Music Video

Tera Mera pyar is a story of two lover that meet and then part ways in every life. The beauty of this song is that it features real actors. It's one of a kind, created by a propriety workflows.

Comment
  • Inspiration

The idea was born from the timeless emotion of eternal love — how it can transcend lifetimes, memories, and even reality. We were inspired by the poetic visuals of The Great Gatsby, the elegance of Mughal-e-Azam, and the visual direction of filmmakers like Satyajit Ray and Wong Kar-Wai. Each era reflects a reincarnated memory of love — from 1920s jazz ballrooms to ancient India’s royal courts.

We asked: What if love was remembered not linearly, but as fragments across lives?

  • What it does

This project is an AI-generated music video that blends live-action aesthetics with pencil-drawn storyboards and stylized eras. Each scene transitions across timelines — 1920s, prince of Persia, ancient India, modern day — as a couple finds each other again and again.

The video uses:

Flash cuts, dolly shots, whip pans (inspired by Ray)

Vertical format (16:9 + 9:16 mobile exports)

Hand-drawn storyboard sequences

Character LoRAs of Arjun and Aisha (Rynaa) powered by ComfyUI and Flux

AI-generated realistic renders for select panels

  • How we built it

Storyboard Design:

Created pencil-style hand-drawn sketches for each major scene using AI image workflows

Inspired by old-school previsualization methods, each scene was directed with attention to cinematography (camera angles, movement, mood)

Tech Stack:

ComfyUI + Flux + Custom LoRAs trained on Arjun (Sa1man) and Aisha (Rynaa)

Workflow scripting for panel-wise storyboard generation

Used prompt-engineered rendering to match lighting/mood of each era

Scene Design:

Scene-by-scene timeline breakdown in vertical format

Visual consistency with character sheets and close-up continuity

Annotated visual direction for AI to recreate human-style camera work

  • Challenges we ran into

Maintaining Character Consistency: Swapping eras while keeping Arjun and Aisha visually consistent (across AI render passes) was tough. Required precise prompt balancing and manual selection.

Stylization vs Realism: Achieving the soft look of pencil-drawn sketches without losing clarity. Also generating realistic outputs that still feel cinematic rather than uncanny.

Camera Direction in AI: Simulating crane-downs, dutch tilts, dolly-outs in AI still-image formats needed creative prompting and scene labeling. We had to embed camera logic in image generation descriptions.

Aspect Ratio Management: Moving from wide cinematic 16:9 to mobile-ready 9:16 without losing narrative clarity.

  • Accomplishments that we're proud of

Created a storyboard that blends ancient and modern eras through cinematic transitions

Achieved visual emotion in sketches — sadness, longing, rediscovery

First time using AI tools to simulate Ray-style scene language and dolly movements

Seamlessly integrated ComfyUI LoRAs and Flux to maintain character fidelity

  • What we learned

Prompt engineering is a form of direction — we learned to direct AI like a cinematographer

AI can generate not just art, but cinematic rhythm if scenes are broken down like film grammar

Iteration is key — we often needed 3-4 versions of each panel to hit the right tone

Character-first storytelling makes even surreal timelines emotionally grounded

-What's next for the Music Video

Render all 20 storyboard panels in high-resolution animation

Add sound design and original score for transitions

Release teaser reels for each timeline on social platforms

Collaborate with musicians and poets to write era-based lyrical versions

Explore voiceover performances from AI-generated actors

Submit to AI Art and Short Film festivals (like SIGGRAPH, Sundance AI labs)

Built With

Share this project:

Updates