In the world of AI it feels like a day with the constant stream of updates to AI models and services powered Large Language Models.
The latest development today was from Google DeepMind. Veo 2 is a state of the art video generation model which promises to transform video creation using artificial intelligence.
The release of Sora by OpenAI is the latest in a series of developments. After using Sora for several days, I can say that it has some limitations (including the inability to understand physics such as rolling dice).
Advanced Video Generation.
Veo 2 allows you to create high quality video clips that match the style and content provided by the user. Veo 2 can also do this at resolutions as high as 4K.
This maximum quality is important, as Sora only goes up to 1080. GenAI is the best option for serious video production. While some projects may only require 1080 quality, GenAI’s ability to produce 4K will be more valuable.
The model is able to interpret instructions, no matter how simple or complex they are, and produce videos that mimic real-world physics, as well as a variety of visual styles. Veo 2 is able to maintain high levels, realism and minimize artifacts. This allows it to produce a more professional output than previous AI video models.
This model is able to understand and execute specific camera controls that are prompted by the user, such as wide shots, POV perspective, and drone shots. This adds a cinematic quality to generated content. This feature is especially exciting for creatives who want to push the limits of traditional filmmaking, without the need for expensive physical resources. Sundar Pichai, CEO of Google and Alphabet made the announcement earlier today in an X Post, sharing a very impressive Veo 2 highlight reel.
Veo 2 is our new video model that takes into account real-world physics and movement. It can be used up to 4K resolution. Join the waitlist at VideoFX. Our new and improved Imagen 3 model also achieves SOTA results, and is coming today to 100+ countries… pic.twitter.com/FtwNKyicEc
— Sundar Pichai (@sundarpichai) December 16, 2024
Realism and Motion
One of the most remarkable features of Veo 2 is its capacity to generate videos with realistic motion. This includes the accurate depiction of human expressions, movements and facial expressions. Veo 2’s understanding physics enhances dynamic sequences. This makes the videos not only visually appealing, but also believable when it comes to how objects and characters interrelate within the scene. User Accessibility
and Impact
Veo 2 will initially be available as part of Google’s VideoFX platform behind a waiting list, signaling Google’s cautious but progressive approach to making such powerful technologies accessible.
Google can gather feedback from the initial users, and refine the model using real-world scenarios. The implications are huge for creators, businesses, and educators. This tool can democratize the production of video by reducing technical and cost barriers.
https://labs.google/experiments
Conclusion
Veo 2 from Google DeepMind represents a significant milestone in AI-driven video production. It allows users to create cinematic visuals by enhancing their control over visual storytelling elements.
As the technology is integrated into more platforms and tools, it will likely lead to a surge of innovative content creation that could redefine how stories are told digitally.
With great power comes great responsibilities, and the ongoing discussion about ethical AI usage will be crucial to harnessing Veo 2’s potential for the better.