However, despite these features, Imagen 2's video generation capabilities are not as advanced as those of competitors like Runway and Stability AI. Google has also faced criticism for its handling of training data, with concerns raised about the potential for copyright infringement. The company has not committed to releasing an opt-out tool or compensating creators for their contributions to the training data. Furthermore, Google's generative AI indemnification policy does not cover "text-to-live images," leaving users to use the feature at their own risk.
Key takeaways:
- Google has launched an enhanced image-generating tool, Imagen 2, within its Vertex AI developer platform, which can create and edit images given a text prompt.
- Imagen 2 can now create short, four-second videos from text prompts, a feature Google is calling 'text-to-live images'.
- To mitigate concerns about deepfakes, Imagen 2 will use SynthID, a Google DeepMind-developed approach, to apply invisible, cryptographic watermarks to live images.
- Despite these advancements, the article suggests that Imagen 2's video generation capabilities may not be competitive with existing tools, citing higher resolution and greater customizability offered by other platforms.