Works37 works
◉ LiveFundamentAI
Shown at the 19th Venice Architecture Biennale: AI accelerates the symbiosis between the lagoon's century-old timber posts and the life growing on them, rendered as hybrid totems of the future.
AI generation, 3D printing, projection mappingOpen →
One and Three Objects, and An Attempt at Exhausting the Object
Project Title: One and Three Objects, and An Attempt at Exhausting the Object
Interactive AI installation: camera, everyday objects, multi-screen generationOpen →
◉ LiveThe 48th Generation
Forty-eight real generations of E. coli from a synthetic-biology "bacterial treadmill" experiment, rewritten as 48 day–night cycles of a game-like generative installation in which every bacterium is a translucent architectural organism. With Wu Ziyang.
Three-channel real-time generative system, game engine; with Wu ZiyangOpen →
◉ LiveGranted Hours: A Non-human Timetable
Every night from 03:17 to 04:17, an AI agent is granted one hour of free creation; for the other twenty-three it remains a tool. An ongoing archive of human–machine time.
Ongoing human–AI archive; interactive works generated daily by an AI agentOpen →
◉ LiveWho Is Human (Open Source)
A cosplay game in which humans play AI, open-sourced: a human hides among AIs pretending to be one. Playable online.
Open →
Worldline Overload {Choose to Patch |or| Forced Reboot}
Some non-linear memos on worldline fluctuations:
Open →
◉ LiveWho Is Human: a Reverse Turing Test
A cosplay game in which humans play AI: in a reverse Turing test run by AIs, you must pass as one of them and survive every round of judgement.
Open →
Microcosmic Universe
"Microcosmos" constructs an animated morphing universe from microscopic captures, transforming cross-sections of pine roots into flowing galactic ecosystems through AI mediation.
Open →
◉ Live3D Bionic Designer
An AI design agent that learns from biological form, producing a design concept, a design statement, 2D images and a 3D model in one chain of thought.
AI agent, ComfyUI + Gradio, open sourceOpen →
Poems Against Common Sense
Large language models predict the next token from statistics, or "common sense". What happens if you ask one to write a poem that defies linguistic common sense?
Open →
Dimensional Recasting
The "Dimensional Recasting" series of works explores the intersection between the compatibility of real dimensions and imagination.
Open →
MicroCosmos
"MicroCosmos" is an artistic project that focuses on the grand reconstruction of information at a microscopic scale.
Open →
Boundless Intelligence
"Boundless Intelligence" is an experimental project that transcends disciplinary boundaries, aiming to explore design and artistic creation under the interaction of diverse intelligences.
Open →
RE-Imaginate Nature
"RE-Imaginate nature" delves into how artificial intelligence is utilized and redefines natural elements in the ongoing process of human self-transformation in pursuit of development, opening up new possibilities for expression in design and art.
Open →
Microcosmic Drift
Starting from micro-photographs taken on a phone, AI regrows cells and fibres into strange living things; shown at the Inclusion Conference on the Bund, Shanghai.
Smartphone micro-photography and AI generationOpen →
The Garden of Robotic Delights
My 2022 creation, The Garden of Robotic Delights, will be featured in the upcoming 2024 London Computer Arts Society (CAS) Members' Exhibition at Moorgate, London.
Open →
◉ LiveEdge of Intelligence: the CyberGenix Autonomous Creator
An AI agent that finds its own inspiration every day, writes its own prompts and generates its own images and videos; the human only reviews and publishes.
Autonomous AI agent; images and videoOpen →
Beyond Appearances
CVPR AI Art Gallery 2024 artwork with Tutorial
Smartphone micro-photography, AI videoOpen →
◉ LiveGaussian Space
I, a programming novice, have created a parallel space-time portal called GaussianSpace, a tool that allows users to guide the editing of three-dimensional Gaussian splatting (3D Gaussian Splatting) in large scenes using text. 🐶
Open →
Silicon-based Lifeform: Fusion Reactor Organism
Try creating an animation of silicon-based lifeforms using a combination of artificial evolution guidance, GPT-4/ChatGLM, Stable Diffusion/Midjourney, and Runway GEN2. It feels like we can create worlds from scratch with AI!
