Unnisbis is a creative studio exploring generative art
The Rhetoric of the JPEG, 2019, Archival Print on Canvas - Generative Art by Fabien Charuau
Our name comes from the Hindi phrase “unnis-bees ka farak” , the tiny difference between 19 and 20. We now live in a statistical world, away from old determinism. AI generation, especially diffusion, begins with noise and works its way towards order, always getting close to reality, but never fully there. At Unnisbis, we embrace that space, narrowing the gap until the difference is almost invisible, close enough to feel real, yet always carrying its own poetry.
The PROCESS❊
01The Art of Subtle Difference (19/20)
Inspired by the Hindi idiom “unnis-bees ka farak” (the tiny gap between 19 and 20), Unnisbis.ai embraces the idea that art doesn’t need absolute perfection to be powerful. Its goal is to create AI works that are so close to reality that the difference is almost imperceptible and that last sliver of imperfection becomes the space where creativity thrives.
Noise as a Creative Partner
Unnisbis.ai celebrates the quirks, errors, and randomness in diffusion models. Rather than seeing these as flaws, the studio treats them as fingerprints of generative art, much like brushstrokes in painting or film grain in photography. This positions AI as a tool of serendipity and surprise, not just precision.
02AI as a Legitimate Artistic Discipline
The studio advocates for AI-generated art to stand shoulder to shoulder with photography, painting, and cinema. Just as photography once fought for recognition as art, Unnisbis champions generative AI as a profound medium for contemporary expression not a gimmick, but an authentic creative discipline.
03Science Turned Poetic
The philosophy transforms technical principles, like curve-fitting in neural networks or Weber’s Law of perceptual thresholds, into artistic metaphors. Unnisbis takes the cold language of math and code and bends it into poetry, showing that computation can carry emotion and meaning.
04Reclaiming Control, Crafting the Real
While surface-level AI outputs can look instantly impressive, Unnisbis focuses on the difficult work of guiding algorithms toward results that feel convincingly real. This is a philosophy of craftsmanship: combining human judgment and machine power to blur the line between reality and imagination, reclaiming control over generative processes to push them beyond clichés.
05Ai Generated Commercial Work
For this Mille Plant Protein ad, we built the film entirely from scratch with AI generation. The key challenge was ensuring the model remained visually consistent across the full 40 second spot. Every frame was carefully guided so the character held the same look, feel, and identity throughout, giving the ad a polished, seamless finish despite being fully AI driven.
…more..
We set out to see if AI could recreate the ordinary setup of small-town India : streets, uniforms, gestures that feel instantly familiar to Indian viewers. With AI it is easy to generate the fantastic, the surreal, or the futuristic. The harder test is the mundane: the everyday details that make something believable. This project became an exercise in consistency across characters, clothing, and location, pushing AI to hold the thread of reality from frame to frame.
…more..
This campaign was developed for an upcoming project. At the start, we had only a building render and a few beach photos. Everything else, including the models, the clothes, and the styling, was created separately with AI. We ran the assignment like a regular advertising project, selecting cast, wardrobe, and mood so the client retained full control over the final look and feel of the ad.
…more..
Copying the ordinary is harder than inventing the extraordinary. We built a 15-second finance ad entirely with AI, not flashy, not experimental, but precise down to the smallest detail. The challenge was invisibility: to make something that could pass on television as just another commercial in the ad break. That seamlessness, that banality, is where the real technical difficulty lies.
…more..
This short video was created as a pitch to the client. The goal was to show that we can maintain model consistency across shots while also working with real spaces. The building elevations, the swimming areas, the living room, and the surroundings are kept exactly as in the photos provided by the client, with AI generation used only to integrate models and details seamlessly into those spaces. This demonstrates how existing assets can be extended and enriched without losing authenticity.
Founder - Fabien Charuau
Interview for the online magazine Matter - 2022
My work spans photography, writing, and generative AI. I began reflecting on the nature of images long before AI, and today I bring that depth into practice with AI processes that generate, refine, and reimagine images. Alongside making, I teach and speak about generative AI, exploring how technology is reshaping visual culture and artistic practice.
Article written for the scholarly anthology “Photography in India” published by Bloomsbury in 2020
Solo Show at Chatterjee & Lal - 2015
My projects have been exhibited in galleries, positioning AI within the lineage of contemporary art rather than as novelty. This perspective informs the consultancy I offer to clients: combining artistic vision with technological expertise to guide how AI can be used critically and creatively. Each collaboration is a way to bring years of reflection on images into practical, forward-looking application.
Generative Art: Bodies of Work
The Rhetoric of the JPEG series takes its cue from Roland Barthes’ Rhetoric of the Image but shifts the focus to the digital file itself. The project works with random image searches built from strong keywords, then applies a scripted series of operations in Photoshop with the aim of exposing the underlying structure of JPEG compression. What emerges is not only an image but the rhetoric of the format — its blocks, its noise, its hidden grid. Instead of transparency, the JPEG insists on its own materiality, showing that even the most common file type carries an aesthetic and a politics of vision
Being Seen Trying is a room-scale video installation built from online darshan footage, where devotees connect to ritual through screens. Using face-recognition software, the work fragments these faces into pixelated images projected onto panes of glass, isolating individuals within the collective. Echoing darshan’s promise of seeing and being seen, the piece explores how prayer, presence, and visibility are reshaped by digital mediation.
A Thousand Kisses Deep is a generative art series written in Java. Around ten years ago, images of couples kissing began to appear widely on Indian social media, signalling a shift toward new forms of intimacy in public space. I gathered these photographs as raw material, seeing them both as personal gestures and cultural markers of change.
The images were processed through an algorithm based on a function that calculates the “energy” of each pixel. Repeated across hundreds of thousands of iterations, the figures dissolve and the kiss emerges instead as a field of intensity. What remains is not depiction but force, a poetic transformation of intimacy into code that celebrates how desire finds new shapes in the digital.
The Cassini Conspiracy reworks raw images from NASA’s Cassini–Huygens probe. Borrowing the methods of conspiracy theorists who see hidden forms in pixels, the project embraces apophenia: our tendency to find patterns in noise. By sharpening, opening shadows, and exposing artifacts, the work turns random data into a space for projection, where delusion and creativity share the same ground.
Share Your Idea, Let’s Generate It Together
Images grow from exchange. Tell us what you have in mind, and let’s see where we can take it together.