Fusion Landscape and Beyond is an interdisciplinary art project, exploring the role of memory in imagination and creation through Artificial Intelligence(AI) and Chinese landscape painting. Inspired by the theory of Cultural Memory, which emphasizes the process of selectively retrieving and updating memory with the responses to the ever-changing present situations or concerns, we find that the current text-to-image AI algorithms can be thought of as analogous to this theory of Cultural Memory, in that they react to diverse human instructions with relatively fixed knowledge, i.e. checkpoints in pre-trained models. It shows different results based on both its formed dataset and the current inputs, producing more fantastic and unbridled images compared with the original ones in its dataset. In addition, cultural memory is also a kind of institution that exists only in disembodied form and requires preservation and re-embodiment. In response to this analogy and the three key components of the cultural memory – tradition, transmission, and transference – we propose a novel concept called AI memory and link that to the Chinese landscape painting (Shan-Shui-Hua), a synthetic embodiment of creativity derived from the artist's memory.
This project contains a video animation titled The Faded Landscape(2022), and a series of still images created with our fine-tuned AI text-to-image generation models. In our artworks, we combine the symbolic elements of landscapes that span time and space to represent the past, present, and future through “AI memory”.
The Exotic Fusions (2022). Fine art paper print, AI generated image, 64 × 22 inches.
The format of display is inspired by traditional Chinese bookbinding technique called the dragon scale bookbinding and a 17th century European composition technique, tabula scalata. Both techniques involve fusing different sets of disconnected elements into one harmonious entity.
From Landscape to Landscape NO. 1 | New York (2022). Fine art matte paper print, AI generated image, 22.05 × 47.25 inches.
The concrete jungle is here. Can you discern city forms in the mountain and river landscape?
From Landscape to Landscape NO.2 | Shanghai (2022). Fine art matte paper print, AI generated image, 22.05 × 47.25 inches.
The floating characters display AI’s effort to learn and imitate scripts of Chinese calligraphy. It is more of an artistic than semantic creation. We call it the “calligraphy of AI.”
From Landscape to Landscape NO.3 | Future? (2022). Fine art matte paper print, AI generated image, 22.05 × 47.25 inches.
What will the future city / landscape look like?
From Landscape to Landscape NO.4 | Future+ (2022). Fine art matte paper print, AI generated image, 22.05 × 47.25 inches.
What harmonious dance between city and nature will define the future, where concrete jungles blend seamlessly with the lush embrace of forests, and the gentle hum of technology is accompanied by the melodious symphony of birdsong?
As an extension of this project, we created an AI Chinese Poem & Painting Generator, Idyllic, bridging the Chinese poem and painting through "AI memory". Click here to see the documentation of the generator.
Artist: Mingyong Cheng
Art Historian & Research and Exhibition Support: Xuexi Dang
Chinese Research Support: Xingwen Zhao
Technical Support: Zetao Yu
Project advised by Memo Akten & Kuiyi Shen
Music: 鹤舞白沙 and 是我的脑海还是云的末端 by Amingliang
Sound/Recording: Water Lap by snog, Rain Moderate by InspectorJ, Road Traffic Environment Ambix by KilianJosselinAlba
This project used open-source AI generated art tools Stable Diffusion and Deforum.
- 37th 2023 NeurIPS Creative AI Track
- 28th Split Film Festival/International Festival of New Film
- CVA Gallery, University of Toledo, Ohio
- 26th Antimatter [media art], Victoria BC Canada
- Here and Now, 8th Digerati Emergent Media Festival, Denver, CO
- 7/2023, FEMeeting: Women in Art, Science and Technology, Taos, NM
- 6/09-8/28 2023, Beyond the Horizon, Annmarie Sculpture Garden & Arts Center, Solomons, MD
- 9/12/2022 at 2022 36th NeurIPS Conference: Workshop on Machine Learning for Creativity and Design Art Gallery (Online)
- 10/27/2022 at California Institute for Telecommunications and Information Technology: Vroom - Initiative for Digital Exploration of Arts and Sciences (IDEAS), Qualcomm Institute