proofof.video

Contributing to Proof of Video

Thank you for helping build the most comprehensive side-by-side comparison of AI/LLM-generated interactive, explorable worlds (2D and 3D)! 🎬

What We're Looking For

  • Videos of interactive, explorable worlds or scenes (2D and 3D) generated by an AI/LLM pipeline
  • Each submission must clearly state the exact model name + version and software + version used
  • Videos should be uploaded to YouTube (unlisted is fine) and embedded via the YouTube ID
  • The prompt or scene description must be included so others can reproduce it
Not in scope: Linear video or film output (e.g. text-to-video models that produce flat MP4s without an underlying explorable world). This project is about interactive, explorable worlds — not video generation.

How to Submit a Video

1

Fork the Repository

Click the Fork button on GitHub and clone your fork locally.

2

Upload Your Video to YouTube

Title format: [Tool Name] [Model] — [Short Prompt Summary]
Description: include the full prompt, model name+version, software name+version, and date.
Copy the video ID from the URL: https://www.youtube.com/watch?v=VIDEO_ID

3

Add an Entry .md File

Create a new file in the _entries/ folder. Name it <tool>-<scene>-<model>{-postfix}.md (e.g. localgpt-school-opus.md).

4

Open a Pull Request

Branch name: add/your-scene-name
PR title: Add: [Tool] — [Prompt Summary]
In the description, briefly describe the model and software you used.

Entry Format

Fill in the following YAML front matter in your .md file: 


---
prompt_id: school
prompt_title: "School Scene"
date: "2025-02-15"
prompts:
  - "use basic primitives to design a school"
models:
  - { name: "Claude", version: "Opus 4.5" }
software:
  - { name: "LocalGPT Gen", version: "v0.2.0" }
cost_usd: 0.5
youtube_id: "cMCGW7eMUNE"
description: "LLM-driven scene generation in Rust/Bevy using Claude Opus."
link: "https://x.com/LocalGPT/status/2023243534090879297"
link_text: "X post"
---
Field Required Description
prompt_id ✅ Slug shared by all entries using the same prompt (e.g. school)
prompt_title ✅ Human-readable title for the comparison group
date ✅ Date of the work in YYYY-MM-DD format
prompts ✅ Array of prompts used (multi-turn conversations)
models ✅ Array of models used, each with name and version
software ✅ Array of software used, each with name and version
cost_usd ✅ Approximate model cost in USD (number, e.g. 0.15, 3.0, N/A if unknown)
youtube_id ✅ YouTube video ID (the part after v=)
description Brief description of the model, tool or result
link URL to a write-up, blog post, social media post, or reference about this work
link_text Display text for the link

Adding to an existing comparison

If a comparison group already exists (e.g. school), use the same prompt_id and copy prompt_title / prompts from an existing entry.

Starting a new comparison

Pick a new prompt_id slug and fill in prompt_title and prompts. Future contributors can add entries to your group by using the same prompt_id.

Guidelines

Rule Detail
Interactive worlds only Submissions must be of explorable worlds or scenes (2D and 3D) — not linear videos or films
Reproducible Include enough info for others to reproduce your result
Honest labeling Use the exact model version (e.g. Claude Opus, not just Claude)
No harmful content Prompts and videos must not contain offensive, hateful, or NSFW material
YouTube only Embed via YouTube for consistent playback and long-term availability

Questions?

Open an issue — we're happy to help!

Fork on GitHub View Repository