
OriginalBest friends
Two friends trade verses under the mic, made from two separate selfies.
Upload one photo of each person, or pet, and they become a rap duo in a bright orange studio, sharing one hanging microphone. Use yourself and a friend, your partner, your grandparents or your dog.
You don't need to write a prompt or film anything. Pick a frame, hit generate, and download the MP4 with the original audio.
Photo 1 · left · Photo 2 · right
No photos handy? Try a sample pair:
Use one clear, front-facing photo per performer, and only photos you have permission to use.
15 seconds · 720p · original audio
Sign in to generate. Credits are refunded if a render fails.
The routine your duo will perform
Your two people stand in an orange studio and take turns at the mic: the one on the left points at the camera and raps first, then the one on the right takes over. The moves come from a built-in reference clip.
Each run comes out a little different. The examples below were made the same way, and show the photos that went in.
Choose your two performersMade with Froging AI
Each clip was made with Froging AI from the two photos shown. The people and pets in the photos are AI-generated too. Turn up the volume to hear the audio, or load any pair into the generator above.

OriginalTwo friends trade verses under the mic, made from two separate selfies.

OriginalGrandma and grandpa trade verses, made from two separate portraits.

OriginalA couple version for an anniversary post.

OriginalA golden retriever and a tabby cat perform the same routine.

OriginalTwo coworkers, for a team intro or launch post.

OriginalAn astronaut and a knight share the mic.

OriginalA mother-and-daughter duet for Mother’s Day or a birthday surprise.

OriginalA pug and a husky, made from two pet photos.
You pick the two people. The moves are already set.
One JPG, PNG or WebP photo per performer. They don't need to be in the same photo. You can also start from a sample pair.
Photo 1 becomes the left performer, who opens the verse; Photo 2 becomes the right. Use the swap button to change roles.
Choose 9:16, 1:1 or 16:9. Every video is the full 15-second routine in 720p, and the button shows the credit cost.
Rendering takes a few minutes. Watch it through, then download the MP4 with sound. Failed renders are refunded automatically.
Your two photos are sent to Seedance 2.0 alongside a fixed orange-studio performance clip. The photos decide who is in the video. The clip decides the setting, timing and gestures, which is why every pair does the same routine.
That's also why there is no prompt box.
The reference clip shows two performers in front of an orange backdrop with one microphone hanging between them, trading lines and gestures.
Pairs that tend to work, with a caption you can post them with.
Pair the quiet friend with the one who never stops talking in the group chat. Whoever is in Photo 1 raps first. Works well as a birthday message.
Caption idea
“When the group chat finally drops a track.”
Watch the exampleGood for an anniversary post. Pick photos with the outfits you want in the video, because clothes usually carry over: see the red jacket and white shirt in our couple example.
Caption idea
“Our anniversary photos took an unexpected turn.”
Watch the exampleSend grandma and grandpa rapping to the family chat. Recent, clear photos work best, so everyone can tell who it is.
Caption idea
“Grandma took the first verse.”
Watch the exampleA dog and a cat, or two dogs, work too. Pets end up standing and gesturing like people, so it looks more cartoonish than real.
Caption idea
“They have opinions about dinner being late.”
Watch the exampleUse it for a launch post or an end-of-project thank-you. Ask both people first and use a separate portrait of each.
Caption idea
“The launch meeting got a new opening act.”
Watch the exampleOur example pairs an astronaut with a knight. Original characters or your own costume photos work as long as the face is visible.
Caption idea
“Nobody asked for this crossover.”
Watch the exampleUse two clear photos with one person in each. They can be from different days. Clothes in the photo usually show up in the video.
Crop out other people and pets. If a photo shows two faces, the model may pick the wrong one or blend them.
Front or three-quarter views with both eyes visible keep identity best. Avoid sunglasses, hands over the face and heavy shadows.
A selfie or waist-up shot works better than a distant full-body photo. Daylight or even indoor light beats a dark night photo.
For animals, pick a photo where the whole head is visible and the pet is looking toward the camera. Distinctive markings help the model keep them recognizable.
Small, blurry or side-on faces give the model too little to work with.
Try next
Use a sharper, closer photo of that person and generate again.
Photo 1 always becomes the left performer and photo 2 the right.
Try next
Press the swap button between the two photos before generating.
Fast gestures and objects held in the hands can produce distortions.
Try next
Use a photo without handheld objects, or generate again; results vary between runs.
The reference is a human performance, so the model may give pets upright poses and human-like limbs.
Try next
Choose a face-forward photo and enjoy it as a stylized performance. The pet examples above show what to expect.
Every video comes with the original audio of the routine, and the download keeps it. Previews on this page start muted; turn up the volume to listen.
Check your social app's music rules before you post. You can also swap in a track from its licensed music library.
Every video costs 200 credits: the full 15-second routine in 720p, in any frame.
The price is shown on the Generate button before you start. If a render fails, the credits are returned automatically. See plans and credit packs.
Photos, credits, music and getting faces right.
It is a photo-to-video trend named after the song "Hotel Lobby". Two people, pets or characters stand in a bright orange studio with one microphone hanging between them and trade rap verses with big hand gestures and head nods. Despite the name, the scene is a music studio, not a hotel.
Upload one photo per performer. Froging AI sends both photos and a built-in orange-studio performance clip to Seedance 2.0. Your photos decide who performs; the reference clip decides the gestures, timing and framing. No prompt or video upload is needed.
Two: one photo per performer. Photo 1 becomes the performer on the left and photo 2 the performer on the right. Clear, front-facing photos with one subject each work best. You can swap the two before generating.
Every video costs 200 credits: the full 15-second routine in 720p, in any frame. The Generate button always shows the price before you start, and failed renders are refunded automatically.
Yes. The original audio of the routine is added to your video from the first second, and the download keeps it. Check your social app's music rules before you post.
You can try friends, couples, siblings, grandparents, pets or original characters. See the examples for generated results using each kind of pairing. For animals, choose a photo with the whole head visible and the pet facing the camera. Pets end up standing and gesturing like people, so it looks more cartoonish than real.
Only use photos of yourself, people who have agreed, or characters you own. Requests that impersonate real public figures or use someone else's photo without permission may be blocked by moderation, and you are responsible for how you share the result.
Use sharp, well-lit photos where the face is large and unobstructed: no sunglasses, hats over the eyes or heavy filters. Crop out other people.
Separate photos are what it's built for. They don't need the same background or location. Put one person in each slot. If you only have a group photo, crop it into two individual portraits before uploading so each role has one clear face.
Every video is the full 15-second routine in 720p. Choose vertical 9:16 for TikTok, Reels and Shorts, square 1:1 for feeds, or widescreen 16:9 for YouTube.
Usually a few minutes; busy queues can make it longer. The page shows whether your video is queued or rendering, and finished videos also appear in History, so there's no need to submit the same pair again.
Previews start muted so several videos don't play over each other. Turn on the volume in the player. The example videos include audio, and downloads keep the audio track.
Yes. Choose the 9:16 frame before you create, and the video is composed for vertical screens. Use photos you have permission to share.