Pet Duo Videos from Two Photos: What Works

Two photos, one video, zero training — and one of the most requested twists we see is pets. A cat duet with a dog. Your childhood dog beside your current one. Pet duo videos generate well, but animals play by different rules than human faces, and knowing them saves you credits.

Do pet photos actually work?

Mostly, yes — with a caveat. The models behind DuetTake were trained overwhelmingly on human faces, so they read human identity cues (eyes, nose proportions, skin texture) far more reliably than fur patterns. Dogs and cats with distinct, front-facing features generate well; heavily patterned or fluffy breeds where the face blends into the body can soften or drift more.

What makes a good pet photo?

The rules from the human photo guide apply, with pet-specific adjustments:

• One animal per photo. The model animates one subject per side; a group shot of three dogs forces it to guess which one you mean.
• Face the camera, eyes visible. A side profile of a snout gives the model almost nothing to hold onto.
• Even light, no flash glare. Shiny eyes and blown-out fur erase texture the model needs.
• Sharp, recent photo. If your pet has grown or been groomed since the photo, the result reflects the photo, not the pet today.

Which combinations work best?

Human + pet is the most reliable pairing — the human anchors the scene and the model leans on the stronger identity. Two pets of different species (cat + dog) works when both faces are clear. Two similar-looking pets — two golden retrievers, two tabby cats — is the hardest: the model can blend them into one performer, especially if the photos differ in lighting.

What about the template and music?

Template motion was designed around human bodies. Pets in human-style choreography can look uncanny or adorable depending on the template — calmer, less gestural templates read best. The music track is embedded regardless, so pick a template whose energy matches the pet; a head-bob template often suits a dog better than a dramatic arm-swing one.

How much will it cost to experiment?

Draft at 480p (20 credits) until the pairing looks right, then export the final at 720p. Failed generations refund automatically, so a weird result costs you time, not credits. If you plan to post the video, the social export guide covers aspect ratios and captions.

Curious what your pet looks like in a duet? Open the generator.

Frequently asked questions

Can I use a photo of a pet that passed away? Yes, and it is one of the most meaningful uses we see. Use the clearest, most recent photo you have — the result reflects that photo faithfully.

Will my pet look exactly like the photo? Close, not identical. Fur detail and color hold well; exact markings can soften, especially on multi-colored breeds.

Can I put a collar or outfit on the pet in the video? Only if it is clearly visible in the source photo. The model animates what it sees — it will not invent accessories.