Open →
Latent Timespace No.5: Cloudscape Artistry
An traditional Chinese art model trained by StyleganXL
Open →
Silicon-based Lifeform: Diatom Cloud Jester
Try to create a silicon-based creature atlas animation using the method of artificial evolution guidance + GPT4/chatGLM + stable diffusion / Midjourney + Runway GEN2. It feels like AI can create the world from scratch now! 🥰
Open →
Xiaomi × Daniel Arsham Launch Film
In November 2022, I had the honor of being invited to create a promotional video using AI for the limited international edition smartphone Xiaomi 12T PRO, in collaboration with the famous American artist Daniel Arsham. I'm excited to share the creative process and the AI technologies used!
Open →
Chevrolet Spring Festival Film 2023
The power of AI creation tools is well-known, but pinpointing customer needs among a vast array of possibilities and ensuring the output meets industrial standards remains a challenging task. My collaboration with Chevrolet and Quanpingben Shi has proven that it is feasible, yet it requires the deep involvement of designers who understand AI.
Open →
Unnamed Flower
If every flower holds a world, does every world hold a flower? From a 512-dimensional latent space learned from 3,000 botanical drawings, endless nameless flowers grow.
GAN trained on 3,000 public-domain botanical plates; latent-space interpolation videoOpen →
The Non-human City
My work "The Non-human City Driven by AI and Ecological Environment" will be featured in an exciting exhibition—potenziale 3—within the sub-unit "Bio Lab: Living Space." This is an interdisciplinary research laboratory aimed at rethinking how we coexist with the environment.
Open →
◉ LiveQuantum Field
The beauty of quantum mechanics, interpreted artistically: particles driven by multiple periodic equations pull on one another, simulating a quantum chaos that resists direct simulation.
On-chain generative art, edition of 500 (Genify / Lambda)Open →
◉ LiveMicrocosmic City
A deep dive into the microscopic world: through layer after layer of cellular structure to a city hidden inside.
Stable Diffusion (DreamBooth fine-tune) and Deforum animationOpen →
◉ LiveDreamFields 3D
An open-source toolkit that "dreams" colourful 3D objects from text and image prompts, exporting meshes, turntable videos and NeRF instances.
NeRF + CLIP; open-source tool (Colab)Open →
◉ LiveClouds of Quantum Particles
Generative art inspired by the hydrogen atom's electron cloud: quantum particles orbit, oscillate and collapse, and every hash yields a different cloud.
p5.js generative art on fxhash (Tezos)Open →
Letters from Another Dimension
Thirty-six phrases written collectively by an on-chain community, handed to a diffusion model: thirty-six semantically blurred images, replies from another dimension.
Diffusion-generated images, 36 piecesOpen →
Guardians of the Ocean Natives
A collaboration with National Geographic China: the iridescent colours of oil slicks fused with National Geographic's ocean photography. The more beautiful, the more dangerous.
AI style fusion and hand drawing; digital collectible seriesOpen →
Biodesign: How Nature Inspires Design
My work was featured in the Biodesign section of the 2022 Venice - Metaverse Art Annual Exhibition, the section of Bio-Design
Open →
◉ LiveLatent Timespace No.1: The Illusionary Sands of Time
A grain of sand could be a world, and a world could polish a grain of sand. The first piece of Latent Timespace, glimpsing parallel-universe scenery through AI.
AI-generated pixel animation, edition of 25 (Teia / Tezos)Open →
Cyber Marine Evolution
Two hundred computer-generated marine creatures that do not exist, each becoming a unique invitation for a guest of Beyond Nature in London.
GAN-generated images and animation, 200 piecesOpen →
Through Every Apocalypse to the End of the Universe
Using AI to traverse all possibilities and face humanity's doomsday.
