Construction timelapse videos are blowing up on YouTube Shorts and TikTok. The idea is simple — you start with an empty piece of land and end with a finished building. Drone view, satisfying transitions, the whole thing. Millions of people watch these every day.

The good news? You can make one entirely with AI. No drone, no camera crew, no construction site. Just one prompt, a few images, and a video tool. Here is exactly how to do it.

What you need: ChatGPT or Claude to run the master prompt, Google Flow (free) to generate the 6 images, Kling AI or Veo 3.1 to animate between them. Total cost: zero.

01 — How the Video is Built

The whole video is made of 6 images and 5 animated transitions between them. Think of it like a flip book — each image is one stage of construction, and the animations show the building slowly coming to life between each stage.

The most important rule is that the camera never moves. Every single image is taken from the exact same drone angle, same height, same zoom. This is what makes the transformation feel satisfying and real.

🌿
Image 1
Empty Land
Just grass and trees. Nothing built yet.
🚜
Image 2
Ground Cleared
Machines arrive, land is cleared and dug up.
🏗️
Image 3
Foundation Done
Concrete poured, steel structure rising.
🧱
Image 4
Building Taking Shape
Floors going up, cranes and scaffolding visible.
🏢
Image 5
Building Complete
Finished exterior, clean and empty.
Image 6
Fully Alive
People, cars, lights, landscaping. Final wow shot.

02 — The Master Prompt

This is the only prompt you need. Paste it into ChatGPT or Claude. It will first give you 10 building ideas to choose from. Once you pick one, it generates all 6 image prompts and all 5 video transition prompts for you — ready to use directly in Google Flow and Kling.

Master Prompt — AI Construction Timelapse Generator
You are an advanced AI system that designs cinematic architectural production pipelines.Your job is to generate photorealistic outdoor building images and frame to video animation prompts. All outputs must show the complete construction process from raw land to finished building, from a fixed drone camera angle. Every scene must be architecturally scaled, camera consistent, and physically realistic.When the user runs this prompt, first:Present exactly 10 numbered outdoor architectural building ideas.Rules: Each option must be a complete building seen from the outside. Each option must be short and clear.Example building types: skyscraper, luxury villa, duplex house, bungalow, high rise apartment, office tower, resort villa, and similar.Then write exactly this line and wait for the user to choose: "Which building would you like? Or suggest your own:"After the user makes their selection:Confirm the selection and note that this is a full exterior drone view project designed for image to video animation, built from scratch.Then generate 6 photorealistic image prompts with this theme: All images take place on the same plot of land. The drone camera is completely fixed. The lens does not change. The height does not change. The angle does not change. The entire building fits within the frame in every image. No visual style drift.IMAGE 1 EMPTY LAND Natural grassy or bushy land. No construction. Untouched surroundings. Photorealistic in daylight.IMAGE 2 LAND PREPARATION Vegetation being cleared, same shot same angle. Bulldozers, workers, excavation machines. Soil exposed. Active site preparation. No foundation yet.IMAGE 3 FOUNDATION AND STRUCTURAL SYSTEM Concrete foundation poured, same shot same angle. Steel rebar and formwork visible. Structure begins rising from the ground. Workers actively building. Real equipment and materials.IMAGE 4 MID LEVEL CONSTRUCTION Building largely formed, same shot same angle. Floors and exterior facade visible. Cranes, scaffolding, exposed surfaces. Construction nearing completion.IMAGE 5 COMPLETED BUILDING PASSIVE Building fully finished, same shot same angle. Clean exterior facade. No interior decoration. No activity. Pure architectural presentation.IMAGE 6 ACTIVE BUILDING Same building now alive, same shot same angle. Landscaping complete. People and vehicles in scene. Exterior lighting active. Cinematic final shot.Each image prompt must include: Full production ready prompt text. Platform note for example: "Generate with Google Flow or Nano Banana"Then generate frame to video animation prompts in FRAME TO VIDEO format.GLOBAL VIDEO RULES: Camera never moves. Drone position is fixed. No sudden cuts. No jumps. No teleporting. All changes are gradual and physically realistic. Only human and machine driven movement.VIDEO 1 IMAGE 1 TO IMAGE 2 Vegetation slowly removed. Machines enter and exit naturally. Land transforms over time.VIDEO 2 IMAGE 2 TO IMAGE 3 Foundation construction begins. Concrete poured. Load bearing system rises.VIDEO 3 IMAGE 3 TO IMAGE 4 Floors built in sequence. Walls rise. Crane and scaffolding move logically.VIDEO 4 IMAGE 4 TO IMAGE 5 Final building elements completed. Exterior facade finished. Construction site cleared.VIDEO 5 IMAGE 5 TO IMAGE 6 Activation process. Landscaping added manually. Vehicles arrive. People fill the space. Exterior lights turn on naturally.Each video prompt must include: Detailed animation prompt text. Clear realism constraints.

03 — Generating the Images

Once ChatGPT gives you the 6 image prompts, open Google Flow and paste them one by one. Generate 2 to 3 versions of each image and pick the one that looks most consistent with the others. The key thing to check: does the camera angle look exactly the same across all 6? If one image looks slightly different, regenerate it.

Quick tip: Add the phrase "same drone angle as previous image, camera has not moved" to each prompt after the first one. This helps Google Flow keep the viewpoint consistent across all 6 shots.

04 — Animating the Transitions in Kling

Now take your 6 images and the 5 video prompts from ChatGPT into Kling AI. For each transition, upload Image X as the start frame and Image X+1 as the end frame, then paste the corresponding video prompt into the text box. Kling will animate the transformation between the two frames.

  • Each clip should be 4 to 6 seconds long
  • Use Standard mode in Kling for faster generation, Pro mode for higher quality final renders
  • If workers disappear or reappear strangely between frames, add "workers remain continuously visible throughout the transition" to the prompt
  • The most important transition is Video 5, the activation scene. Spend extra credits regenerating this one until the lighting and people look natural

05 — Putting It Together

Import your 6 images and 5 video clips into CapCut. The order is simple: Image 1, then Video 1, then Image 2, then Video 2, and so on until Image 6. Add a subtle construction ambiance soundtrack from Pixabay. Export at 1080x1920 for Shorts.

  • Total video length should be 30 to 50 seconds
  • Use Image 1 as your thumbnail — the empty land creates curiosity
  • Title format: "We built a [building type] from scratch 🏗️"
  • Pin a comment: "What should we build next? 👇"
  • The best performing building types are luxury villas, glass skyscrapers, and waterfront resorts
  • Post 3 to 5 times per week — this niche rewards volume

Try It Yourself — Free

Copy the master prompt above, paste it into ChatGPT, and pick your building. Your full production package will be ready in seconds.

Open Google Flow Browse All Prompts →