The Mandem notebook
Mandem in Disguise AI explained: the moment before and after the flash
The joke works because the setup is ordinary. Someone looks into a camera, a flash fills the frame, and the same scene returns with black balaclavas and puffer jackets. Here is what to look for in the reference before making your own version.
Before the flash
After the flashThe reveal has three beats
Watch the opening before focusing on the disguise. The mirror template establishes two people standing together; the close-up template puts one person behind the other. That arrangement gives the viewer a clear before image to remember.
The bright flash hides the change of clothing. The reveal then returns to the same composition, with a masked face and dark jacket replacing the everyday appearance. A convincing version keeps those positions and reactions readable rather than introducing an unrelated scene.
Choose the scene, not just the cover image
Solo is the simplest option when only one person should appear. Mirror selfie gives two people a shared bathroom-mirror scene, including the phone held by the person on the right. Close-up flash brings the right person closer to the camera while the other stays at the back left.
These are three references available in this studio. The two duo previews are about 12 seconds long; our personalized output is 10 seconds. The preview player shows the source duration, not a promise to reproduce every frame.
Your photo and the video have different jobs
Your portrait supplies the identity. The selected reference supplies the scene, positions, movement and transition. For Duo, upload two separate single-person portraits: Left person first, Right person second. You do not need to take a photo together.
Replicate documents labeled image and video references for Seedance 2.0. Our studio uses that input pattern, but facial consistency is still a generative outcome. A reference is guidance, not a literal frame-by-frame face replacement guarantee.
Review the opening as carefully as the ending
Pause before the flash and check whether each face is recognizable. Then compare the eye region, body proportions and position after the reveal. In a duo, make sure the two identities have not traded places and that neither person disappears.
The clips shown here are reference material supplied for the template. They are not customer results or proof of a particular success rate. We do not attach unsupported view counts, a definitive origin story or a promise that your post will go viral.
Watch the reference
Sources & reference material
Generation can change identity, motion and audio. Preview your result before sharing.
