top of page
Dunhuang Galaxy
Fusing Dunhuang's celestial beings with cosmic skies via AI art, our 'Dunhuang Cosmos' merges Eastern and Western cultures. Using TouchDesigner's visuals and Arduino's interaction, we craft an emotive, multi-layered narrative, inviting a dynamic dialogue between audience and art.
General View
Design opportunities
How can we use existing AI technology and software to collaborate on innovative breakthroughs in the design field.
Design opportunities
Our theme revolves around the captivating fusion of Dunhuang and the cosmos. The celestial imagery within Dunhuang culture has left an indelible mark, depicting figures that gracefully traverse the skies, alternating between dance and music, creating an enthralling narrative. In the existing Dunhuang murals, the primary focus lies on depicting celestial beings and an array of exotic creatures from the Western Regions, with limited emphasis on the aerial environment.
Simultaneously, in Western art, the starry sky symbolizes a realm of mystery and contemplation. Renowned Western artists like Vincent van Gogh each interpret and understand the vast expanse of stars in their unique ways.
Thus, a daring innovative concept emerges: "Dunhuang Cosmos." By utilizing AI-generated art, we aim to place these celestial beings originating from Dunhuang's Western Regions amidst the cosmic night sky. This amalgamation of Western artistic sentiments with the imagery of Dunhuang culture presents an intricate tableau where Eastern and Western cultures intertwine through time and space.
Leveraging the unique sound visualization capabilities of TouchDesigner, along with a plethora of visual effects, we will infuse life into scenes, orchestrating movement and transformation. This audio-visual fusion aims to evoke emotional resonance deep within the audience, creating multiple layers of storytelling and meaning. The integration of Arduino-driven human-machine interaction further enhances our goals, fostering a dynamic dialogue between visitors and the unfolding artistic expression before them. This promises to be an extraordinary journey of creative exploration.
Design elements choose
Design Tools
· Touchdesigner
· Arduino
· Logic prox
Part of different AI tools test
Reflection of using image AI-- Midjourney
The three smart ai image generation tools all produce images based on keywords and scene descriptions from the point of view of their use. The description of the picture content directly affects the quality of the picture content.
After testing, we found that Midjourney's database is the most powerful, but if we rely only on keywords to determine the art style, the output is more similar to the painting style of the images, and it is not possible to produce works with unique art styles. Adobe Firefly is an experimental project, which is only suitable for creating materials that are available in the database, and does not support uploading images like Midjourney. Therefore, it can't "learn" the style of Dunhuang murals. Runway's model library is more suitable for combining an image to generate a small video, and the quality of its output varies depending on the model selected, making it difficult to continue creating videos in touchdesigner. Therefore, we chose to use Midjourney to create the keyframes.
Reflection of using music AI-- Tuneflow
During the process of testing AI music production software, Mubert allows users to freely input keywords, and the software generates random music based on the provided keywords. However, the generated music tends to have a narrow style and lacks quality, with limited post-production editing capabilities. Tuneflow software enables basic style and rhythm selection, but doesn't allow users to input keywords freely. The software generates random monophonic melodies based on user prompts and supports post-production editing. AIVA software offers a limited selection of keywords, resulting in more mature randomly generated music that can be edited post-production. However, due to its highly polished outputs, ethical concerns might arise.
Final design -- visual display
Image Changes and Integration: The overall image will undergo various transformations between multiple Dunhuang elements using buttons, with the mouse replacing physical buttons as in a real exhibition. Additionally, the elements themselves will undergo dynamic changes through overlapping effects to achieve better integration with the background, creating a particle-like dynamic effect.
Interaction:
Using Arduino to control the brightness of the background image. When people cover the interactive device with their hands, the overall image becomes brighter as more people gather, creating interaction among them.
Presentation:
In the flat display, we configure the overall image into a symmetrical structure suitable for spherical presentation.
Screenshots of project file
Background
Elements part
Whole project
Final Video
bottom of page