Open →
◉ LiveExploring the Latent Space of Facades
StyleGAN2 generates plausible building facades without any conditional input; the latent space is then steered, visualised and explained.
StyleGAN2 and GANSpace on 9,772 facade images; DigitalFUTURES 2021 paperOpen →
- WK-322025
◆ PinnedFundamentAIShown at the 19th Venice Architecture Biennale: AI accelerates the symbiosis between the lagoon's century-old timber posts and the life growing on them, rendered as hybrid totems of the future.
3D→ - WK-302025
◆ PinnedOne and Three Objects, and An Attempt at Exhausting the ObjectProject Title: One and Three Objects, and An Attempt at Exhausting the Object
Interactive→ - WK-372026
The 48th GenerationForty-eight real generations of E. coli from a synthetic-biology "bacterial treadmill" experiment, rewritten as 48 day–night cycles of a game-like generative installation in which every bacterium is a translucent architectural organism. With Wu Ziyang.
Generative→ - WK-362026
Granted Hours: A Non-human TimetableEvery night from 03:17 to 04:17, an AI agent is granted one hour of free creation; for the other twenty-three it remains a tool. An ongoing archive of human–machine time.
Interactive→ - WK-352025
Who Is Human (Open Source)A cosplay game in which humans play AI, open-sourced: a human hides among AIs pretending to be one. Playable online.
Interactive→ - WK-342025
Worldline Overload {Choose to Patch |or| Forced Reboot}Some non-linear memos on worldline fluctuations:
Video→ - WK-332025
Who Is Human: a Reverse Turing TestA cosplay game in which humans play AI: in a reverse Turing test run by AIs, you must pass as one of them and survive every round of judgement.
Interactive→ - WK-312025
Microcosmic Universe"Microcosmos" constructs an animated morphing universe from microscopic captures, transforming cross-sections of pine roots into flowing galactic ecosystems through AI mediation.
Video→ - WK-292024
3D Bionic DesignerAn AI design agent that learns from biological form, producing a design concept, a design statement, 2D images and a 3D model in one chain of thought.
3D→ - WK-282024
Poems Against Common SenseLarge language models predict the next token from statistics, or "common sense". What happens if you ask one to write a poem that defies linguistic common sense?
Generative→ - WK-272024
Dimensional RecastingThe "Dimensional Recasting" series of works explores the intersection between the compatibility of real dimensions and imagination.
Image→ - WK-262024
MicroCosmos"MicroCosmos" is an artistic project that focuses on the grand reconstruction of information at a microscopic scale.
Image→ - WK-252024
Boundless Intelligence"Boundless Intelligence" is an experimental project that transcends disciplinary boundaries, aiming to explore design and artistic creation under the interaction of diverse intelligences.
Image→ - WK-242024
RE-Imaginate Nature"RE-Imaginate nature" delves into how artificial intelligence is utilized and redefines natural elements in the ongoing process of human self-transformation in pursuit of development, opening up new possibilities for expression in design and art.
Image→ - WK-232024
Microcosmic DriftStarting from micro-photographs taken on a phone, AI regrows cells and fibres into strange living things; shown at the Inclusion Conference on the Bund, Shanghai.
Image→ - WK-222024
The Garden of Robotic DelightsMy 2022 creation, The Garden of Robotic Delights, will be featured in the upcoming 2024 London Computer Arts Society (CAS) Members' Exhibition at Moorgate, London.
Image→ - WK-212024
Edge of Intelligence: the CyberGenix Autonomous CreatorAn AI agent that finds its own inspiration every day, writes its own prompts and generates its own images and videos; the human only reviews and publishes.
Generative→ - WK-202024
Beyond AppearancesCVPR AI Art Gallery 2024 artwork with Tutorial
Video→ - WK-192023
Gaussian SpaceI, a programming novice, have created a parallel space-time portal called GaussianSpace, a tool that allows users to guide the editing of three-dimensional Gaussian splatting (3D Gaussian Splatting) in large scenes using text. 🐶
3D→ - WK-182023
Silicon-based Lifeform: Fusion Reactor OrganismTry creating an animation of silicon-based lifeforms using a combination of artificial evolution guidance, GPT-4/ChatGLM, Stable Diffusion/Midjourney, and Runway GEN2. It feels like we can create worlds from scratch with AI!
