With the talent of thousands of researchers, programmers, visual effects artists, archivists, and producers from Google's different verticals like DeepMind and Cloud, along with people from Sphere Studios, Magnopus, Warner Bros. Discovery, and others in the film and technology industries, Google has come up to the point of releasing "The Wizard of Oz," a Hollywood classic from the 1930s, reimagined solely using AI models. The movie will be released on August 28 at Sphere, the colossal Las Vegas venue. 

It's not the very first movie to be reconceptualized with AI, but it will soon be known for the same, Google’s blog post states. It is because of the amount of effort and detail that has gone into bringing it across on a 16K and 160,000 square feet screen. For instance, to keep the originality of the movie intact, neither a line of new dialogue nor a note of new music has been introduced in enriching this classic for the spherical theater. Interestingly, the theatre boasts its capacity to seat around 17,600 people in one go.

Turning Tiny Frames into Ultra-High HD Images 

The very first task in reimagining the film was to fit the old visuals of the movie to suit the 160,000-square-foot screen. It was addressed by the Google teams and their partners, who developed an AI-based “super-resolution” tool to turn those tiny celluloid frames from 1939 into ultra-ultra-high-definition imagery. 

Afterward, the teams performed AI outpainting to expand the scope of scenes to both fill the space and fill in the gaps created by camera cuts and framing limitations.

Read about OpenAI's latest image generation model at: Explore How OpenAI’s "4o Image Generation" Delivers Accurate, Context-Aware Images

Fine-Tuning the AI Models 

Other than the original film, the teams went with other conventional traditions of cinema, where teams have a blank yet filled canvas to paint. Simply put, the team referred to resources like old footage and supplementary material, such as the shooting script, production illustrations, photographs, set plans, and scores to fine-tune the AI model. This helped the team to equip models like Veo and Gemini to suit their needs and understand the situation with more clarity. 

This helped the models to adapt the specific details of the original characters, their environments, and even elements of the production, like camera focal lengths for specific scenes.

Wildly Innovative Models 

“The models, they’re wildly innovative,” Dr. Steven Hickson, a Google DeepMind researcher on the project, says. “We’d find something we can't do, we think it's impossible, and then a month later we're like, actually, maybe we can do that.”

To read more on AI, click here to visit our category page! 

This illustrates one of the instances of innovation and creativity that AI models can exhibit. It is equally good and bad to witness this. Good, because the technological leap is good enough to be praised, while bad because it is going to affect the employment metrics of the entertainment industry. Apart from being good or bad, one thing can be said safely: the models are utterly innovative and up to the creative mark.