workflow. eroq's guide explains its own method. These are third-party public descriptions. They are not tests conducted by this site or current compatibility and output guarantees.
Prepare your remix plan. The planner organizes decisions in your browser without uploading photos, saving the plan or generating a video.
For the account-based upload and generation workflow, use the homepage generator. The local planner below is a separate preparation aid.