Pika H3 Flex Prompt Guide: How to Write Prompts for Video and Sound (with Examples)

Pika H3 Flex generates picture and sound in one pass, so a good prompt describes both. This guide gives you a repeatable structure, tips for image-to-video, settings advice and a set of example prompts you can paste straight into the generator.

The five-part prompt structure

Write prompts in this order — it mirrors how a director would brief a shot:

1. Subject — who or what, with one or two defining details. “A golden retriever with a red bandana”
2. Action — one clear thing that happens. “steals a slice of birthday cake off the counter”
3. Setting and light — where and when. “bright 90s sitcom kitchen, flat studio lighting”
4. Camera — shot size and movement. “static wide shot”, “slow dolly in”, “handheld orbit”
5. Sound — start with the word Sound: and list what we hear. “Sound: studio audience laughter, plate clink, upbeat sitcom sting.”

Put together:

A golden retriever with a red bandana steals a slice of birthday cake off the
counter while the family isn’t looking. Bright 90s sitcom kitchen, flat studio
lighting, static wide shot. Sound: studio audience laughter, plate clink,
upbeat sitcom sting.

Tips that make the biggest difference

• One action per clip. Five to fifteen seconds is short. Two or three events can work in 15 s; five events usually turn to mush.
• Name the camera move. Pika H3 Flex follows camera language reasonably well, but ComfyUI’s docs call camera moves advisory — keep them simple and consistent with the action.
• Always describe sound. If you leave audio out, the model guesses. Specific sounds (“rain on a tin roof”, “muffled bass through a wall”) beat generic ones (“ambient noise”).
• Dialogue in quotes. For speech, write the exact line in quotes and say who speaks: He looks into the lens and says “The sea always gives back what it takes.”
• Avoid fine contact. Hands tightly gripping small objects and detailed physical interactions are a known weak spot; frame around them.
• Use the prompt enhancer for short ideas. Our generator’s Prompt enhancer (Pika’s rewrite option) expands short prompts. Turn it off when you have written a detailed prompt and want it followed literally.

Image-to-video prompts

When you upload a start frame, the image already defines subject, setting and style — so the prompt should focus on motion, camera and sound:

She turns from the window toward the camera and smiles, hair moving in a light
breeze, slow push-in. Sound: soft rain, distant thunder, quiet piano.

Add an end frame to control where the shot lands — great for product reveals, transformations and matching cuts between two shots. Keep both frames the same aspect ratio as the output.

Choosing settings

Draft cheaply at 480p, then re-render the winner at a higher resolution. Higher resolutions are upscaled from a 768p generation, so judge motion and composition at 768p first.

12 example prompts

Cinematic

• Night street in a rain-soaked city, neon signs reflecting in puddles, a woman in a yellow raincoat walks toward camera, slow dolly back, shallow depth of field. Sound: steady rain, distant traffic, soft synth pad.
• Close-up of an astronaut’s visor reflecting a red planet as the ship shakes on re-entry, handheld micro-shake. Sound: muffled radio chatter, rumbling hull, rising alarm.

Product and ads

• A matte black wireless earbud case opens on a wet stone pedestal, droplets bounce in slow motion, studio rim light, macro orbit. Sound: crisp click, water drops, deep bass hit.
• Iced coffee poured over clear ice in a tall glass, condensation running down, top-down slow motion, warm morning light. Sound: ice cracking, pouring liquid, light acoustic guitar.

Social and comedy

• Vertical 9:16, a cat in sunglasses slides across a polished kitchen floor on a skateboard, low tracking shot. Sound: wheels on tile, surprised gasp, punchy beat drop.
• Bright 90s sitcom living room, a grandpa tries to take a selfie with a flip phone and the flash goes off in his eyes, static wide. Sound: audience laughter, camera shutter, sitcom sting.

Nature and travel

• Drone rises over a misty pine ridge at sunrise to reveal a turquoise alpine lake, smooth slow ascent. Sound: wind, distant birds, swelling orchestral strings.
• Waves roll into a black-sand beach at golden hour, low angle at water level, foam washing past the lens. Sound: crashing surf, hiss of retreating water, gulls.

Music and dance

• Vertical 9:16, a dancer in an orange cape hits a sharp popping routine on a rooftop at dusk, handheld camera circles her. Sound: punchy hip-hop beat, sneaker squeaks.
• A jazz trio plays in a smoky basement club, slow push-in on the saxophonist, warm tungsten light. Sound: live saxophone solo, upright bass, quiet crowd murmur.

Dialogue

• Close-up of an elderly fisherman on a wooden dock, he looks into the lens and says “The sea always gives back what it takes.” Warm afternoon light. Sound: his voice, gulls, gentle waves.
• Two kids whisper under a blanket fort lit by a flashlight; the girl says “Did you hear that?” Sound: whispering, creaking floorboard, distant thunder.

Troubleshooting

Ready to try one? Pick an example, paste it into the Pika H3 Flex generator and render a 480p draft first.