Quick answer
Starrd's Big River Flows East template recreates the viral 大河向东流 ('the big river flows east') monkey clip from one photo of you or your pet. A tourist sings the old folk song at the wild monkeys on a lake park's railings and every one of them freaks out; your subject takes the monkey's place in the first three shots and copies the real monkeys' movements and faces: bolting along a wooden railing and creeping back with a mad, confused glare, dropping a green soda bottle and baring their teeth on a carved stone post, then ducking behind a white balustrade and peeking back over it. The original singing plays on top. 15 seconds, vertical 9:16, Seedance 2.5, 150 credits (about $7.50).
One photo in, and out comes you or your pet taking the wild monkeys' place in the viral 大河向东流 clip: someone sings the old folk song at the lake-park railings and your subject bolts, glares, drops their soda bottle in shock and ducks behind the balustrade, copying the real monkeys' moves and faces. Here it is with a puppy:
Big River Flows East
One photo of you or your pet. Someone sings 大河向东流 at the lake-park railings and you freak out exactly like the monkeys.
What You Get
- The real monkeys' performance. The bolt, the creep back, the dropped bottle, the duck and the peek all land at the same moments as in the original.
- Three startles in three shots. Every time the singing starts, your subject reacts, just like the monkeys did.
- Real reactions on the face. A mad, confused glare in the first shot (knitted brows, one cocked brow, a tight crooked frown), then wide eyes and bared teeth.
- The original places. The dark wooden lakeside railing under a pointed green peak, the carved white-stone post above a bright blue lake with red buoys, and the white balustrade in front of thick bushes.
- The original singing on top, at full volume.
- 15 seconds, vertical 9:16, ready for TikTok, Reels and Shorts.
What the Meme Actually Is
The clip was posted by @sunapereal on September 30, 2026, captioned 大河向东流 ('the big river flows east') with #monkey #lovemonkey #funny, and passed 2.2 million plays, 339,000 likes and 75,000 shares. It's a minute of phone footage from a lake park in southern China, where wild macaques sit on the boardwalk railings.
Someone off-camera sings the opening of 好汉歌 ('Song of the Heroes'), the theme of the 1998 TV series Water Margin, at each monkey in turn. The monkeys do not appreciate it. One jumps up and bolts along the railing, one drops its soda bottle and bares its teeth, one ducks behind the balustrade and peeks back over it, another runs off into the trees.
Why it suits AI: the joke is the reaction, and a reaction works on anyone. Put your dog, your cat or yourself on that railing and the same song gets the same response.
The Clip, Shot by Shot
| Time | Place | What happens |
|---|---|---|
| 0:00 | Wooden lakeside railing | Sitting on a post, back to the camera, looking out at the lake |
| 1.2s | The singing starts: a full-body jolt, then a scramble away along the railing | |
| 3s | Creeps back toward the phone and glares into the lens, mad and confused | |
| 5.6s | Carved white-stone post | Hugging a green soda bottle |
| 6.3s | The second line: jolts upright and the bottle drops | |
| 6.8s | Bares its teeth at the camera, eyes wide, then freezes | |
| 10.5s | White balustrade | Reaching toward the bushes |
| 11s | The third line: ducks down behind the balustrade, out of sight | |
| 12.3s | Pops back up, head and shoulders over the rail, and bares its teeth to the end |
Your Subject, Their Lake Park
- One photo of a person or a pet. People keep their own face, hair and clothes; animals keep their own breed, coat, ears and paws.
- People move like the monkeys. Squatting on the posts, creeping along the railing on hands and feet: that's the joke.
- Pets stay real animals. Paws, never monkey hands, and no monkey face.
- Their places, their light. Bright daylight, slightly soft handheld phone footage, exactly like the original.
Use a clear, well-lit photo where your subject's face is easy to see. The best moments are close to the lens (the glare at the end of the first shot and the bared teeth in the second), so the face carries the video. For pets, a photo with the whole animal in it works best.
The Fastest Way
Big River Flows East
One photo of you or your pet, the first three shots of the 大河向东流 monkey clip, the original singing.
One photo, 150 credits (about $7.50), no prompt. It renders on Seedance 2.5 at 480p, vertical 9:16.
Build It Yourself
1. Pick the shots with the startles. The original runs a minute; the first 15 seconds hold three clean startles, one per sung phrase, and the editor already cut in the gaps between phrases. Keep those cuts.
2. Turn the clip into a depth map, shot by shot. Run it through a depth-estimation model (Depth Anything V2 works) and set the contrast per shot. Push anything far away to black: the lake and sky drop out on their own, but the bushes behind the balustrade sit close to the monkey and need their own curve. Put the song inside the depth video so the timing travels with it.
3. Erase anything you don't want back. A second monkey leaps through the corner of the third shot. Mask it out of the depth map, or it comes back in your video as a furry blob.
4. Recreate the original frames with your subject. Make a 2×2 storyboard from four frames of the real clip and have an image model redraw it with the monkey replaced by your subject, keeping the places, the framing and the phone-video look.
Reference image 2 is a sheet of four real frames from a tourist's iPhone video at a lake park in southern China. Recreate it exactly: the same four panels in the same order, the same places, camera angles, framing, light, sky, water and slightly soft handheld phone-video quality. Make one change only: in every panel the monkey is gone, and the subject from reference image 1 is in its spot, in its pose. In the bottom-right panel the subject is behind the white balustrade, the rail between them and the camera. No monkeys anywhere.
5. Drive it with the depth map. Feed the storyboard, the photo and the depth video to a model that copies motion from a reference video (we recommend Seedance 2.5), then write a timed beat list: when each startle happens, what the body does, and what the face does, by part (brows, eyes, mouth, jaw). Lay the original song back on at the end.
Common Mistakes
Driving with the raw clip. Monkey fur, monkey faces and monkey hands come straight through, and a dog ends up with fingers. Use the depth map.
Naming a feeling instead of the face. "Angry" or "scared" gets you one flat expression held for the whole shot. Describe the parts: knitted brows, one cocked brow, a narrowed glare, a tight crooked frown; or wide eyes, high brows and bared teeth.
Letting the subject hide at the end. People and pets are a different shape from a monkey, and a vague "peeks over the rail" leaves only the eyes showing. Say the whole head and shoulders come up over the rail and stay there.
Leaving the second monkey in. It's only half a second, but it renders as a brown furry shape in the corner.
Asking the model for the music. Video models tend to cut their own music at dramatic beats, which is exactly where the startles are. Add the real song afterwards.
FAQ
How many photos? One: you or your pet.
Does it work with a cat? Yes. Any animal stays a real animal and moves like the monkeys did.
How long is it? 15 seconds, vertical 9:16.
What does it cost? 150 credits, about $7.50.
Related Reading
- How to Make the Kung Fu Dog at the Vet AI Video: another one-photo pet-or-you template where the reaction is the joke
- How to Make the 'Deep Down In My Heart' AI Video: a viral clip rebuilt from a depth map, people or pets
- How to Make the 'Sleeping in the Attic' AI Video: a viral sound with a secret reveal, people or pets
- Viral AI Video Trends 2026: every trend we've turned into a template this year
