AI is rewriting Sandalwood's songbook
Music isn't just a backdrop in Kannada cinema. It's a core driver of emotion and box office. Now, AI is stepping right into that core - writing lyrics, composing melodies, and even rendering vocals on demand.
Composer, actor, and director V Manohar put it plainly: "AI is being used to write lyrics, especially in low-budget movies… When you push the button, the song is ready with an AI-based singer belting out an AI-based song of AI-composed music from AI-powered lyrics." His concern is clear - if this runs unchecked, many musicians and singers could see less work. Yet he also notes that audiences often prefer authenticity, much like how widespread dubbing never fully replaced originals.
What's already changing on set and in studio
National award-winning filmmaker Avinash U Shetty believes the industry has barely started. In his view, only a slice of AI's potential is in use. Resistance looks familiar; the analog-to-digital shift drew the same skepticism until it didn't. What feels off today might feel normal in five years.
Proof of the cost and speed shift is already here. Sangamesh produced a three-minute music video end-to-end with AI for Rs 4,500 - work that could have cost Rs 15 lakh - and finished it in three days. The only real spend? A tool subscription.
Action, crowds, and "is that real?"
Senior stunt master Thriller Manju sees the line between real and synthetic blurring. War scenes, big fights, and dance numbers can be shot with fewer people, then multiplied with VFX and AI. That saves money and time, but it also changes how stunt teams, dancers, and junior artistes get hired.
Period films on startup budgets
Director P Seshadri wants to make a Kannada film on Ferdinand Kittel, who helped print Kannada dictionaries in the 1890s. He estimates a traditional approach could cross Rs 100 crore. With AI, he sees a path near Rs 1 crore - using AI to prototype visuals, build sets digitally, and trigger sharper decisions before spending on physical production.
If you're curious about Kittel's legacy, here's a quick primer: Ferdinand Kittel on Wikipedia.
Adaptation, identity, and where to draw the line
Actor Ramesh Aravind is pragmatic: change creates new kinds of work. Learn the tools early, or risk being left out of the next call sheet. That mindset helps replace fear with ownership.
Actor-director DP Raghuram raises a key point: earlier, people chased the big-screen experience with a different kind of passion. Today, cinema comes to the phone, but the fever is softer. For him, AI should help improvise - not become the creative core. That tension matters.
There's also the human side. Hundreds of junior artistes once filled crowd scenes - not just for wages, but for meals, memories, and real proximity to stars like Dr. Rajkumar, Vishnuvardhan, and Ambareesh. Technology compresses budgets, but it can also compress community. That deserves intention, not indifference.
What this means for creatives: a practical playbook
- Define the line: Use AI for drafts, references, and previz. Keep final taste, voice, and performance human.
- Budget with clarity: Automate low-signal tasks (scratch melodies, temp vocals, crowd fill). Spend on high-signal tasks (lead vocals, live instruments, craft edits).
- Set consent rules: Don't clone voices or faces without written permission. Keep licenses and model sources documented.
- Build a minimal AI stack: One tool for lyrics, one for composition, one for voices, one for visual previz. Keep version control and naming tight.
- Test with real viewers: A/B test AI vs human takes on small cuts. Track retention, replays, and comments before you lock the master.
- Upskill your crew: Train a point person on prompts, references, and model selection. If you need a jumpstart, see practical options for creatives here: Courses by Job and Generative Video Tools.
- Create new roles: Prompt editor, AI music editor, dataset wrangler, ethics lead. Small teams can combine roles; large ones should specialize.
- Keep junior artistes in the loop: Use hybrid shoots - film a base group on set, extend with AI later. Rotate extras, keep meals, maintain that bond.
A simple workflow you can try this week
- Brief the intent: emotion, tempo, references, instruments, language, and scene length.
- Generate 5-10 lyric drafts with clear prompts. Merge the best lines into one coherent version.
- Create scratch melodies from 3-5 styles. Pick a top 2 and iterate until it fits the scene timing.
- Render temp vocals for timing. Then hire a session singer to record the final take over the best arrangement.
- Previz complex scenes: block movements, camera angles, and crowds with AI and VFX placeholders before shoot day.
- Swap AI elements with human performances where it matters most. Lock rights, credits, and consent in writing.
Costs, speed, and taste
Sangamesh's Rs 4,500 video proves the new math: ideas move faster and cheaper. But taste is still the lever that matters. Audiences will forgive tool choice; they won't forgive bland work.
If you're a creative in Sandalwood, the path is simple: keep the soul human, let the machines do the grunt work, and make room for people who make sets feel like family. That's how you protect identity while shipping better, faster stories.
Bottom line: Use AI to cut noise, not corners. Let it draft. You decide what lives in the final cut.
Your membership also unlocks:
