Earth Zoom Out AI

Earth zoom out AI from one photo of you: the camera rises from a park or a rooftop through the city and clouds to the whole Earth. Vertical, 63 credits.

Earth Zoom Out AI turns one photo of you into a camera pull-back: you wave on the grass or on a rooftop, the camera rises over the city, the coastline and the clouds, and the clip ends on the whole Earth floating in space. MiniMax H3 Max draws the whole fall away from one prompt, and the scenes ask for a rising wind sound. The globe is a drawing, not a map. The park scene needs 10 seconds, which is 125 credits; the rooftop scene works at 5 seconds, which is 63 credits, quoted before you generate.

  • From 63 credits per clip
  • From your photos
  • 5s / 10s / 15s
  • 480P
  • Audio available

Loading your account…

Model

MiniMax H3 Max

Browse by task

One clear photo of yourself or an adult who agreed to appear, facing the camera, ideally full-length. You only appear in the opening seconds, so a sharp face and clothes matter most. JPG, PNG or WebP, up to 10 MB each.

What scene would you like to create?

Choose a scene. The prompt is fixed and cannot be edited.

Prompt672/2000

This page runs only the scenes above, so the prompt cannot be edited.

Actual output: 5s, 9:16, 480P. Writing “4K” in the prompt does not change this resolution. Faces and details may change; successful results still cost credits if you dislike them.

More settings · 9:16 · 5s · Audio
Aspect ratio
Duration
Resolution
Audio

Checking generation and private storage configuration.

Billing and delivery

Generate charges the displayed quote for the current length and resolution. If generation fails or the file fails our check, the credits come back automatically; a finished clip you don’t like is still charged. Finished clips stay in My videos.

Estimated cost: 63 credits
Sign in to generate

Sample result

10s · MP4
Park to Earth: ten seconds from the grass to space.Made on Cicadas · MiniMax H3 Max · 10s
Made from these photos (AI-made, fictional)
  • Your photoYour photo
Use this recipe

About Earth zoom out AI on Cicadas

The earth zoom out effect starts on a person and pulls the camera straight up and away until the whole planet is in frame. On this page one photo of you is a reference, and MiniMax H3 Max draws the rise from a new park or rooftop through the streets and clouds to a globe in black space, in one shot with no cuts. Pick the scene, add one clear photo of yourself, and choose 5 or 10 seconds. It uses no map data, so the continents are invented.

To start from a still image rather than a person, use image to video; to turn one picture into another, the AI morph video. For a change of clothes on a spin, see the AI outfit change video, and to write your own scene around your photo, AI reference to video. The MiniMax H3 page lists what the model behind this page costs per second. The full November round, including the bicep heart pose we cut, is in AI video trends, November 2026: we tested seven from one photo.

Earth zoom out AI examples made on Cicadas

From your photos10s · 480P

Park to Earth: ten seconds from the grass to space.

MiniMax H3 Max reference to video, 10 seconds at 480P with sound, vertical, from the site's sample photo of a woman. She waves on the grass of a park, the camera rises over the park, the city blocks, a coastline and clouds, and the clip ends on the whole globe in space. It differs from the prompt in two ways: the park turns into a different square between two moments instead of one unbroken pull-back, and the continents on the globe are not real geography. The sound was not listened to.

Made from these photos (AI-made, fictional)
  • Your photoYour photo
Use this recipe ↗
From your photos5s · 480P

Rooftop to Earth: five seconds, very fast.

MiniMax H3 Max reference to video, 5 seconds at 480P with sound, vertical, from the site's sample photo of a man. He waves on a rooftop at sunset; the camera passes the rooftop, the streets, a headland and clouds, and ends on the globe. It differs from the prompt in two ways: five seconds makes the middle a blur of streets, and the continents on the globe are approximate. The sound was not listened to.

Made from these photos (AI-made, fictional)
  • Your photoYour photo
Use this recipe ↗

Why use Cicadas for Earth zoom out AI

One shot from you to the planet

Both scenes start on you waving, then the camera rises faster and faster: park or rooftop, city blocks, a coastline, thin clouds, the curve of the planet and the whole Earth in space with stars. The scene asks for no cuts and the clip is made in one pass.

