Give the photo and video different jobs
The photo is the appearance reference: the person’s face, hair, skin, visible clothing and body proportions. The video is the performance reference: where the person sits, how the chair rocks, when the head turns and how the mouth moves in the singing shots.
The supplied prompt describes a young woman with a black braid, black T-shirt and jeans while also requesting the photographed person’s own appearance and clothes. Choose a clear full-body photo so the intended identity and wardrobe are readable; a prompt cannot supply details hidden outside that image.
A reusable structure for the editing instructions
A compact outline is: ‘Replace the only performer in the reference video with the person in Image 1. Preserve that person’s face, hair, visible clothes and proportions. Copy the reference performance, mouth movements, chair motion, camera angles and cuts. Keep the green office and full landscape sequence unchanged.’
This is an explanatory outline, not the complete stored prompt or a guarantee that any video model accepts the same wording. The built-in instructions also call out close-up identity, blinking, facial expression, the chair’s backward tips and the flash near 15.3 seconds.
Lip-sync follows the reference performance
The instruction asks the new performer’s lips and jaw to open and close at the same moments as the source. It also asks for the same looks into the camera and facial motion rather than a frozen face over a moving body.
The finished edit restores audio from the supplied clip. This is not voice cloning or a new vocal recording. If the mouth appears early or late, compare several close shots before judging the entire sequence; keeping the audio track does not itself prove perfect lip-sync.
The prompt preserves the scene’s editing
The wide and close shots should stay in their original order, with the room, chair, lighting and color treatment carried through the cuts. The prepared instructions do not ask for added camera moves, subtitles, logos or a different office.
The studio keeps the complete 19.04-second picture in 16:9. No text prompt is required from you. Use the upload tutorial for the practical steps, and inspect the finished result because identity, fingers, clothing and fast transitions can still vary.
Follow the upload-to-download tutorial · See the scene sequence and source credits
Reviewed against the EasyVideoAI studio on October 30, 2026. Check the live studio for current availability and the final credit quote.