Intelliflicks Builds India's First AI Film Maharaja in Denims

Intelliflicks Studios is producing Maharaja in Denims, positioning it as one of the world's earliest feature-length films created largely with generative AI. The project is co-founded by author Khushwant Singh and former Microsoft executive Gurdeep Singh Pall, and co-produced with Twenty21 Studios. The team operates with a micro-crew of six, claiming production costs reduced to roughly one tenth of a standard budget while eliminating traditional actors, sets, and camera logistics.

The production pipeline substitutes AI-driven image and scene generation for conventional filming, but realism remains a hard technical problem. The team reports model limitations on fine-grained facial realism for South Asian physiognomy and historical Sikh figures, forcing iterative fixes and manual intervention. Rapid cadence in generative-image model releases creates a software-churn problem: newer models make earlier renders look dated, prompting re-renders and additional licensing or compute spend. Key technical constraints practitioners should note:

• •Data bias and domain coverage, with current models undertrained on Indian faces and cultural artifacts
• •Temporal-consistency and motion realism across long takes, which generative models struggle to maintain
• •Asset and pipeline versioning costs due to fast model upgrades and format incompatibilities

The team retains human creative control for score and voice by commissioning a title song from Sukhwinder Singh. Production prioritizes AI for visuals and scene assembly while keeping a director and cinematographer to curate outputs and manage continuity. The project has publicly framed these choices as productivity and cost wins, not replacement of creative authorship; Khushwant Singh summarized the shift plainly: "There is no actor fee. There is no fuss over them coming late or causing delays. There are no sets."

India produces more than 2,000 movies annually, and its industry is comparatively receptive to AI-driven workflows because cost savings scale across low- to mid-budget films. Unlike Hollywood, where strikes and union rules have constrained AI adoption, parts of Indian cinema are experimenting aggressively with generative tools. That said, the genre matters: mythological and sci-fi productions can mask artifacts, but realism and historical narratives expose model weaknesses. Competing projects such as Naisha, Chiranjeevi Hanuman: The Eternal, and regional titles have also marketed AI-first approaches; some have postponed release dates citing technical issues, underscoring the gap between proof-of-concept clips and a stable feature-length pipeline.

This is a production-scale stress test of generative-video workflows that surfaces practical engineering problems: dataset gaps for non-Western demographics, asset versioning under model churn, compute and licensing cost volatility, and human-in-the-loop curation needs. For ML engineers, those are actionable signals for dataset collection, evaluation metrics for facial and motion fidelity, and pipeline design that supports reproducible renders and cheap rollbacks.

Track public artifacts or technical writeups from Intelliflicks and partners for concrete pipeline choices, metadata standards they adopt for versioning, and any licensing or rights disputes around synthesized likenesses. The outcome will influence whether other regional studios scale generative pipelines or remain experimental.