Low-poly 3D floating diorama
A clean low-poly 3D render of a tiny floating island scene on a plain background. Paste this prompt into ChatGPT, Gemini or Grok, change the parts in brackets to suit you, and generate.
- Written for
- ChatGPT, Gemini, Grok
- Status
- Tested 27 Sep 2026 in Gemini app (web), Flash mode with built-in image generation
- Test settings
- Default copy text pasted unchanged into a new Gemini chat. No reference image. First result used; only converted to WebP.
- Variability
- Layout, colours and small details change between runs. Gemini adds a small sparkle watermark in the corner.
Scene: a small floating chunk of land shaped like a rough cube, with a tiny lighthouse, two pine trees and a winding path on top, rocks and roots hanging underneath. Geometry: low-poly faceted modelling, flat-shaded triangles, visible crisp edges. Colour: limited palette of coral, teal and cream, soft pastel background gradient. Light: soft global illumination with one warm key light, gentle ambient occlusion. Camera: isometric three-quarter view, centred subject, generous empty space around it. Style: polished 3D illustration render, calm and minimal. Format: square 1:1 Avoid: smooth high-poly surfaces, realistic textures, text, logos, cluttered background, watermarks.
Customise
For example: a tiny lighthouse, two pine trees and a winding path.
For example: coral, teal and cream.
Fields start with illustrative values; replace them with your own facts. The prompt updates as you type. Copy prompt copies your version.
How to use it
- Open an external image tool named above. Check Status for the provider actually tested.
- Paste the prompt. Replace anything in [BRACKETS] first, or use Customise.
- Send, then ask for small changes in follow-up messages.
How to use this prompt
- This prompt makes a new picture. Check the AI app named above and the real test recorded in Status.
- Fill Customise with what you want, or replace [BRACKETS] with the right details. Press Copy prompt and paste it into the image app you chose.
- When the picture is ready, check objects, hands, faces, writing and any detail that matters. Ask for one clear correction if something is missing.
Good to know
- Some models add smooth surfaces; repeat 'flat-shaded low-poly facets' if the render looks too realistic.




