
A miniature diorama shows a real place as a tiny handmade model. This guide gives you six AI prompts: a felt-craft image prompt, four prompts for other materials and moods (clay, paper, night and snow globe), and a motion prompt that turns any of them into a short video.
You can use tools such as Google Gemini, Runway and Pika. Features and limits vary by plan, so check what yours allows.
Step 1: Create the Image
Create a vertical 9:16 image, the usual shape for Reels, Shorts and TikTok. Replace {FAMOUS_LOCATION}, {CITY_NAME} and {COUNTRY} with your place.
Miniature Diorama Image Prompt
Replace the highlighted text with your own details before you run the prompt.
Check the buildings and figures before moving on, because any flaws will carry over into the video.
Step 2: Animate the Scene
Upload your finished image and paste this motion prompt. Starting from an image gives you more control than generating the whole video from text.
Miniature Felt Diorama Motion Prompt
Replace the highlighted text with your own details before you run the prompt.
Watch for warped buildings or figures in the result. If you see them, ask for slower camera movement and try again.
Try Other Materials
The Step 1 prompt makes a felt-craft model. Swap it for one of these four to change the material or the mood. Each uses the same three placeholders, and each image works with the Step 2 prompt if you want to animate it.

Clay Miniature Diorama
Example: Eiffel Tower, Paris, France. Image made with Gemini (Nano Banana) from this prompt. Replace the highlighted text with your own place.

Paper-Craft Diorama
Example: Taj Mahal, Agra, India. Image made with Gemini (Nano Banana) from this prompt. Replace the highlighted text with your own place.

Night-Time Felt Diorama
Example: Charles Bridge, Prague, Czech Republic. Image made with Gemini (Nano Banana) from this prompt. Replace the highlighted text with your own place.

Snow Globe Diorama
Example: Big Ben, London, United Kingdom. Image made with Gemini (Nano Banana) from this prompt. Replace the highlighted text with your own place.
Why Two Steps Work Better
The image step lets you get the layout, light and details right first. The video step then only has to add movement. It also saves effort, since you can fix a weak image before spending time or credits on video.
Locations to Try
Recognisable places work best, because people can identify them even at a tiny scale. Some good options:
- Eiffel Tower, Paris, France
- Times Square, New York City, USA
- Shibuya Crossing, Tokyo, Japan
- Colosseum, Rome, Italy
- Burj Khalifa, Dubai, UAE
- Marine Drive, Mumbai, India
More in AI Video Prompts and Workflows
This guide is part of our AI Video Prompts and Workflows hub, where you’ll find every related guide, along with tips and common questions.