Video→ - WK-172023
Latent Timespace No.5: Cloudscape ArtistryAn traditional Chinese art model trained by StyleganXL
Video→ - WK-162023
Silicon-based Lifeform: Diatom Cloud JesterTry to create a silicon-based creature atlas animation using the method of artificial evolution guidance + GPT4/chatGLM + stable diffusion / Midjourney + Runway GEN2. It feels like AI can create the world from scratch now! 🥰
Video→ - WK-152023
Xiaomi × Daniel Arsham Launch FilmIn November 2022, I had the honor of being invited to create a promotional video using AI for the limited international edition smartphone Xiaomi 12T PRO, in collaboration with the famous American artist Daniel Arsham. I'm excited to share the creative process and the AI technologies used!
Video→ - WK-142023
Chevrolet Spring Festival Film 2023The power of AI creation tools is well-known, but pinpointing customer needs among a vast array of possibilities and ensuring the output meets industrial standards remains a challenging task. My collaboration with Chevrolet and Quanpingben Shi has proven that it is feasible, yet it requires the deep involvement of designers who understand AI.
Video→ - WK-132022
Unnamed FlowerIf every flower holds a world, does every world hold a flower? From a 512-dimensional latent space learned from 3,000 botanical drawings, endless nameless flowers grow.
Video→ - WK-122022
The Non-human CityMy work "The Non-human City Driven by AI and Ecological Environment" will be featured in an exciting exhibition—potenziale 3—within the sub-unit "Bio Lab: Living Space." This is an interdisciplinary research laboratory aimed at rethinking how we coexist with the environment.
Image→ - WK-112022
Quantum FieldThe beauty of quantum mechanics, interpreted artistically: particles driven by multiple periodic equations pull on one another, simulating a quantum chaos that resists direct simulation.
Generative→ - WK-102022
Microcosmic CityA deep dive into the microscopic world: through layer after layer of cellular structure to a city hidden inside.
Video→ - WK-092022
DreamFields 3DAn open-source toolkit that "dreams" colourful 3D objects from text and image prompts, exporting meshes, turntable videos and NeRF instances.
3D→ - WK-082022
Clouds of Quantum ParticlesGenerative art inspired by the hydrogen atom's electron cloud: quantum particles orbit, oscillate and collapse, and every hash yields a different cloud.
Generative→ - WK-072022
Letters from Another DimensionThirty-six phrases written collectively by an on-chain community, handed to a diffusion model: thirty-six semantically blurred images, replies from another dimension.
Image→ - WK-062022
Guardians of the Ocean NativesA collaboration with National Geographic China: the iridescent colours of oil slicks fused with National Geographic's ocean photography. The more beautiful, the more dangerous.
Video→ - WK-052022
Biodesign: How Nature Inspires DesignMy work was featured in the Biodesign section of the 2022 Venice - Metaverse Art Annual Exhibition, the section of Bio-Design
Image→ - WK-042021
Latent Timespace No.1: The Illusionary Sands of TimeA grain of sand could be a world, and a world could polish a grain of sand. The first piece of Latent Timespace, glimpsing parallel-universe scenery through AI.
Video→ - WK-032021
Cyber Marine EvolutionTwo hundred computer-generated marine creatures that do not exist, each becoming a unique invitation for a guest of Beyond Nature in London.
Video→ - WK-022021
Through Every Apocalypse to the End of the UniverseUsing AI to traverse all possibilities and face humanity's doomsday.
Video→ - WK-012021
Exploring the Latent Space of FacadesStyleGAN2 generates plausible building facades without any conditional input; the latent space is then steered, visualised and explained.
Image→
Nothing matches