The lecture description language isn’t public at the moment. It’s a core part of the technology we’re building the company around.

Also, if you’d like, you can share the video as a public link by clicking the share button!

Also, you’re not chatting with the transcript; you’re chatting with the video. The AI has much more than the transcript in its context. See this, for example: https://academa.ai/lectures/diffusion-models-learning-to-den...

And in the coming months, we’ll make the answers themselves real-time videos, which is kind of trivial for us at this point. We’ll ask the LLM to answer by writing code, and our software will render that code and show you the response directly as a video.

Without our system, there isn’t really an equivalent target language you can ask the LLM to write. Manim exists, but you won’t get videos like these by simply asking an LLM to generate Manim code.

Actually, the economics are quite good.

We’re not using video generation models. We ask an LLM to write the lecture as code, then render it deterministically with computer graphics and TTS.

So the main costs are the LLM call, TTS, and some cloud GPU time for rendering. That ends up being much cheaper than generating long form video directly with video models.

The one that asks questions at the beginning is Gemini Flash 3.7.

We thought: what if lecture videos were written as code and compiled into video using computer graphics and TTS?

Then LLMs could write them, and lectures could be fixed with code edits instead of video production.

On correctness: LLMs may make mistakes. But these videos are code sitting in our repository, and we can maintain them. Every report and review becomes a fix in the source, and everyone who watches after that gets the corrected lecture. Every recorded lecture on the internet is stuck with its mistakes. Ours will continuously improve.

Each video also comes with an AI chat that understands everything said and shown in the lecture.

There is a longer write-up at the bottom of the page.

Happy to answer any questions.

Comments URL: https://news.ycombinator.com/item?id=49503421

Points: 30

# Comments: 25

Comments URL: https://news.ycombinator.com/item?id=49193968

Points: 2

# Comments: 0

Computer graphics will have enormous applications because they are directly controllable by LLM-generated code. Video models are probabilistic and less suitable when precision matters. In education, for example, we need exact visuals. If an AI wants to plot y = sin(x), it should generate the precise graph through computer graphics rather than approximate it with a video model.

We've made the AI chat free up to a certain usage limit, so if you're interested, you can give it a try now!

We never intended to use 3Blue1Brown's name in a negative or misleading way. We simply wanted to communicate that we have made the Manim API created by 3Blue1Brown accessible in the browser with WebGPU, without requiring any installation.

Downloading videos is completely free. Only the AI-powered features are behind the paywall, and there's a 7-day free trial.

We'll put more pre-written examples, thank you!

We reimplemented Manim with the same Python API, but the implementation underneath is Rust, connected to Python through PyO3. The Rust code uses wgpu, so rendering happens on the GPU.

To run it in the browser, we compiled the Rust parts to WebAssembly so the PyO3 extension loads in Pyodide. In the browser, wgpu targets the WebGPU API, so animations render in real time on your GPU through the browser.

The editor is Monaco (the editor that powers VS Code) with a live preview: write your Manim code on one side, watch the animation update on the other.

There's also a built-in AI agent if you'd rather vibe-code your animations.

Comments URL: https://news.ycombinator.com/item?id=49091703

Points: 49

# Comments: 32

Our content is text, not MP4 files. So we review it with humans and AI, just like how software companies review their software. We may make mistakes, but that's okay because we maintain the content. We continuously improve it and fix issues as we find them.

It'll even become more accurate than cam-recorded lecture videos over time, because those aren't being updated.