Three Free Motion and 3D Libraries for Your Videos

Three free libraries make a plain video look designed: one gives you the exact words for camera moves, one gives you looping backgrounds, and one gives you more than 1,300 ready-made 3D models. Here's how to use each one, step by step.

What this helps you do

• Motion Prompt gives you 192 named moves, each with a moving example and a ready-made prompt. Use it when "make it cinematic" keeps giving you the wrong shot.
• Scenario Motion Loops gives you 80 looping animations, 40 backgrounds and 40 shape animations. Put one behind a title or between shots.
• Scenario 3D Collection gives you 1,342 AI-made 3D models you can download and use. Skip modeling from scratch.

"Cinematic" doesn't tell an AI which way the camera moves, how fast, or what stays in frame. Named moves do. Loops and 3D models add movement and depth without filming anything new.

Before you start

• A web browser. All three libraries open in any browser. No account needed to browse or download.
• An AI video or image tool for the camera-move prompts, such as Veo, or a coding agent like Claude Code or Codex.
• A screen recorder if you want to record a loop. On a Mac, press Shift, Command, and 5.
• Optional, for the agent routes: Node.js 18 or newer. The Motion Loops skill also needs ffmpeg and Playwright to render video files.
• Optional, for 3D: Blender (free) or any editor that opens GLB files.

Terms to know

• Camera move: how the camera travels in a shot. A dolly in moves the camera toward the subject. A zoom only magnifies with the lens.
• Loop: an animation whose last frame flows into its first, so it can repeat forever with no visible jump.
• HTML file: a web page saved as one file. Double-click it and it opens in your browser.
• Agent skill: a folder of instructions an AI coding agent like Claude Code reads to do a specific job.
• MP4: the standard video file format that every editor and phone can play.
• GLB: a common 3D file format that holds the model, its colors, and its textures in one file.
• License: the rules for how you can use something. CC BY 4.0 means you can use it, including commercially, as long as you give credit.

Step-by-step

Library 1: Motion Prompt (camera moves and motion names)

1. Open the gallery at showheyohtaki.github.io/motion-prompt. Click EN at the top if it opens in Japanese.
2. Browse the sets. There are camera moves and transitions, motion graphics, UI parts, landing page effects, and photo looks. Every card plays its move and shows a number, a name, what it means, and an example prompt.
3. Press Pick on the moves you want. Add a short note in the note field at the bottom, such as "for the opening shot."
4. Press Copy picks at the bottom of the screen. This copies your picks with links to their demos and specs.
5. Paste into your AI chat, then add your own subject. Or fill in the camera-move template in "Copy this" below.
6. Compare. Run one clip and put it next to the example. If it's off, change the subject or the timing until it matches.

Using Claude Code or Codex? Install the skill with npx skills add showheyohtaki/motion-prompt in Terminal. Then ask your agent to "Open the motion-prompt library," pick your moves, and press Send to AI instead of copying. Details are in the Motion Prompt README.

Library 2: Scenario Motion Loops (looping backgrounds)

1. Open the page at scenario.com/explorations/motion-loops. You'll see 40 backgrounds and 40 shape animations playing live.
2. Click a loop to open it. The backgrounds include moving grids and waves (Grid Tunnel, Wave Ridges, Halftone Wave, Single Wave), spinning rings (Gyro Rings), and dozens more such as Perspective Floor, Warp Stars, and Kaleidoscope. The shape animations include Shockwave, Spirograph, Equalizer, and Loader Dots. Dot Tunnel, number 04 in Backgrounds, has rings of dots spiraling toward you.
3. Press Download .html (or Copy code and paste it into an empty .html file).
4. Change the colors and size. Open the file in a text editor. Edit the colors in P and the size in W and H. For a vertical video, set the width to 1080 and the height to 1920.
5. Double-click the file to open it in your browser. Each loop repeats every 12 seconds with no visible seam.
6. Record it with your screen recorder and place it behind your title. Since it never stops moving, it fits a video of any length.

