If your Reels or Shorts feed has looked like a tiny world lately — toy-sized cities, miniature street markets, tilt-shift traffic crawling like ants — you've already seen the trend. Miniature-style AI video is one of the fastest-growing formats on Instagram and YouTube right now, and almost every viral clip starts the same way: someone typing a carefully worded prompt into an AI video tool.
What Diorama Actually Does
Diorama is a free, browser-based prompt generator. It doesn't create the video itself — no AI tool does that reliably from a vague idea. Instead, it takes something as simple as "a rainy night market in Bangkok" and expands it into a full, cinematic prompt covering:
- Camera angle and movement — dolly-ins, orbiting pans, crane rises, locked-off shots
- The tilt-shift / miniature effect itself — the exact depth-of-field language that makes an AI model render a scene as scale-model-sized
- Lighting — golden hour, neon night, soft studio light, and more
- Motion — sped-up toy-like traffic, gentle ambient movement, tiny pedestrians
- Texture and color grade — from photorealistic miniature detail to a candy-colored toy palette
- Exact duration — stated in seconds, so the AI tool knows how long the clip should run
Every generation is reshuffled, so hitting "Generate" again on the same idea gives you a genuinely different variation to try — not a copy-paste template.
Built for Instagram Reels, YouTube Shorts, and TikTok
Diorama includes a video size selector built around real publishing formats, not generic aspect ratios:
- YouTube — Landscape (16:9) and Shorts (9:16)
- Instagram — Reels/Stories (9:16), Feed Square (1:1), Feed Portrait (4:5)
- Facebook — Feed (4:5) and Stories/Reels (9:16)
- TikTok (9:16), X/Twitter (16:9), LinkedIn (1:1)
Whichever you pick, the correct aspect ratio and resolution get written directly into every generated prompt — so you're not guessing at dimensions after the video's already rendered.
One Idea, Six AI Tools
Diorama writes a universal prompt first, then adapts it into the typical style of six popular AI video generators: Gemini (Veo), Seedance, Omni, Kling, Runway, and Luma. Each has its own prompting conventions — flowing cinematic prose for Gemini, labeled structured fields for Seedance, bracketed camera tags for Kling — so you don't have to manually reformat the same idea six different ways.
Every prompt has a one-tap Share button. On mobile, it hands the prompt straight to your chosen AI app through your phone's native share sheet. On desktop, it copies the prompt and opens the platform for you, ready to paste.
Single Shot or Full Multi-Scene Sequences
For a quick 5–30 second clip, Single Shot mode gives you one continuous prompt. For a longer sequence, Multi-Scene mode breaks your idea into 2–6 timed shots — an establishing wide, a closer detail shot, a moment of action, a resolving shot — that add up to a full 15–60 second sequence, each with its own start and end timecode.
How to Use Diorama
- Type one idea in plain language, or tap 🎲 Surprise me for inspiration
- Choose Single Shot or Multi-Scene, and set your duration
- Pick the platform you're publishing to
- Select which AI tools you want prompt variations for
- Hit Generate prompts
- Copy or Share the prompt straight to your AI video app
- Generate the clip, then bring it into your editor to finish
Runs Entirely in Your Browser
Diorama needs no login, no API key, and no server. Your idea and the prompts it generates never leave your device — everything happens client-side, right in the browser tab. Use it as many times as you like, for free.
Frequently Asked Questions
Does Diorama generate the actual video?
No — Diorama only writes the prompt text. You paste that prompt into an AI video tool (Gemini, Kling, Runway, and so on), and that tool generates the actual clip.
Do I need an account or API key to use this?
No. Diorama runs entirely in your browser and needs no login, key, or server. You'll need your own account on whichever AI video platform you use to actually generate the clip.
Is my idea or generated prompt uploaded anywhere?
No. Everything runs locally in your browser — nothing is sent to a server, stored, or logged.
Why doesn't Share paste the prompt automatically on my computer?
Browser security stops any website from typing into another website. On phones, Share uses your device's native share sheet, which can hand text straight to an installed app. Desktop browsers have no equivalent, so Share copies the prompt and opens the site for you to paste.
Why doesn't Seedance open automatically like the others?
Seedance doesn't have one single official public web app — most "Seedance" sites you'll find are third-party wrappers. The Seedance button copies the prompt instead of guessing a link, so you can paste it into whichever Seedance-access app you already use and trust.
Can I edit a prompt before using it?
Yes — copy it into your AI tool's input box and adjust the wording however you like before generating.
Which video size should I pick?
Match wherever you're publishing: 9:16 for Reels, Shorts, TikTok, and Stories; 1:1 or 4:5 for feed posts; 16:9 for YouTube landscape.
Does Diorama control the exact length the AI produces?
It states your intended duration in the prompt, but the final length depends on the AI platform — some cap clips at a fixed length (e.g. 5–10s) regardless of what's requested, so check that tool's current limits.
Will regenerating give a different prompt for the same idea?
Yes — camera angle, lighting, motion, and texture are reshuffled each time, so hitting Generate again on the same idea gives you a fresh variation to try.
Is Diorama really free?
Yes. It's part of the Cutting Room toolkit — free, browser-based tools with no login or subscription required.
Try Diorama below, and if you build something with it, we'd love to see the result.
