Generation AI director Dave Clark tried “Every tool under the sun” to make his own artificially intelligent films, most famous for his virus Trump Battalion. But even since he has released this video, the Gen-A technology has been very advanced and believes that what was considered innovative then can now be even more promising.
On Tuesday, Google unveiled Flow, a Great Video Generation Tool, which is powered by Google AI, VEO 3 at its I/O event. Clark was one of AI’s filmmakers in collaboration with Google Labs about the functionality and interface of the new tool, so that it was created with filmmakers and creators. It is enough to say that he was fascinated by the results.
More out of indiewire
-
Christian Petzold made his debut in his Cannes and how his actors changed the script: “I have to destroy my specific behavior”
-
Creating a series of book transactions for all your films, starting with Once in Hollywood.
“My mind was still blowing in control,” Clark said before starting Flow’s release. “This is what I waited for me as a director. You put 10,000 hours, you are fighting these calls and using these different systems. But we really hope it is the level of control that I feel, as we finally understood, and it will come so early.”
Flow has camera controllers that allow the filmmakers to specifically arrange the motion and the visual angle they want to create to direct the cinematographer. It has a feature called “scenesuilder” that allows filmmakers to generate the scene and then expand it by maintaining all its characters’ designs and places consistent and intact. Creators can upload and combine several reference images that will match things like faces, clothes and space, doing so with a particularly specific detail.
Flow also allows you to quickly write natural language to understand the script and granular technical details such as the choice of lens, lighting conditions, fireplace distance or film grain, visual language. It even manages to generate real -time sound, including sound effects, background music and dialogue.
Clark and his AI production promise are using Flow to create a new AI short, called Freelancers about two adopted brothers, each becoming international spies and hits. He described one scene, creating two of his characters sitting in a restaurant, along with a shot with a slow delly. He then urged the model to use a dynamic, hand camera, followed by a shot of two acting characters.
“And it did! To see how the camera was shaking, it is exactly the same as I probably filmed if I were in that restaurant. It was pretty cool and I didn’t see that level [control]”Said Clark.
Not long ago, Open introduced their Sora image model; Then the Ai filmmakers we talked to were confused that it had a “quick adherence” where you can arrange one thing to happen, and then that figure would happen to something else. Flow and Veo 3 have shown even more complex opportunities when stage creators generate, then drag and expand by time zone to effectively continue the action and even extend the person’s frame for 6-8 seconds, the model usually generates.
Clark equated this to practically able to create his “1917” style on the Oner, if he would like, and the action continues from place to place, even if the character remains the same.
“Let’s say it’s a guy driving a car around the Amalfi Coast. When the ‘SEEBUILDER INSIDE FLOW’, the director was able to block the rest of the scene in me,” Clark explained. The character looks like the Amalfi coast and the road seems to be still.
Clark said it became as specific as arranging the scene to look as if it was shot with a 50 mm lens up to 135 mm long lens, and the flow not only understands, but can marry this work with specific blocking.
AI created image created using Google’s Imagen 4Google Labs courtesy
Matthieu Kim Lorrain, Google Deepmind’s creative manager, and Thomas Iljic, Google Labs Product Manager, talked to Indiewire about how to understand a person like Clark’s creative process to inform how Flow should operate, functionality.
“Showing and saying is important. They don’t just want to enter the text,” Iljic said. “They want to include these elements, these ingredients. They want consistency during shots. They want projects to start the structure because it is so many files, wonder how I am going to put this thing?”
Lorrain added that the main word of AI filmmakers is “customization”, and five different AI filmmakers can have five different workflows, or even one director may have a different method for a different project, all of which are built into flow.
“Sometimes you need a video because you really want to start with the composition. You want to really make the perfect picture and then expand,” said Lorrain. “Sometimes it is in the world.
Clarko’s promise has previously announced additional funding for the Google AI Foundation, as well as Crossbeam Venture Partners, Kiwu Ventures and Saga Ventures. However, Google also cooperates with two other filmmakers about Flow’s launch, including Henry Daubrez and Junie Lau. Flow is also now available in Google AI Pro and Google AI Ultra Ultra Ultra Plans in the US, and more countries will soon appear.
Clark is not sure what the Next Frontier with rapidly changing technology is, but he thinks that the tools become more accessible, it must at least understand it.
“I always refer to James Cameronons and George Lucazes, some of the best filmmakers have always been technologists, and I feel we will see that other level, the other George Lucas School, which will be by nature technologists,” Clark said. “You will need to understand technology, whether you are using it or not.”
The best indiewire
Sign up to get an indiewire newsletter. To get the latest news, follow us on Facebook, Twitter and Instagram.