Veo, the newest generative AI video model from Google, is now accessible for businesses looking to integrate it into their content creation workflows. Initially revealed in May—three months after OpenAI showcased its rival product, Sora—Veo has made its debut in a private preview through Google’s Vertex AI platform.
What are the capabilities of Veo?
This model can produce high-quality videos in 1080p resolution across various visual and cinematic styles, utilizing either text or image prompts. When the model was first introduced, the generated clips were described as being longer than a minute in duration, though Google has not provided specific length limitations for this preview version. Recent example clips shared by Google align with what has been previously observed from Veo; without careful scrutiny, it is extremely challenging to distinguish these videos as AI-generated.
Expansion of Imagen 3
Additionally, the updated version of Google’s Imagen 3 text-to-image generator will be available to all Google Cloud customers via Vertex starting next week, following its initial US rollout on Google’s AI Test Kitchen back in August. Users on Google’s allow list can also explore new features, including prompt-based photo editing and the option to infuse their own brand, style, logo, subject, or product features into the generated images.
Safeguards and technology integration
Google asserts that both Veo and Imagen 3 include built-in safeguards to prevent the generation of harmful content or breaches of copyright protections—although previous observations suggest that bypassing these safeguards may not be particularly challenging. Every output generated by Veo and Imagen 3 is integrated with DeepMind’s SynthID technology, which acts as an invisible digital watermark aimed at reducing misinformation and misattribution concerns. This concept is akin to Adobe’s Content Credentials system, which can be embedded in content created by the company’s own image and video generative AI models.
Competitive landscape and market trends
With Google’s video model now launched, OpenAI finds itself lagging behind its competitors and facing a tight timeline to fulfill its commitment to release Sora by the end of 2024. AI-generated content is already making its way into advertisements, such as Coca-Cola’s recent holiday campaign, and businesses have a strong incentive not to delay for Sora, as Google reports that a significant percentage of organizations currently utilizing generative AI are experiencing revenue growth.
