Living Florals
/ Research
/AIGC
/Cultural Heritage
/AIGC
/Cultural Heritage
Artificial Intelligence Generated Content (AIGC) is increasingly applied in cultural heritage, from restoring damaged artifacts to reimagining traditional crafts such as batik, ink painting, and ceramics. It also supports new cultural dissemination and learning through VR and interactive platforms. At the same time, research shows that AIGC and AI technologies can benefit older adults by enhancing well-being, personalization, and digital inclusion. However, most work focuses on emotional value, industry use, or technical design, while little attention has been paid to how AIGC can connect cultural heritage knowledge with elderly learning.
Extraction of Cultural Genes
We applied the framework to Song Dynasty flower arrangements, extracting cultural “genes” from texts and paintings across five dimensions—vessels, flowers, colors, forms, and connotations—to build the rule base for generation.
Front-end Interaction Flow
Building on the extracted cultural genes, the platform’s front-end interaction is designed as a four-step learning and creation process—Learn → Select → Create → Validate— guiding users through a complete experience of understanding and generating Song Dynasty flower arrangements.
Learn: Explore historical background and cultural genes.
Select: Choose vessel–flower pairs with embedded aesthetic guidance.
Create: Upload or draw a sketch; the system generates a stylized Song-style image.
Validate: View AI output alongside the original sketch for comparison and reflection.
Implementation in ComfyUI
On the back end, ComfyUI executes the specific image generation tasks. ComfyUI coordinates ControlNet, LoRA, and other modules to process user inputs and generate images that meet the structural and stylistic constraints of the system.
Overall workflow
The system workflow begins with the user selecting a vessel and flowers on the front-end webpage and drawing a sketch. These inputs are sent to the back-end server, which dynamically constructs an API call to a ComfyUI instance. The four layers of the framework then process the inputs collaboratively, generating the final image that is returned to the user.
We conducted a mixed-method user study with six retired participants to evaluate the platform’s usability, learning experience, and cultural engagement.
Key Findings：
Discussion & Future Work：
The platform demonstrates strong potential for helping older adults learn and practice Song-style flower arrangement through AIGC. Participants showed clear interest, cultural engagement, and enjoyment. However, several usability challenges remain. The sketch-based creation tool was the most common barrier, requiring fine motor control and causing frustration for many older users. These interaction difficulties affected task completion and limited the system’s overall effectiveness. In addition, accessibility features—such as text size, clarity of instructions, and visual feedback—need further refinement to better support age-related cognitive and physical needs.
Going forward, we will focus on several key improvements:
• Simplify the creative module by replacing freehand sketching with component-based or guided composition tools.
• Enhance accessibility through larger text, clearer labels, stronger visual contrast, and more intuitive interaction flows.
• Add multimodal guidance, including audio prompts, step-by-step tutorials, and real-time feedback to support learning.
• Introduce save/share functions so users can keep, revisit, or showcase their creations.
• Expand to a 3D interactive version for more realistic spatial understanding of flower arrangement aesthetics.
These updates aim to make the platform easier to use, more inclusive, and more immersive for older adults.