The no-recording route: download the Agent skill (ZIP) from the same page and unzip it into ~/.claude/skills so the folder sits at ~/.claude/skills/scenario-motion-loops. Restart Claude Code. Then ask it to turn your loop into a vertical MP4, using the loop prompt in "Copy this." The skill checks the loop seam and renders a 12-second MP4 for you.

Library 3: Scenario 3D Collection (free 3D models)

1. Open the collection on Hugging Face at huggingface.co/datasets/scenario-labs/3d-collection.
2. Browse the previews. The Dataset Viewer shows a thumbnail for each model. Scroll until you find an object that fits your video, such as a coffee cup.
3. Check the format. The formats column should say glb. Most models are GLB. A few are Blender, PLY, or other formats.
4. Download it. The output_files column has the link to output.glb. Open that link and the file downloads.
5. Open it. In Blender, choose File, then Import, then glTF 2.0 (.glb/.gltf). Or drop it into any editor that opens GLB files.
6. Use it. Try a simple explainer: a 3D coffee cup next to the words you're teaching, turning as the story changes.
7. Check the license in the license column for the model you picked. In this collection it's CC BY 4.0. Credit "Scenario" wherever you use it.

Copy this

Camera-move prompt template

Use this in any AI video tool. Take the move name and number from Motion Prompt, then fill in the rest.

[MOVE NAME, e.g. Slow dolly in] (Motion Prompt #[NUMBER]) toward [SUBJECT, e.g. a coffee cup on a wooden desk]. Move the camera, do not zoom. Speed: [slow / medium / fast], over [SECONDS] seconds. Keep [WHAT STAYS IN FRAME, e.g. the cup centered and fully visible] the whole time. Start framing: [e.g. wide, whole desk in view]. End framing: [e.g. close-up of the cup]. Style: [LIGHTING AND MOOD, e.g. soft morning window light]. Vertical 9:16, no cuts, no text.

Loop-to-MP4 prompt (for Claude Code with the Motion Loops skill)

Use the scenario-motion-loops skill. Here is the code for a loop from Scenario's Motion Loops page: [PASTE THE COPIED CODE]. Put it into the skill's loop template, set the colors to [HEX CODES], check that the loop seam passes, then render a vertical 1080x1920 MP4 named [FILE NAME].mp4. Ask before installing anything.

Things to know

• Motion Prompt is MIT licensed. Commercial use is allowed for its code, videos, and images. If you reuse an example's code, keep its first license line.
• Ask by number once you know them. "Use #027 and #043" is faster than describing the move.
• Add timing and strength to a name. "Mask reveal from below, 0.6 seconds" gets closer than "mask reveal."
• Motion Loops are drawn by code, so nothing warps or flickers and they render at any size. For paid client work, read Scenario's terms first.
• The Motion Loops skill renders horizontal by default. Ask for 1080x1920 to get vertical.
• The 3D Collection is CC BY 4.0. You can use the models commercially when you credit Scenario. A separate Scenario set made with Hunyuan models has its own licenses and needs approval to download, so stick to the main collection unless you've read those terms.
• The 3D models are AI-generated. Check yours from every side before you use it. Some have rough backs or odd details.
• Want AI to put it all together? See how to edit a short video with Claude Code.

If something goes wrong

• Motion Prompt is in Japanese: click EN at the top of the gallery.
• The AI clip doesn't match the example: add the speed, the seconds, and what stays in frame. Those three details fix most misses.
• The downloaded loop opens as text: make sure the file ends in .html, not .html.txt, then double-click it.
• The loop jumps when it repeats: you changed a timing value. Download a fresh copy and change only the colors and size.
• The Motion Loops skill won't render: it needs Node 18 or newer, ffmpeg, and Playwright. Ask Claude Code to check which one is missing before it installs anything.
• The GLB won't open in Blender: use File, Import, glTF 2.0, not File, Open.
• The model looks gray in your editor: switch the viewport to material or rendered view so the textures show.

Final check

You copied one named camera move into an AI tool and got a matching clip, you have one loop playing behind a title in your colors, and you have one GLB model open in an editor with "Scenario" credited.