Now it is likely saw short videos created using AI video generation tools that allow you to create photorealistic clips of several seconds from a simple text prompt. An Indian startup is now pushing the technology to its limits: by the end of 2025, it plans to release a feature-length film created almost entirely using generative artificial intelligence tools.
Chandigarh-based Intelliflicks Studios is the brainchild of author Khushwant Singh and Gurdeep Pal, president of AI strategy at Qualtrics in Seattle and former corporate vice president of AI incubation at Microsoft. The studio is making a film adaptation of Singh’s 2014 novel Maharaja in denimwhich tells the story of a young man from the present who believes he is the reincarnation of Maharaja Ranjit Singh, the 19th century founder of the Sikh Empire.
Singh says studio heads in Bollywood bought film rights to the book twice, but the complexity and cost of telling the story over multiple time periods meant the film was never made. So when Pall, Singh’s childhood friend, told him about the rapidly improving capabilities of AI video generators, the pair decided to join forces to create what they believe to be the first full-length AI generative film. “We’re trying to make a breakthrough to show the capabilities of this technology,” says Singh.
What generative AI tools do they use?
The company is using a suite of commercial and open-source artificial intelligence tools to make the film, and is developing its own software to manage new workflows, according to Pall. It uses image generation models to create character, scene, and object designs that are then fed into video generation models. Other AI tools are used to create audio, lip sync dialogue and sharpen images. Pall says his team also uses conventional video production tools for simpler tasks, such as matching lighting and color between scenes.
The developers primarily use pre-trained models, and Pall says they have also fine-tuned some models for India-specific data. However, in some cases fine-tuning is not enough. One scene involves a woman performing a traditional northern Indian dance called kathak, and Pall says collecting enough data to train the model would be impractical. Instead, they plan to record an actual Kathak performance and use AI to exchange face-to-face with the AI-generated character.
Intelliflicks Studios has released this trailer for the AI-generated feature film it plans to release this year. IntelliFlicks Studios
According to Palla, the biggest challenge the team has faced is consistency. Generative AI is inherently probabilistic, so the model’s response to a particular challenge will be different each time. This can make things tricky when a character has to have the same look throughout a feature film.
This challenge has become significantly more manageable in the last year as many models can now add a digital tag to each output. This tag can be added to future calls to ensure that the model follows a similar style when generating a new clip. But the re-creations are never perfect, Pall says, adding that his team adapts to the limitations of the technology. “You have to look at it as a new medium,” he explains. “You can’t paint the same with water colors as with oil.”
What do external experts think?
Jamie Umphersonhead of creative at video startup AI Pathin New York, says the most successful AI video projects are those that understand the technology’s limitations and lean on them to enhance the storytelling. Still, technology is constantly improving, he adds, so some of these restrictions may be short-lived.
Still, making a feature film with today’s technology is a bit of a chore. Umpherson says most of Runway’s customers—which include film studios, ad agencies and independent artists—use the technology to quickly iterate on ideas early in the creative process or create visual effects to complement live action. “Making a fully formed film is definitely possible,” he declares, but it will require “an incredible amount of artistry.”
Many of today’s video generators now provide a tag with each generated clip that can be added to the next prompt to improve continuity.
IntelliFlicks Studios
Part of the challenge, says Abe Davis, an assistant professor of computer science at Cornell University, is that these tools are designed to generate high-fidelity video with minimal user input — taking control of details that would normally require human decisions. -production. This automation allows a layperson to quickly generate a clip, but can frustrate someone with expertise and vision. “People underestimate the number of relevant or important decisions that a filmmaker actually wants to make,” says Davis.
The AI-generated film takes place in both the modern world and the 19th century. IntelliFlicks Studios
Take, for example, the decision about how an actor should deliver a line; this direction may be difficult to express in a text prompt. And yet all of these details must remain consistent throughout the video, Davis adds, which becomes increasingly difficult as the video gets longer.
Singh admits that the first feature film created by artificial intelligence is likely to be significantly different from those produced conventionally. But he hopes the technology will break down the structural barriers that keep people from expressing their creativity. AI is a game-changer, says Singh: “I think it’s going to democratize filmmaking in a huge way.”
