Which face is the singer and which reacts
The singer is the foreground face under the tortilla. The reaction face is behind and above on the left.
In the final outdoor frame, the singer is the foreground face under the tortilla. The reaction performer is behind and above her on the left. Photo 1 goes in the singer slot. Photo 2 goes to the performer who drops his jaw.
Use the role map before you generate, especially when two people are uploading:
Open Role map above the workbench. Both roles shows two slots. Singer and Reaction each show one.
Find the singer in the final outdoor frame: the face below the tortilla. That person also appears in the outdoor verse and the indoor close-up, gestures toward the bearded performer outside, and brings the tortilla forward.
Find the reaction performer behind and above the singer on the left. In the source he is the blond performer in green headphones. His part is the final jaw-drop through the last frame. The original singer delivers the line when you leave that role unchanged.
Put photo 1 on the person who should sing and hold the tortilla. Put photo 2 on the person who should drop his jaw. Read the Singer and Reaction performer labels before you generate.
Example. Mina should deliver the line. Chris should look shocked. On Both roles, Mina's portrait is photo 1 and Chris's portrait is photo 2. In the final frame, Mina is the face under the tortilla and Chris is behind her on the left. If the files are reversed, Chris sings and Mina does the jaw-drop. The soundtrack and the six-second landscape timing stay the same either way.
What you trade. Photo 1 puts that person on the lyric and the tortilla. Photo 2 spends the joke on the reaction. Leave photo 1 empty and the original singer stays, which is useful when only the shocked face should change. One photo can do either job. Two photos are for doing both.
If the wrong friend is holding the tortilla, swap the uploads. Photo 1 is the foreground singer. Photo 2 is the jaw-drop. If you meant to change only one face, clear the other slot so the original performer remains. On a small screen, use the final outdoor frame: the face under the tortilla is the singer, and the face higher and to the left is the reaction.
Limits. These are the only two mapped faces. The bearded performer the singer gestures toward is not a separate upload slot. The reaction slot is the headphone performer who drops his jaw, not every face in the outdoor group. Use a well-lit, full-face portrait for each slot you fill. A normal expression is fine. The pages do not say the role map can move these people to new positions in the frame.
Which face is the singer and which face is the reaction?
In the final outdoor frame, the singer is the foreground face under the tortilla. The reaction performer is behind and above her on the left, the blond performer in green headphones who drops his jaw. Photo 1 is the singer. Photo 2 is the reaction.
Which upload is photo 1 and which is photo 2?
Photo 1 goes in the foreground singer slot. That person delivers the line, appears in the outdoor verse and the indoor close-up, and holds up the tortilla. Photo 2 goes to the performer who drops his jaw behind her. Swap those files and you swap who sings and who reacts.
How do I check the role map before generating?
Open Role map above the workbench and look at the final outdoor frame. The face below the tortilla is the singer. The face behind and above on the left is the reaction performer. Match those positions to the Singer and Reaction performer labels, then add the portraits.
What if I only want to replace one of the two faces?
Fill only that slot. Use Singer or photo 1 for the face under the tortilla, and leave the reaction empty. Use Reaction or photo 2 for the jaw-drop, and leave the singer empty. An empty slot keeps the original performer. One well-lit, full-face portrait is enough.