How to Create AI Duet Videos from Two Photos

A Quick Recap: What's Actually Being Merged

What Makes Two Photos Merge Cleanly (and What Doesn't)

• Framing and scale should roughly match. A tight close-up paired with a waist-up shot puts two very differently sized faces into the same scene, and that size mismatch is one of the fastest ways a merge reads as artificial. General guidance on merging photos with AI backs this up directly: photos with similar perspective and scale avoid the "obviously pasted in" look.
• Lighting direction matters more than lighting quality. Two well-lit photos can still clash if one is lit from the left and the other from the right, since the shadows on each face will point in conflicting directions once they share a frame. AI image-blending guidance names consistent lighting and tone as one of the biggest single factors separating a believable blend from an obvious one.
• Similar camera angle and eye-line. A photo shot from slightly above paired with one shot from slightly below tends to look like two different photo shoots stitched together. Photos taken at roughly the same height and angle composite more naturally.
• Resolution should be reasonably close between the two. Pairing a sharp, high-resolution photo with a blurry or heavily compressed one gives the AI uneven detail to work with. General photo-merge guidance makes the point directly: the source images matter more than the tool itself.
• Faces should be mostly unobstructed. Sunglasses, a hand near the chin, hair covering part of the face, or a hat casting a shadow over the eyes all reduce how much the model has to work with when animating that face to lip-sync.

Before You Start

How to Build the Duet in singingphoto.ai

1. Step 1

   Choose Duet Karaoke Mode

   Select Duet from the three Karaoke Modes: Solo, Duet, Pet Karaoke.

2. Step 2

   Upload the two photos you've already checked against the criteria above

   Asset Management lets you pull in a photo you've uploaded before instead of re-uploading it.

3. Step 3

   Upload the song

   Add the audio both faces will lip-sync to together.

4. Step 4

   Pick an Instant Stage Preset, or use Custom Scene Editing

   Choose one of six presets (Studio, Jazz Club, Home, Bar, Supercar, Fisheye), or write your own prompt for the stage, lighting, camera, or atmosphere if none of the six fits.

5. Step 5

   Generate, then review the merge specifically

   Not just the lip-sync. Check that both faces sit at a believable scale relative to each other and that neither looks out of place against the chosen scene.

6. Step 6

   Export

   Download the finished video in HD, free, with no watermark.

If the Merge Doesn't Look Right

• One face looks noticeably bigger or smaller than the other. This is almost always a framing mismatch between the two source photos. Try re-cropping one photo so both faces occupy a similar proportion of the frame, or swap in a photo shot from a similar distance.
• The lighting looks inconsistent between the two faces, like they were lit in different rooms. Because they probably were. Swap in a photo where the lighting direction is closer to the other one, even if it means using a slightly less flattering photo of one person.
• The merge looks technically fine but somehow flat or generic. This is usually a stage-preset mismatch rather than a photo problem; try a different Instant Stage Preset, or describe a more specific atmosphere through Custom Scene Editing.
• A face looks blurry or under-detailed compared to the other. Swap that photo for a higher-resolution version if one exists.

Common Questions