.jpg)
The short film production landscape is set to change with the launch of Google Flow at Google I/O this morning. This is a new service from Google that can generate short videos using a combination of the previously announced Veo, Imagen and Gemini models.
Flow is unique in that assets in the form of images can be generated using Imagen which are stored as props using the “Ingredients” feature. Flow can then be directed using Ingredients in the video generated using Veo. The prompts can be written in plain language that Gemini understands.
Flow will remember all the objects and videos that the user generates. Short videos can be generated from one scene to another continuously with characters and objects remaining consistent. This is an early issue with some previous video generative AI models which were too random. Control is placed in the hands of the video producer at all times. What’s more powerful is that users can direct the desired camera movement in the video before it is generated.
Flow also has basic editing features built in allowing for odd-looking scenes to be cut out immediately before the next scene is generated. If the resulting scene is too short, the Extend feature can be used to extend it to the desired ending.
Google worked with several creators to develop Flow, and the videos above and below are among the short films that have been produced entirely with the generative AI service. Google Flow can be used with a Google AI Pro subscription for $19.99 (~RM 86) per month or Google AI Ultra for $249.99 (~RM 1074) per month, which offers more features and generation credits.