# Refik Anadol Refik Anadol’s AI Data Paintings and Sculptures translate the logic of new media technology into spatial design to create immersive experiences based on vast datasets of nature, history, and human activity. ## Large Nature Model: Living Archive ### *Large Nature Model* 2024 Refik Anadol Studio is pioneering a groundbreaking project—the Large Nature Model (LNM). This is a unique open-source, generative AI model focused on nature, using the largest ethically sourced dataset of natural elements. This multi-modal LNM will be able to understand and generate content across multiple forms of media, such as text, images, audio, and video. [Watch Now](#) ### *Generative Reality Research* 2024 Refik Anadol Studio's interactive AI research focuses on producing generative realities through real-time rendering, scent, and sensory inputs. In this project, fully interactive generative 3-D worlds are envisioned by AI but realized by 3D artists. The interactive component lets the audience navigate through millions of raw datasets from the Large Natural Model and discover how machines categorize bird sounds or images from nature. As they travel through these immersive worlds, the viewers can also verbally or textually interact with an AI-driven character that uses custom LLM models. [Watch Now](#) ## LNM—AI Data Paintings **Generative Reality—LNM: Fauna** **Generative Reality—LNM: Flora** **Generative Reality—LNM: Landscapes** **Machine Hallucinations—LNM: Fauna** **Machine Hallucinations—LNM: Flora** **Machine Hallucinations—LNM: Landscapes** ## The Process **Data Collection Partnerships**—For the Large Nature Model, Refik Anadol Studio collects as much data as they can on the flora, fauna, and fungi of the world’s rainforests. They enhance the breadth of this nature-focused AI research by working with some of the world’s leading organizations, including The Smithsonian Institution, London’s Natural History Museum, and National Geographic, among others. **Physical Data Collection**—In addition to the open archives of their data partners, the team is venturing to 16 unique rainforest locations, deploying data collection technologies ranging from lidar to photogrammetry, and capturing ambisonic audio and high-resolution visuals of these diverse ecosystems. NVIDIA’s neural network algorithms and tools let them create advanced visualizations and graphics. **AI Research**—The LNM’s AI research is built on industry standards and widely accepted diffusion architectures, using tools developed by a wide range of open software communities. Diffusion models in AI generate high-quality images by first adding noise to simple distributions to create a series of increasingly noisy images. Then, a neural network is used to reverse this, systematically removing the noise to reconstruct the original image with clarity. **Generative AI Video Research**—This research integrates the aesthetic principles of the Studio’s multi-year project, Machine Hallucinations, with innovative AI techniques crafted for the LNM’s latest outputs. It showcases a fine-tuned diffusion model for video generation, which draws upon previous fluid simulation videos from Machine Hallucinations. This enhanced model transforms realistic AI video outputs to align with the distinct fluid simulation visuals characteristic of the Refik Anadol Studio. **Building Generative Realities**—Large Nature Model: Living Archive is an immersive, multisensory artwork in a growing body of work that’s created using the LNM. It's the longest continuous generative AI visualization of nature to date, offering vibrant AI simulations of the world’s rainforests in a multi-channel sound and video experience that envelopes viewers in the rainforest flora, fauna, and landscapes. **Audio Classification**—Using a subset of thousands of recordings from Cornell University's Macaulay Library—particularly of Amazonian birds—Anadol and his team use cutting-edge pre-trained audio neural networks for audio tagging. A new audio is generated using a custom audio latent diffusion model trained on a broad spectrum of unlabeled audio, enabling the replication of complex sounds. The power of the Studio's model is its capability to produce realistic and immersive