Home > AI Prompts > AI Video JSON Prompts

AI Video JSON Prompts for Structured Shot Control

These ai video json prompts put the brief in fields, not a pile of adjectives. Use them as json video prompt examples and structured video prompts, including Veo-style JSON as a format, then generate a short clip in Media.io Text to Video.

Fill the Fields, Then Generate From Structured JSON

Use Media.io Text to Video with a short structured prompt: subject, camera, action, style, and audio. Generate the shot, then change one field and try again.

1

Choose "Text to Video"

On Media.io, click Text to Video. You can paste a structured prompt here—subject, camera, action, style, audio—and generate a shot you can rewrite field by field.

2

Enter Your Prompt

Copy an ai video json prompt from this page. Keep the JSON small. Fill subject, camera, action, style, and audio instead of writing a novel in one field.

3

Generate and Download

Click Generate, then change one field and try again. Download the clip that matches the structure. If the shot ignores a field, shorten that value and generate again.

Veo-Style JSON, Camera Fields, and Structured Prompt Examples

Street Track JSON

{"subject":"woman in a wet black coat, identity locked","camera":"slow chest-height tracking shot, 35mm","action":"walks through neon puddles, glances back once","style":"photoreal night, teal magenta practicals","audio":"rain on fabric, distant train","duration":"4s","constraints":"no extra people, no readable signs"}

Product Rotate JSON

{"subject":"unbranded glass bottle on marble","camera":"locked tabletop, tiny push at the end","action":"quarter turn, highlight slides","style":"studio soft key plus rim","audio":"quiet room tone","constraints":"no readable label, no extra SKUs"}

Veo-Style Portrait JSON

{"subject":"same woman, window light, identity locked","camera":"slow dolly in, 85mm look","action":"inhale then a small smile","style":"cinematic skin, shallow depth","audio":"quiet room, faint clock","notes":"veo json prompts format example, generate in Media.io Text to Video"}

Kitchen Line JSON

{"subject":"man at a sink, morning","camera":"locked close-up","action":"turns slightly toward camera","audio":{"dialogue":"We still have time.","room":"kitchen"},"style":"photoreal, window key","constraints":"one speaker, no extra voices"}

Wildlife Tele JSON

{"subject":"red fox at a snow treeline","camera":"locked long lens","action":"pause, breath fog, one step","style":"overcast wildlife documentary","audio":"wind, distant crow","constraints":"no extra animals"}

Gym Lift JSON

{"subject":"lifter, chalk, rubber floor","camera":"side 90 locked","action":"deadlift lockout then hold","style":"fitness documentary, fluorescent","audio":"effort breath, plates unreadable","constraints":"no extra lifters"}

One-Take Hall JSON

{"subject":"same person, gray hoodie","camera":"continuous follow, walking pace, no cuts","action":"walks hall, opens back door to garden","style":"photoreal afternoon","audio":"footsteps then birds","constraints":"no teleporting rooms"}

Timelapse City JSON

{"subject":"locked skyline axis","camera":"locked wide","action":"day compresses to night, windows light","style":"photoreal city","audio":"low traffic bed","constraints":"no readable billboards, no camera wander"}

Multi-Shot JSON List

{"cast":"same woman, red coat","shots":[{"id":1,"camera":"window ECU rain"},{"id":2,"camera":"medium turn"},{"id":3,"camera":"cab step"}],"style":"wet night","audio":"rain","constraints":"identity locked"}

Architecture Walk JSON

{"subject":"concrete gallery interior","camera":"slow walkthrough, chest height","action":"advance toward a window of light","style":"architectural visualization, overcast","audio":"soft footsteps","constraints":"no furniture clutter logos"}

Surreal Office Tide JSON

{"subject":"ordinary open office","camera":"locked wide","action":"slow tide moves under desks, workers type normally","style":"photoreal except one impossible law","audio":"fluorescent hum plus water","constraints":"no extra impossibilities"}

Sports Jumper JSON

{"subject":"basketball player at the arc","camera":"sideline locked then tiny crash-in","action":"jumper, net splash","style":"photoreal hardwood","audio":"crowd as bed, no readable score","constraints":"one play"}

I2V Still Field JSON

{"first_frame":"red coat against a gray sea wall","camera":"locked then she walks right","action":"wind on coat only at first","style":"overcast coast","audio":"surf","constraints":"identity and wardrobe locked"}

Explainer Toggle JSON

{"subject":"desk padlock and a switch","camera":"close-up locked","action":"finger toggles switch, lock opens","style":"clean educational tabletop","audio":"click then quiet","constraints":"no readable UI"}

Loop Steam JSON

{"subject":"ceramic cup, table still","camera":"locked","action":"steam loops, first frame matches last","style":"cinemagraph","audio":"quiet room","constraints":"only steam moves"}

