DeepMind’s AI Creates 4K Video

▶ Source –  TechCrunch ▶ Google DeepMind has unveiled Veo 2, a video-generating AI capable of producing up to two-minute clips at up to 4K resolution, exceeding OpenAI’s Sora capabilities in duration and resolution. Currently, VideoFX limits Veo 2 output to eight seconds at 720p. The model features enhanced physics understanding, motion realism, and cinematic camera control, with ongoing refinement through creative collaborations. Google plans future integration into its Vertex AI platform while addressing challenges like coherence, hallucinations, and safeguarding against deepfakes.
▶ Editor’s Comment – DeepMind’s Veo 2 pushes forward the possibilities of AI video generation, particularly with improved cinematic effects and physics realism. However, its current limitations highlight ongoing challenges in coherence and character consistency. Nevertheless, the potential to create shortform videos for training and education look very promising.
▶ Image –  Google DeepMind ▶