/miniature: Miniature AI Media Prompt

What this creates

Turns your house into a tiny handcrafted diorama. Supply the source, bound the intended change, and state exactly what must survive.

Use when: Use for visualization of real or proposed spaces, property, landscaping, staging, or renovation concepts.
Avoid when: Avoid presenting a visualization as an as-built condition, guaranteed renovation result, survey, or code-compliant plan.
Change: Replace the bracketed variables with the actual subject, destination, composition, and production constraints.
Preserve: Keep supplied identity, product geometry, logos, labels, architecture, brand colors, and every unaffected region accurate.
Failure mode: Vague style language can overpower the subject, mutate reference details, or produce a crop that fails in the real destination.
Output decision: Choose from the actual destination before generation. Confirm the destination safe area before final generation.

Complete prompt

Edit the supplied image to create **/miniature — Turns your house into a tiny handcrafted diorama**.

OUTPUT JOB
Create a production-ready visual for [destination/use]. The main subject must read immediately at the intended display size.

INPUT
Use: [describe/attach source assets]. The image to edit, plus any exact logo/brand/product reference needed to preserve identity.

PRESERVE LOCK
Preserve the supplied subject/product identity, geometry, logos, labels, and unaffected areas exactly when they are present.
Explicitly preserve: [identity / product shape / logo / label / architecture / composition / colors / other].

COMPOSITION
[shot size and viewpoint]. Place [primary subject] at [position]. Reserve [amount/side] negative space for [copy/CTA] when needed. Avoid background elements competing with the subject.

CAMERA / LIGHT / MATERIAL
Use only details that control the result: [viewpoint/lens behavior], [light direction/softness/contrast], [surface/material/texture], [environment/time of day].

STYLE
Turns your house into a tiny handcrafted diorama. Keep the treatment coherent and physically plausible where realism is expected. Follow supplied brand/design references rather than generic AI aesthetics.

TEXT / LOGO
Use only supplied text/brand assets. Keep generated text short and verify spelling character by character. If exact typography is critical, leave clean space for final typesetting outside the generator.

OUTPUT
Choose from the actual destination before generation. Compose for the real destination and its current safe area/crop behavior.

VERIFY
Check identity, anatomy, geometry, hands, logos, label text, reflections, shadows, scale, perspective, duplicated objects, edge artifacts, and brand accuracy. Change one controlling variable at a time during refinement.

Verify before publishing

The intended subject is clear at final display size.
The preserve lock survived generation or editing.
Logos, labels, text, UI, data, and claims match supplied truth.
Anatomy, geometry, scale, reflections, shadows, and perspective are plausible where realism matters.
Video motion avoids flicker, drift, duplication, and mutated text.
The crop fits the real destination and synthetic events cannot be mistaken for documentary proof.

Universal production contract

Read the 17-part contract

# Universal AI Media Prompt Contract

Every shortcut inherits this production contract. The shortcut supplies the specific creative recipe; the contract prevents a pretty result from becoming an unusable deliverable.

## 1. Output job
What should the asset accomplish?
Where will it appear?
What should a viewer understand at thumbnail size?

## 2. Workflow
Choose intentionally:
- text-to-image;
- reference/image-to-image;
- targeted image edit;
- text-to-video;
- image-to-video;
- video edit/transformation when supported.

## 3. Input assets
List exactly what was supplied:
person/selfie · product · logo/brand board · room/property · UI screenshot · packaging · reference photo · sketch/blueprint · first frame · data/labels.

Do not pretend an unsupplied asset exists.

## 4. Preserve lock
State what must remain unchanged:
identity · face · product geometry · logo · label · brand colors · packaging text · environment · camera angle · composition · clothing · architecture · UI content · other.

## 5. Subject
Describe concrete visual facts, not empty praise.

## 6. Composition
Use only controls that matter:
shot size · viewpoint · camera height · focal position · negative space · background separation · crop · symmetry/asymmetry · depth.

## 7. Camera / optics
Use when useful:
lens/focal behavior · depth of field · distortion · shutter/motion feel · viewpoint · sensor/film character.

Camera jargon should change a visible decision.

## 8. Light
Define direction · softness · contrast · practical lights · time of day · reflections · rim/fill · shadow behavior.

## 9. Material / texture
Define finish · wear · translucency · roughness · reflectivity · grain · fabric · condensation · dust · scratches · scale.

## 10. Environment
Define location · supporting objects · weather · era · time · context.
Supporting elements must not compete with the subject.

## 11. Color / brand
Use actual supplied brand tokens/references where possible.
Do not invent a brand color from memory.

## 12. Text / logo
Quote only short critical text.
Ask for legibility and exact spelling, then verify the output character by character.
If typography must be exact, leave clean space for final typesetting outside the generator.

## 13. Style
Describe a coherent visual system:
photographic · editorial · catalog · documentary · technical · flat vector · clay · paper · 3D · watercolor · other.

Do not stack unrelated style adjectives.

## 14. Video motion
For video define:
- subject motion;
- environmental motion;
- camera motion;
- timing/speed;
- temporal progression;
- start/end state;
- loop/transition;
- audio/dialogue only when supported and needed.

For image-to-video, spend most of the text prompt on what changes over time. The image already defines the starting frame.

## 15. Output
Aspect ratio · orientation · safe zones · intended crop · resolution target · transparency where supported.

## 16. Verification
Check identity · anatomy · geometry · hands · logos · labels · generated text · materials · reflections · shadows · scale · perspective · object count · background artifacts · edge halos · brand accuracy · continuity/flicker for video.

## 17. Iteration
Change one controlling variable at a time:
composition OR lighting OR styling OR motion OR background.

Keep the preserve lock unchanged unless the user explicitly changes it.