rainforest soundscapes. Previous Next 1. Slide 1 2. Slide 2 3. Slide 3 4. Slide 4 5. Slide 5 6. Slide 6 ## Machine Hallucinations—MoMA ### Series of Artworks from Machine Hallucinations—MoMA Dreams 2021-2022 Fluid Dreams AI Data Paintings Fluid dynamics has been a source of artistic inspiration for Anadol since the inception of his broader project, Machine Hallucinations. The artist’s exploration of digital pigmentation and light-through-fluid solver algorithms are accelerated by GPU computation and real-time ray-traced lighting. This speedup helps manifest his inspiration by showcasing the most innovative methods available to AI-based media artists. [Play Video](#) ### Series of Artworks from Machine Hallucinations—MoMA Dreams 2021-2022 Generative Study I AI Data Paintings Anadol and his team created a unique exhibition of AI data paintings by training an AI model with the public metadata of The Museum of Modern Art’s collection spanning more than 200 years of art. Generative Study is from a series of algorithmic AI data paintings showcasing Anadol’s collaboration with GAN algorithms at the intersection of technology and aesthetics. Each frame in the series displays a cluster of chosen “latent space sequences,” as the artist goes through serendipitous allusions to modern visual expressions in the machine-mind. [Play Video](#) Previous Next 1. Slider 1 2. Slider 2 ## Data Paintings [![](https://images.nvidia.com/deep-learning-ai/ai-art-gallery/refik-anadol/paintings/gtc22-spring-refik-anadol-data-painting-a-3c33-d.jpg)](#data-paintings-1) [### Machine Hallucinations MoMA Dreams — A](#data-paintings-1) [![](https://images.nvidia.com/deep-learning-ai/ai-art-gallery/refik-anadol/paintings/gtc22-spring-refik-anadol-data-painting-b-3c33-d.jpg)](#data-paintings-2) [### Machine Hallucinations MoMA Dreams — B](#data-paintings-2) [![](https://images.nvidia.com/deep-learning-ai/ai-art-gallery/refik-anadol/paintings/gtc22-spring-refik-anadol-data-painting-c-3c33-d.jpg)](#data-paintings-3) [### Machine Hallucinations MoMA Dreams — C](#data-paintings-3) [![](https://images.nvidia.com/deep-learning-ai/ai-art-gallery/refik-anadol/paintings/gtc22-spring-refik-anadol-data-painting-d-3c33-d.jpg)](#data-paintings-4) [### Machine Hallucinations MoMA Dreams — D](#data-paintings-4) [![](https://images.nvidia.com/deep-learning-ai/ai-art-gallery/refik-anadol/paintings/gtc22-spring-refik-anadol-data-painting-e-3c33-d.jpg)](#data-paintings-5) [### Machine Hallucinations MoMA Dreams — E](#data-paintings-5) [![](https://images.nvidia.com/deep-learning-ai/ai-art-gallery/refik-anadol/paintings/gtc22-spring-refik-anadol-data-painting-f-3c33-d.jpg)](#data-paintings-6) [### Machine Hallucinations MoMA Dreams — F](#data-paintings-6) [![](https://images.nvidia.com/deep-learning-ai/ai-art-gallery/refik-anadol/paintings/gtc22-spring-refik-anadol-data-painting-g-3c33-d.jpg)](#data-paintings-7) [### Machine Hallucinations MoMA Dreams — G](#data-paintings-7) [![](https://images.nvidia.com/deep-learning-ai/ai-art-gallery/refik-anadol/paintings/gtc22-spring-refik-anadol-data-painting-h-3c33-d.jpg)](#data-paintings-8) [### Machine Hallucinations MoMA Dreams — H](#data-paintings-8) [![](https://images.nvidia.com/deep-learning-ai/ai-art-gallery/refik-anadol/paintings/gtc22-spring-refik-anadol-data-painting-i-3c33-d.jpg)](#data-paintings-9) [### Machine Hallucinations MoMA Dreams — I](#data-paintings-9) ## The Process Machine Hallucinations is a multi-year research project from Refik Anadol Studio (RAS) that investigates data aesthetics based on collective visual memories of humanity. Machine Hallucinations - MoMA expands on this vision by processing 138,151 pieces of metadata from the entire MoMA archives in the mind of a machine. Data Universe - MoMA is a global AI Data Painting simulating a latent walk among the museum’s digitized collection. It combines RAS’s vision of handling data within a universe that it creates for itself with their approach to data visualization’s latent space as a locus for never-ending, self-generating contemplation. Using NVIDIA StyleGAN2 ADA to capture the machine’s “hallucinations” of MoMA’s vast archive of modern art in a multi-dimensional space, RAS trains a unique AI model