The globe is drawn, not mapped

No map or satellite data goes into the clip. In both of our clips the globe is a convincing blue planet whose continents are approximate, not accurate geography, so don't use it where the real place matters. For that, use a map tool.

10 seconds or 5

The park scene is written for 10 seconds, because the pull-back has five steps to pass through. At 5 seconds the rooftop scene still reaches the globe, but the middle becomes a blur of streets. If you have the budget, choose 10 seconds for either scene; 10 seconds costs 125 credits, while the default 5 seconds costs 63.

Price before you generate

The default 5 seconds at 480P is 63 credits, and 10 seconds, which the park scene is written for, is 125; 15 seconds is 188. At 768P they are 100, 200 and 300. The amount appears on the button before you generate, and a job that fails is refunded.

How to use Earth zoom out AI

  1. 01

    Sign in and pick a scene: Park to Earth (written for 10 seconds, 125 credits) or Rooftop to Earth (works at 5 seconds, 63 credits).

  2. 02

    Add one clear photo of yourself, set 5 or 10 seconds to match the scene, and check the quote.

  3. 03

    Tick the photo confirmation, generate, and download the MP4 from My videos; add your own caption in your editor.

Park scene or rooftop scene

The park scene has you on a lawn on a sunny day and passes the whole park, the city blocks, a coastline and clouds on the way up. The rooftop scene puts you on a flat roof at sunset and moves faster. Pick the park for a calmer rise and the rooftop for a shorter, more dramatic one.

Where the effect breaks

Longer pull-backs ask the model to keep one place consistent while the scale changes by orders of magnitude. In our park clip the park became a different square between two moments, and in the 5-second rooftop clip the middle was a blur. You only appear at the start, so check the first second against your photo.

A photo for the first seconds

A sharp, front-facing photo in even light, with your face and clothes clearly visible and nobody else in the frame. The camera leaves you within seconds, so the opening frames are your one chance to look like yourself.

Earth zoom out AI prompt ideas

These are the tested scenes on this page. Each scene is fixed and its prompt is read-only: pick one above, or press “Use this prompt”.

  • “Vertical, one continuous shot with no cuts. It starts as a medium shot of the person from Image 1 standing on the grass in a city park on a sunny day, looking up at the camera and waving. The camera then rises straight up and away from them, faster and faster: first the whole park with its paths and trees is seen from above, then the surrounding city blocks, then the coastline and thin clouds, then the curve of the planet, and it ends on the whole Earth floating in black space with stars. Keep their face, hair and clothes exactly as in the photo at the start. Sound: a rising whoosh of wind that fades to silence in space, no music. No text, no logos, no map labels.”

  • “Vertical, one continuous shot with no cuts. It starts as a medium shot of the person from Image 1 standing on a flat city rooftop at sunset, looking up at the camera with one arm raised. The camera then rises straight up and away from them very fast: the rooftop shrinks among the city blocks, the city becomes a patch on the coastline under thin clouds, then the curve of the planet appears, and it ends on the whole Earth floating in black space with stars. Keep their face, hair and clothes exactly as in the photo at the start. Sound: a rising whoosh of wind that fades to silence in space, no music. No text, no logos, no map labels.”

Earth zoom out AI FAQ

Tested on Cicadas · reviewed 2026-11-10

Local implementation, 2026-11-10. Two locked MiniMax H3 Max reference-to-video scenes from one photo, recorded from the site's sample woman (park, 10 seconds, USD 0.50) and sample man (rooftop, 5 seconds, USD 0.25), 480P, vertical, sound on (docs/research/claude-trends-20261110/). Park clip: the park turns into a different square between two moments; both globes have approximate continents; the 5-second rooftop middle is a blur. Not run: 15 seconds, 768P. The audio was measured, not listened to. Uploads of children or public figures are ruled out by the consent tick on the page.

Updated Nov 10, 2026 · How we test

Related guides