AIFV — AI First Video Format (.aifv)

Part of the AI First Exchange (AIFX)

AIFV (AI First Video Format) is an open, ZIP-based standard for AI-generated video. It packages video assets, prompts, scripts, generation parameters, audio tracks, subtitles, thumbnails, licensing information, and declared provenance into a single .aifv container.

AIFV is designed for transparent, inspectable, and interoperable AI-native video workflows across platforms and tools.

---

🎬 Overview

An .aifv file represents a single AI-generated video with structured context describing how the video was created, with what tools, and under what parameters.

AIFV supports documentation of:

• AI generation inputs (prompts, scripts, settings)
• Media assets (video, audio, subtitles, thumbnails)
• Declared provenance and timestamps
• Licensing and attribution context

AIFV does not guarantee reproducibility or legal authorship and does not verify metadata beyond what is recorded in the container.

---

📦 AIFV Container Structure

An .aifv file is a ZIP container with the following standardized layout:

/
manifest.json # REQUIRED: primary video manifest
video/
main.mp4 # REQUIRED: primary video (mp4/webm/mov)
audio/
voiceover.wav # OPTIONAL
music.wav # OPTIONAL
frames/
thumbnails/ # OPTIONAL: preview images
keyframes/ # OPTIONAL: extracted frames
prompts/
generation.txt # OPTIONAL: main prompt
negative.txt # OPTIONAL
script.txt # OPTIONAL: scene or narration script
settings.json # OPTIONAL: model/generator parameters
subtitles/
captions.srt # OPTIONAL: captions or transcripts
legal/
license.txt # OPTIONAL: licensing terms
terms.txt # OPTIONAL
extra/
notes.md # OPTIONAL: creator notes

yaml Copy code

The manifest.json anchors the container by defining metadata, declared AI sources, provenance, and file references.

---

🧠 Key Capabilities

✔ AI-Native Video Metadata

AIFV may record declared information such as:

• AI tool or platform identifier (e.g., Veo, Runway, Sora, Pika)
• Model or pipeline name
• Prompt and script sources
• Negative prompts (where applicable)
• Generation parameters (resolution, FPS, duration, aspect ratio)
• Multi-track audio references
• Declared creator identity and timestamps

---

✔ Unified Video Container

A single .aifv file may include:

• Final rendered video
• Prompts and scripts
• Generator settings
• Audio tracks or stems
• Subtitles or transcripts
• Thumbnails or preview frames
• Licensing and attribution metadata

This supports exchange, archiving, review, and platform ingestion without reliance on external sidecar files.

---

✔ Declared Provenance & Transparency

AIFV enables documentation of:

• Who declared authorship
• Which tools were used
• What inputs shaped the output
• When the video was created

AIFV records declared provenance, not absolute proof of origin.

---

✔ Open & Extensible Design

AIFV is:

• ZIP-based
• JSON-driven
• Language-agnostic
• Forward-compatible

Developers may extend AIFV for:

• Multi-scene or episodic video
• VFX or CGI pipelines
• Storyboards and animatics
• Multi-model or hybrid workflows

---

📑 manifest.json (Simplified Example)

{
  "aifv_version": "1.0.0",
  "id": "aifv-2025-000001",
  "title": "Sample Video",
  "creator": "CreatorName",
  "duration_seconds": 12.5,
  "resolution": "1920x1080",
  "fps": 24,
  "aspect_ratio": "16:9",
  "ai_generated": true,

  "ai_source": {
    "tool_name": "Veo",
    "model_name": "Veo-Gen",
    "prompt_source": "prompts/generation.txt",
    "script_source": "prompts/script.txt",
    "settings_source": "prompts/settings.json"
  },

  "provenance": {
    "creator_handle": "YourName",
    "creation_utc": "2025-12-01T04:22:00Z",
    "tool_chain": ["Veo", "AIFX Converter v1.0"]
  },

  "files": {
    "main_video": "video/main.mp4",
    "audio_tracks": [
      "audio/voiceover.wav",
      "audio/music.wav"
    ],
    "captions": "subtitles/captions.srt",
    "thumbnail": "frames/thumbnails/default.jpg"
  }
}

🌐 MIME Type Proposed MIME type:

bash Copy code video/aifv 🧰 Tooling (Planned) Reference implementations may include:

aifvwrap — package video assets into .aifv

aifv-validate — validate AIFV containers

aifv-core — libraries for reading/writing AIFV files

🛠 Use Cases AI filmmaking and short-form video

Storyboarding and pre-visualization

Platform ingestion of AI video

Training datasets with full metadata context

Archival of AI-native video projects

Multi-scene or episodic AI video workflows

🔄 Versioning AIFV follows semantic versioning:

1.x.x — backward-compatible enhancements

2.x.x — backward-incompatible schema changes

Unknown fields should be ignored for forward compatibility.

📬 Contributing AIFV is an open specification under the AI First Exchange (AIFX). Contributions via issues and pull requests are welcome.

📝 License Released under the MIT License.