Abhishek Das

Artist

Abhishek Das · abhishekdas.com/art ↗

Based

Born in New Delhi, India. Lives and works in the United States.

Practice

Generative artist and AI researcher collaborating with Devi Parikh on long-form Art Blocks projects that investigate algorithmic textures and emergent organic structures.

Now

Webpage designer and co-creator for Cushions (200 editions, 2021); contributor to Aragnation.

Raster

raster.art/artist/abhishek-das ↗

In the set

Cushions ↗ · 200 editions

Links

Full bibliography ↗

Abhishek Das approaches generative art from a foundation in artificial intelligence research and software development. Born and raised in New Delhi, India, he moved to the United States in 2016 to pursue a PhD at Georgia Tech, where he contributed to early work on agents that can see, talk, and act.¹ His day job involves building algorithms to mimic intelligence, and he has extended this interest to algorithms that mimic textures from the physical world such as paper, ink, cloth, and sand.²

Collaboration with Devi Parikh, an AI researcher and generative artist, dates back to 2015 across numerous projects in AI and art. Abhishek introduced Devi to generative art in 2018, and together they have produced works that question boundaries between human and machine creativity.¹

Beauty often lives in our unique, individual oddities. To break the uniformity and add an element of imperfection, some regions may be rendered as a dark void.

The Cushions project exemplifies this shared sensibility. Launched in December 2021 on Art Blocks, it consists of 200 editions generated from a system of Truchet tiles and connected component analysis that identifies cushion-like shapes in negative space.³⁴⁵

The Cushions Series

Cushions begins with a grid of randomly oriented diagonal lines. Connected component analysis then isolates contiguous regions resembling cushions or pillows. These regions receive uniform color fills drawn from palettes inspired by digital illustrations, pastels, and diverse artistic traditions. Optional dark voids within some regions incorporate textures to heighten the organic quality. Thirteen features control aspects such as scale, curvature, density, bias, thickness, style, color count, contrast, composition, accents, and void characteristics, yielding more than five billion possible variations.³

Abhishek Das contributed the webpage design and participated in the collaborative development of the project, which Art Blocks credits to both artists.⁴⁸ The series sits within the broader movement of long-form generative art, where artists define systems that produce diverse, high-quality outputs over many iterations rather than single images.⁶

The six examples below illustrate the range of compositions, from dense interlocking forms to sparse arrangements with prominent voids and textured accents. Each token draws randomness from the transaction hash at mint time, ensuring unique yet coherent results across the edition of 200. Vibrant yet restrained palettes shift between soft pastels and bolder accents, while line weights and curvatures produce effects that range from architectural to fluid. Some outputs emphasize packed, pillow-like clusters; others open into airy fields punctuated by dark, textured voids that suggest depth and material tactility.

Live on Art Blocks: cushions-by-devi-parikh ↗

Collaborative Context and Practice

Beyond Cushions, Das and Parikh developed Aragnation, a long-form AI generative art project on Art Blocks that employs tiny probabilistic models rather than large neural networks to construct abstract imagery with watercolor and art-paper aesthetics.¹ A probabilistic graphical model learns prototype shapes and colors from training photographs taken by Das, while a small multilayer perceptron generates color gradients. The project name is an anagram of "No-GAN AI Art," signaling a deliberate turn away from contemporary diffusion or transformer pipelines toward hand-crafted, lightweight systems trained from scratch without GPUs. Features include paper textures, subject categories such as flower, bird, urban, and landscape, and optional reflections that add depth. The approach underscores Das's interest in constrained systems that still permit rich emergence and variation.

Das maintains a personal practice in p5.js, creating generative pieces that simulate physical media and processes. He describes himself as relatively new to the field yet draws on a lifelong engagement with code that began in middle school and a parallel passion for music as a trained pianist.² His scientific research on AI for catalyst discovery in renewable energy storage informs a precise, systems-oriented approach to art making, developed through academic and industry roles.⁹

These efforts position Das within a lineage of artist-scientists who treat code as both tool and medium. The resulting works emphasize emergence, imperfection, and the beauty of constrained systems, qualities evident across the Cushions variations and his other collaborative outputs. Secondary market activity on platforms hosting the edition further attests to sustained collector interest in the series.⁷

Bibliography

The references cited in this essay are listed below. For the complete bibliography see the Abhishek Das links page ↗.