with subsets of the collection, creating embeddings in 1024 dimensions. These hallucinations construct new aesthetic images and color combinations through unique lines drawn by algorithmic connections. For MoMA - Fluid Dreams, Refik Anadol Studio’s signature fluid dynamics algorithm infinitely dreams about the MoMA archive. RAS synthesizes the vast data collected from MoMA archives into ethereal data pigments, and eventually into a representational form of fluid-inspired movements with the help of custom software and generative algorithms. Previous Next 1. Slide 1 2. Slide 2 3. Slide 3 4. Slide 4 ## About Refik Anadol Refik Anadol (b. 1985, Istanbul, Turkey) is an internationally renowned media artist, director, and pioneer in the aesthetics of data and machine intelligence. He’s the Director of Refik Anadol Studio in Los Angeles and teaches at UCLA’s Department of Design Media Arts. Anadol’s work locates creativity at the intersection of humans and machines. Taking the data that surrounds us as primary material and the neural network of a computerized mind as a collaborator, Anadol offers us radical visualizations of our digitized memories and expands the possibilities of interdisciplinary arts. Anadol’s site-specific data paintings and sculptures, live audio/visual performances, and immersive installations take many forms, while encouraging us to rethink our engagement with the physical world, collective experiences, public art, decentralized networks, and the creative potential of AI. Alex Morozov, Alistair Ramage, Caio Villela, Christian Burke, Cosku Turhan, Daniel Seungmin Lee, Dasha Orlova, Delaney Kough, Dogukan Yesilcimen, Efe Mert Kaya, Efsun Erkilic, Ho Man Leung, Hye Min Cho, Jules Hyun, Kelian Maissen, Kerim Karaoglu, Kyle McLean, Laura Cohen, Linda Berke, Maurizio Braggiotti, Mert Cobanov, Michael Hsiu, Nidhi Parsana, Pelin Kivrak, Sebastian Monroy, Simon Burke, Tobias Heinemann, Xi Wang, and Yufan Xie [Website](http://refikanadolstudio.com/) | [Instagram](http://instagram.com/refikanadol) | [X](http://twitter.com/refikanadol) | [Vimeo](http://vimeo.com/refo) | [YouTube](http://youtube.com/user/refikanadol) | [LinkedIn](http://linkedin.com/in/refikanadol/) | [Gallery](http://nft.refikanadol.com/) ## Featured Sessions ## AI and the Data Universe: Refik Anadol’s Artwork for Global Impact Refik Anadol and members of his studio discuss with NVIDIA’s Brian Dowdy how their unique AI workflow, as well as NVIDIA GPUs, SDKs, and state-of-the-art AI models, help use their work to make a positive impact on the world. [Watch Session](https://www.nvidia.com/gtc/session-catalog/?tab.allsessions=1700692987788001F1cG&search=S63138#/) ## Generating Modern Masterpieces: MoMA Dreams Become a Reality Join multimedia artist and director, Refik Anadol, and Museum of Modern Art curators, Michelle Kuo and Paola Antonelli, to explore the evolution of technology in the making of modern art. [Watch Session](https://www.nvidia.com/en-us/on-demand/session/gtcspring23-s52060/) ## Building an AI Data Sculpture from Brain Waves Refik Anadol Studio (RAS) embarks upon a new journey to explore the architecture of the human brain by combining advanced neuroimaging techniques with cutting-edge AI and multi-modal data visualization tools. [Watch Session](https://www.nvidia.com/en-us/on-demand/session/gtcspring21-art3339/) ## Virtual Studio Tour with Refik Anadol Go behind the scenes with one of our amazing AI Art Gallery artists for a virtual tour of their innovative studios, and learn how AI has helped shape their creative process. [Watch Session](https://www.nvidia.com/en-us/on-demand/session/gtcfall20-a22161/) ## More Artists [### 64/1 and Harshit Agrawal Custom datasets, GANs, and image classification](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/harshit-agrawal.md) [### AIVA Artificial intelligence virtual artist-composed music](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/aiva.md) [### Allison Parrish AI-generated poetry](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/allison-parrish.md) [### Amelia Winger-Bearskin Questioning ownership of the sky through AI-altered landscapes](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/amelia-winger-bearskin.md) [### Anna