How the New NotebookLM Tool for Creating “cinematic Videos” Works
NotebookLM is a more sophisticated, research-focused AI tool from Google, built on the same core models as Gemini, but with a less interactive and user-friendly interface. Like Gemini, it receives regular updates and new features, and the latest update adds “cinematic” video summaries.
One of the tasks NotebookLM is ideal for is collecting large amounts of information in a notebook (PDFs, web links, YouTube videos) and then explaining and summarizing it ( in this regard, it’s a great study tool). You may remember that a couple of years ago, the program added the ability to create realistic-sounding podcasts called “Audio Reviews” from your notebooks.
Last year, we introduced video summaries , but they were more like slideshows than mini-movies—they essentially distilled the contents of a notebook into a format that could be created in PowerPoint. However, with this new cinematic update, you’ll get video summaries that are much more animated and three-dimensional.
There are a few significant caveats: access to the new “Video Reviews” category requires a subscription to Google’s top-tier AI Ultra plan at $250 per month; furthermore, it’s only available to users over 18 and only in English. This feature may become generally available over time, but for now, access requires a relatively wealthy AI enthusiast.
Monthly AI costs are significant, but Google AI Ultra offers much more , including a YouTube Premium subscription and 30 TB of Google Drive storage for all your AI-generated images and videos. Gemini and NotebookLM offer higher usage rates for almost everything.
To test the new cinematic reviews, I created a new notebook in NotebookLM based on a single source: a paper published by Apple researchers on the “illusion of thinking” demonstrated by long-form reasoning models (LRMs) like those used by NotebookLM. It’s a massive 39-page study, exactly the kind of lengthy document one might need for AI-powered summarization.
cinematic difference
While NotebookLM has mobile apps, the web interface is easier to use. If you’re new to the tool, you can click “Create a new notebook” to get started, then tell NotebookLM your sources—whether it’s simple text you want to paste or Apple’s AI research you want to download. The app can even search the web for other relevant sources.
Once you’ve gathered the necessary information, the ” Video Overview” option will appear in the right-hand panel of the studio. You can choose between a quick overview or a more detailed explainer video, as well as select a template for the overview. There’s also a space to specify the video structure. A “Cinematic Video ” option is available for paid Google AI Ultra subscribers.
First, I commissioned a video review of Apple’s AI article in an “explainer” format. It took about 15 minutes to create, but ended up being just over six minutes long. As you can see above, it does a good job of explaining the article’s content and some of the reasons why LRMs get stuck when solving complex problems.
It’s a very static presentation, but it’s well-structured, and the illustrations are generally logical, with one or two exceptions: Claude Sonnet’s first chart has two extra lines that shouldn’t be there. It gives a decent overview of the article, but I wouldn’t rely on it 100 percent (Google places a disclaimer on Gemini and NotebookLM that the AI ”may be inaccurate”).
The cinematic-style video took over 50 minutes to create and is over seven minutes long, and you can watch it below. The upside is that it goes into more detail, and after watching it, I felt like I learned more about the topic. All the diagrams were accurately copied from the document, and some of the animations were truly helpful.
However, NotebookLM clearly struggled with some animations, such as when trying to show someone drawing on a page or stacking blocks in the Tower of Hanoi puzzle. These issues are also noticeable in other AI-generated videos , as these models don’t truly understand the real world or physics—the vast amounts of video footage they were trained on give them a good idea of where to place pixels, but not how objects should interact with each other.
Overall, I preferred the cinematic video review, although the results will obviously vary depending on the sources you use in NotebookLM—outside of academic papers, it can be a bit more creative and engaging. However, the visual glitches are somewhat distracting, and ultimately, the standard video reviews available to everyone work well enough without any improvements.