Horror Door JSON

{"subject":"same person, white shirt, dark hall","camera":"slow push","action":"listens, door moves an inch","style":"low practicals","audio":"wood creak, no score","constraints":"no extra creatures"}

Wedding First Look JSON

{"subject":"couple in a garden, identities locked","camera":"slow push to two-shot","action":"she turns, he reacts","style":"soft late light","audio":"birds, distant strings low","constraints":"no extra party, no readable programs"}

News Fictional Desk JSON

{"subject":"fictional news anchor, simulated desk, not a real network","camera":"locked medium","action":"looks to camera, short pause","audio":{"dialogue":"We go live to the scene.","room":"studio"},"style":"broadcast lighting","constraints":"no real names, no real events, no logos"}

Pika-Style Effect JSON

{"subject":"still portrait identity locked","camera":"locked","action":"one wind hit on hair and coat","style":"photoreal, punchy social","audio":"one gust","constraints":"no extra effects stack"}

Higgsfield Camera JSON

{"subject":"car on a wet street","camera":"low lateral track, one pass","action":"slow roll-by","style":"cinematic night practicals","audio":"tire hiss","constraints":"no readable plates, no extra cameras"}

Nature Forest JSON

{"subject":"moss forest path, no people","camera":"slow dolly in","action":"fog drifts, a leaf falls","style":"cinematic nature, overcast","audio":"drip and distant bird","constraints":"no extra locations"}

Miniature Kitchen JSON

{"subject":"tiny cooks on a real strawberry, macro illusion","camera":"locked macro","action":"one tiny pour of syrup","style":"photoreal tiny-world","audio":"tiny clink","constraints":"no extra giant hands"}

Historical Forge JSON

{"subject":"medieval smith, wool and iron, no magic","camera":"medium to close","action":"one hammer strike, sparks","style":"historical cinematic, fire key","audio":"hammer and fire","constraints":"period materials, no fantasy runes"}

Audio-Led Hit JSON

{"subject":"empty hallway","camera":"locked","action":"dust falls after an off-screen slam","style":"practical interior","audio":"door slam then silence","constraints":"picture hits with the sound"}

Constraints-Heavy JSON

{"subject":"same man, brown coat, identity locked","camera":"locked medium","action":"one look left, then hold","style":"overcast street","audio":"soft traffic","duration":"3s","constraints":"no extra people, no morph, no readable signs, no second camera"}

Subject, Camera, Action, Style, Audio. One Object.

A usable video prompt json template is short. If every field is a paragraph, you are back to an unstructured dump.

A structured JSON prompt template

  • subject: who or what, with identity locks.
  • camera: one move, one height, one lens note.
  • action: one beat that can finish.
  • style: light, palette, photoreal or not.
  • audio: dialogue, SFX, or ambience. One bed.
  • constraints: no extra cameras, no readable type, duration if needed.

What Veo JSON and structured-field briefs usually include

  • Json prompt for ai video work is field control, not a new model claim.
  • Veo json prompts and veo 3 json prompts are format examples here, generated in Media.io Text to Video, not a claim that this is Veo.
  • Ai json video prompts fail when camera and action share one messy string.
  • Structured ai video prompts should let you rewrite one key at a time.
  • Keep JSON valid enough to copy. Trailing commentary belongs outside the object.
  • Timing fields help: duration, hold, or a 3-second push.

Why Prompt Teams Generate From Fields Instead of Paragraphs

A camera key you can rewrite is the product. Generate from the object, then change one field without rebuilding the world.

One key, one job

Camera does not live inside subject. Split them.

Rewrite by field

If motion fails, change camera or action only.

Short values

Phrases beat nested essays.

Audio as a field

A quoted line or one SFX is enough.

AI Video JSON Prompts: Structured Fields and Templates

1. What makes good ai video json prompts?
faq faq

Short fields: subject, camera, action, style, audio. A 3-second push beats a nested screenplay.

It helps. If you paste fields as labeled lines, keep the same order.

No. They are structured examples. Generate them in Media.io Text to Video.

Yes. Paste the object into Media.io Text to Video, generate a short clip, then rewrite one field and try again.

Five to eight. More keys usually fight each other.

A short constraints field is enough. Do not dump a legal document.

How Engineers Generate From JSON Briefs

user
Chris T.

Prompt Engineer

Rewrite One Key. If camera is its own field, I can iterate. I paste that JSON into Media.io Text to Video, generate a short clip, and download the take that followed the path.

user
Mina L.

Pipeline TD

Short Values Survive. Nested essays in one key are just paragraphs again. I run the prompt in Media.io Text to Video first, then download only the test that respects duration.

user
Hugo B.

Editorial Tools

Audio Belongs in a Field. A quoted line in audio is cleaner than hiding speech in action. Media.io Text to Video is where I paste the rewrite, generate, and keep the cleaner object.

Veo, Multi-Shot, and Camera Prompts Nearby

media.io

AI Video Generator star

Easily generate videos from text or images

Generate