Google Omni model joins AI video generator Flow

google introduced 'omni flash' within its flow platform, expanding its generative video capabilities by allowing users to combine video uploads with text prompts to create and edit footage. unlike its predecessor veo, omni incorporates real-world knowledge to ostensibly maintain character consistency and supports iterative editing through text instructions, such as adjusting facial expressions or removing objects. testing indicates that while the tool follows narrative prompts better than veo, it frequently fails at object permanence, often transforming props like a jar of honey into a water bottle or incorrectly adding features like antlers to characters. generation requires a $20 pro subscription, with costs ranging from 15 to 40 credits per clip or edit, making experimental workflows expensive due to the high credit consumption.