Ridler Visualization of time through AI-generated flowers](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/anna-ridler.md) [### Daniel Ambrosi Computational photography and artificial intelligence](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/daniel-ambrosi.md) [### Emanuel Gollob Interactive robotic installation](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/emanuel-gollob.md) [### Fashion Innovation Agency, London College of Fashion The Machine Muses: AI in Fashion](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/fashion-innovation-agency.md) [### Helena Sarin Generative mixed media](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/helena-sarin.md) [### Holly Herndon and Mat Dryhurst Performance and multichannel A/V with TensorFlow](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/holly-herndon.md) [### Instant NeRF Artists Transformation of 2D photos into 3D scenes with AI](https://www.nvidia.com/en-us/research/ai-art-gallery/instant-nerf.md) [### Jesse Woolston Machine learning-generated, multi-sensory art experiences](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/jesse-woolston.md) [### Lunar Ring Human-machine interactive experience](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/lunar-ring.md) [### Lyricstudio AI-trained lyrics inspiration engine](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/lyricstudio.md) [### Madeline Gannon Robot-taming interactive performance](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/madeline-gannon.md) [### Nao Tokui and Qosmo AI-guided music experiments](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/nao-tokui-and-qosmo.md) [### Oxia Palus GANs, GPUs, and multispectral imaging](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/oxia-palus.md) [### Pindar Van Arman Algorithmic robotic painting](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/pindar-van-arman.md) [### Refik Anadol AI Data Paintings and Sculptures](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/refik-anadol.md) [### Scott Eaton Generative AI drawing and sculpture](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/scott-eaton.md) [### Sofia Crespo and Entangled Others 3D GAN and 3D style-transfer](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/sofia-crespo.md) [### Stephanie Dinkins Conversational deep learning chatbot](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/stephanie-dinkins.md) [### Ting Song AI-visualized Chinese poetry with GANs and Kaolin](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/ting-song.md) [### Vanessa Rosa The story of physical-turned-digital AI-powered avatars](https://www.nvidia.com/en-us/research/ai-art-gallery/artists/vanessa-rosa.md) Previous Next 1. x 2. Area ReSTIR: Resampling for Real-Time Defocus and Anti-Aliasing 3. Walkin’ Robin: Walk on Stars With Robin Boundary Conditions 4. Decorrelating ReSTIR Samplers via MCMC Mutations 5. fVDB: A Deep Learning Framework for Sparse, Large-Scale, and High-Performance Spatial Intelligence 6. From Microfacets to Participating Media: A Unified Theory of Light Transport With Stochastic Geometry 7. A Free-Space Diffraction BSDF 8. NeuralVDB: High-Resolution Sparse Volume Representation Using Hierarchical Neural Networks 9. ConsiStory: Training-Free Consistent Text-to-Image Generation 10. Modeling Hair Strands With Roving Capsules 11. SuperPADL: Scaling Language-Directed Physics-Based Control With Progressive Supervised Distillation 12. A Differential Monte Carlo Solver for the Poisson Equation 13. Diffusion Texture Painting 14. Simplicits: Mesh-Free, Geometry-Agnostic, Elastic Simulation 15. Stabler Neo-Hookean Simulation: Absolute Eigenvalue Filtering for Projected Newton 16. Fluid Control With Laplacian Eigenfunctions 17. Real-Time Neural Appearance Models 18. Surface-Filling Curve Flows via Implicit Medial Axes 19. Interactive Character Control With Auto-Regressive Motion Diffusion Models 20. Flexible Motion In-Betweening With Diffusion Models 21. Flexible Motion In-Betweening With Diffusion Models 22. Flexible Motion In-Betweening With Diffusion Models 23. Flexible Motion In-Betweening With Diffusion Models 24. Flexible Motion In-Betweening With Diffusion Models 25. Vanessa Rosa