Zipline Through a Waterfall With Your Dog: The AI Video Prompt
A free GPT Image 2 and Seedance 2.5 prompt. You and your dog fly on a zipline over a jungle canyon, straight through the mist of a giant waterfall, and come out soaking wet. 12 seconds, filmed like an action-cam selfie.
Have it running in three steps
- 1
- 2Open OpenArt
- 3Paste it and generate
What you'll need
30 min- OpenArt
You hold an action camera at arm’s length. You fly on a zipline over a deep jungle canyon, and your dog hangs next to you in its own harness. You shout. Then you both fly straight through the white mist of a giant waterfall. You come out the other side soaking wet, with water drops on the lens. You ask your dog if it liked it, and the dog barks back.
You make it in two steps. First a still photo of you and your dog on the zipline, then the video that starts from that photo. You need one photo of yourself and one photo of your dog.
The whole process
- Pick one clear photo of yourself and one clear photo of your dog.
- Open OpenArt, go to image and pick GPT Image 2.
- Add both photos, paste the first prompt and make the first frame.
- Go to video and pick Seedance 2.5. Add the first frame, your photo and your dog photo.
- Set 12 seconds and 9:16, paste the video prompt and press Generate.
What happens in 12 seconds
This is what happened in our video. The model started the mist early, so we are already in the white spray in the first second.
| Time | What happens |
|---|---|
| 0–2s | You fly inside the waterfall mist and shout. Everything around you is white |
| 2–4s | The spray gets thicker and covers the whole frame for a second |
| 4–6s | You come out soaking wet, with drops on the lens. The waterfall and the canyon open up behind you |
| 6–9s | You laugh and turn to the dog: “Did you like it, Jack?” The dog barks back, tongue out |
| 9–12s | You raise your fist and both of you look into the camera. “Yeah! Let’s go!” |
Make the first frame
Open OpenArt, go to image and pick GPT Image 2. Add two references: your photo first, then your dog photo. Set the shape to 9:16.
The prompt below is written for a woman and a German Shepherd, because that is what we made. For a different dog, change “German Shepherd, about 35 kg” to your breed and weight. For a small dog, you can drop the size sentence.
Ultra-realistic action-camera selfie, vertical 9:16, shot on a GoPro-style ultra-wide lens (14mm equivalent, slight fisheye curve at the edges), camera held at arm’s length slightly below face level.
@Image1 defines the woman: same face, hair colour, skin and proportions as the reference. Use only the woman from @Image1, not her background or clothes. @Image2 defines the dog, Jack: same breed, fur colour, markings, face and ears as the reference. Use only the dog from @Image2.
The woman is riding a high-speed zipline over a deep jungle canyon. She wears a bright lime-green climbing helmet (vivid yellow-green, like a fresh lime, matte finish) with a black chin strap, a black climbing harness over a fitted black athletic t-shirt, and holds the camera toward herself with one extended arm. Her mouth is wide open in a thrilled shout, eyes bright, long hair blowing back in the wind.
Jack is a full-size adult German Shepherd, about 35 kg, shown at his true size: his head is level with the woman’s head, his body is as long as her torso, his chest is wider than her shoulders. He hangs beside her, to the right, in his own black full-body dog climbing harness with a padded chest plate and belly straps, clipped by a short strap and a steel carabiner to the same zipline pulley. His legs hang down relaxed. He wears small dark tinted dog goggles. His ears and fur blow back hard in the wind, mouth open in a happy pant, tongue out.
Both heads are fully inside the frame. Above them: the zipline cable and the steel pulley trolley, straps running up to it. Behind them: a sheer canyon of dark red-terracotta rock walls, emerald rainforest far below, and ahead a giant waterfall throwing a huge white cloud of mist into the air.
Soft overcast daylight, no harsh shadows on the face. Colours: deep jungle green, terracotta rock, white mist, bright lime-green helmet. Wind motion in the hair, straps and fur. Real skin texture, individual hair and fur strands. No text, no logos, no watermark.
This is the frame we picked:
Make three or four and pick the most real one. Check three things. The dog is its real size. Both heads are fully in the frame. The straps from both of you go up to the same pulley. If the dog sits on your hip or the straps split in two, generate again. The video copies everything in this frame.
Make the video
In OpenArt, go to video and pick Seedance 2.5. Add three references in this order: the first frame, your photo, your dog photo. Set the length to 12 seconds and the shape to 9:16.
Paste the prompt below. Change “Jack” to your dog’s name. That is the only word you need to change.
[LOOK — LOCKED FOR THE WHOLE CLIP]
Photoreal action-camera footage, one continuous selfie shot, no cuts. Ultra-wide GoPro-style lens with slight fisheye curve, camera held at arm’s length by the woman, shaking lightly with the speed of the zipline. Soft overcast daylight. Colours: deep jungle green, terracotta-red rock, white water mist, bright lime-green helmet. Strong wind physics on hair, straps, clothes and fur.
@Image1 is the first frame. Start exactly from it. @Image2 defines the woman: same face and hair in every moment. @Image3 defines the dog, Jack: same breed, fur and markings in every moment. One woman, one dog.
[Generation Goal]
A 12-second vertical selfie video: the woman and her German Shepherd zipline through the mist of a giant waterfall and burst out into a huge canyon view.
0-2s: They race along the cable. The woman shouts “Wooo!” into the camera, mouth wide open. The German Shepherd beside her has his ears flapping hard in the wind, tongue out, legs hanging relaxed in his harness.
2-4s: The cable carries them toward the waterfall on the right. A thick wall of white spray from the falls sweeps across them and fully covers the frame for about one second, water hits the lens, the image turns white and blurred.
4-7s: They burst out of the mist. Both are soaking wet: water drips from the lime-green helmet, her t-shirt and hair are dark with water, the dog’s fur is wet and spiky, drops sit on the lens. Behind them opens the full view: the wide multi-step waterfall pouring into the gorge, a white river running far below through emerald jungle. The woman laughs, turns her head to the dog and says “Did you like it, Jack?”
7-9s: The dog turns his head to her, gives one happy loud bark “Woof!”, tongue out, ears pressed back by the wind but no fear.
9-12s: The woman turns back to the camera, raises her free fist and shouts “Yeah! Let’s go!”. They keep flying along the cable over the canyon.
[Maintain Consistency]
The dog is a full-size adult German Shepherd and keeps his true size in every moment. He hangs beside her, to the right, in his own black harness clipped to the same pulley, the whole time, and never leaves it. The dog wears the goggles the whole time. The helmet stays lime green. The camera stays in her hand, the selfie framing never changes to a third-person view. The dog does not speak, it only barks. No text or logos in frame.
Audio: (loud rushing wind and the whirr of the zipline pulley) <roar of the waterfall rising, then muffled splash as they pass through the mist> <woman shouting “Wooo!”> <woman laughing, “Did you like it, Jack?”> <one happy dog bark> <woman shouting “Yeah! Let’s go!”>.
Why this prompt works
The first frame is already made. The photo from step 1 fixes you, the dog, the harness and the canyon. The video model does not have to invent any of it. It only has to make it move.
The mist is the cut. The whole video is one shot. When the spray covers the lens for a second, it works like a jump cut. Before it you are dry, after it you are wet, and the view is suddenly huge.
Your photos come back in the video. The first frame shows the scene, but faces can drift in 12 seconds. So your photo and your dog photo go in again as @Image2 and @Image3, to hold the faces.
The dog only barks. “The dog does not speak, it only barks.” Without this line, the model sometimes gives the dog a human voice line.
The waterfall is where the frame puts it. In our frame the waterfall is on the right and has many steps. The video prompt says the same thing. If the prompt and the frame disagree, the model can swap the whole background.
Make it your own
- A man instead of a woman. Change “woman” to “man” and “her” to “his” in both prompts.
- A small dog. Put it in a chest carrier: “rides strapped to her chest in a padded dog carrier harness clipped to her climbing harness”. Then change “hangs beside her” to “rides on her chest” in the video prompt.
- A cat. Change “dog” to “cat” and the bark to “one loud meow”.
- Another helmet colour. Change “lime green” in both prompts. Orange and yellow stand out best against the jungle.
Where it goes wrong
| What you got | The fix |
|---|---|
| The dog is too small | Write the breed and the weight, and say his head is level with yours |
| The dog sits on your hip | Keep “hangs beside her, clipped to the same pulley” and generate the frame again |
| The dog changes size in the video | Keep “keeps his true size in every moment” |
| The camera jumps to a view from the side | Keep “the selfie framing never changes to a third-person view” |
| The dog talks | Keep “the dog does not speak, it only barks” |
| The background changes after the mist | Describe the waterfall the way it looks in your first frame |
| Your face changes | Add your photo again as @Image2 in the video step |
| The video starts inside the mist | This happened to us. It still works, because you come out of the mist wet. For a dry start, generate again |
Ready? Open OpenArt, add your photos and paste the first prompt.
Why this workflow runs here
Many models, one account
Free credits every day
Compare before you commit
Make yours in the next 30 min
You have the whole prompt and every setting. The only thing left is an account, and that takes a minute.
- Free to start
- Prompts ready to paste
- Upgrade only if you want more
Referral link — if you subscribe I earn a commission, at no extra cost to you. More on that.
Related prompts
Effect Free 24 GTR AI Prompt: Swap the Car for Your Dragon
A 24 GTR AI prompt for the AIReel effect. You flash the UNO cards, the scene cuts to a stone mountain, and a dragon stands behind you instead of a car.
Effect Free Fan Blade Outfit Change: The Top-Down AI Prompt
A free Seedance 2.5 prompt. She lies on a rug, a slow ceiling fan blade sweeps over her, and her outfit is new. Five looks, 20 seconds, five reference images.
Effect Free Fit Check Hit by a Car: The AI Video Prompt
A free Seedance 2.5 prompt. She films a street fit check on her phone, shows the outfit from every side, and a car takes her out of frame. 12 seconds, one reference image.