Kang Sira

Concept and Point of Departure
This work takes as its starting point the materiality of color and surface structure evident in the paintings of Ayako Rokkaku, expanding painterly matter into the possibility of an environment and a form of civilization through moving images. The core of the work does not lie in reproducing specific paintings, but in translating the layers of paint, diffusion, density, collision, and traces of pressure into the conditions of a world that could plausibly exist.

In this context, color no longer functions as a purely visual element on a flat surface, but operates as a material that forms terrain, structure, boundaries, traces, and even social rules. Thick impasto surfaces accumulate like geological strata; diffusion becomes pathways and borders; hardened ridges of pigment function as laws that organize a form of civil order. Within this material environment, non-human entities move, leave traces, form collectives, and ultimately construct, collapse, and cyclically regenerate a civilization.

Connection to the Hackathon / Exhibition Dataset
This work originates from the density of color, the traces of the hand, and the tension and materiality of surfaces found in Rokkaku’s paintings. Rather than treating the painterly surface as a static image, it is reimagined as a dynamic environment in which the flow of color and the movement of material are already embedded. Within this imagined environment, processes of existence, structure, and social order unfold and are expanded into video form. The exhibition data thus functions not as a visual reference, but as a foundational condition for constructing a world in which material becomes environment and evolves into civilization.

Artistic Intent and Interpretive Direction

Key Elements of Focus

• The thickness and density of paint directly pushed and rubbed by the artist’s fingers
• The flow of color as it collides, absorbs, and forms relationships across the surface
• The structural potential of reading the flat pictorial surface as a form of terrain
• The expansion of painterly traces from expressive marks into environmental conditions and social rules

Method of Interpretation and Translation
Rather than appropriating painterly imagery, this work treats the material properties of painting as a fully realized environment. It poses the central question: if formal elements already present within the painterly surface were to begin moving, what forms of society and civilization would emerge? Diffusion becomes boundary; pigment ridges become law; thick surfaces become the foundation of terrain and architecture; and unexplained bursts of color are translated into mythic events. In this way, the painterly surface is no longer an object of contemplation, but an environment in which events occur. The resulting video constructs a non-human civilizational narrative spanning birth, tribe formation, law, myth, urban accumulation, exploration, conflict, collapse, and cyclical renewal.

Rationale for the Chosen Format
A generative AI-based video format was selected in order to extend the static surface of painting into a domain of temporality and event. This work is not concerned with animating images per se, but with designing the conditions of matter, the laws of a world, the behaviors of entities, and the structures of society through textual prompts. Generative AI enables these conditions to be articulated as narrative-driven prompts and effectively constructs painterly textures and structures as functioning environments.

A key principle in this work is that “only what already exists may move.” Accordingly, the generative process was not left fully open-ended; instead, prompts were precisely controlled so that only formal elements already present within the composed environment images could act as agents of movement. This constraint was essential for maintaining the overall aesthetic coherence of the work.

Production Process

Development and Evolution
The process began by translating the material qualities of Rokkaku’s paintings into images that could be perceived as tangible environments. This involved compositing painterly and material sources. Painterly sources included vivid color collisions, the directional flow of finger-applied paint, thick impasto textures, and layered diffusion. Material sources included the fibrous structure of corrugated cardboard, the cracked and compressed surfaces of clay, and the fractured textures of wood. These elements were combined into a single environmental image, which served as the foundational model for video production.

Based on this environment, scene-specific prompts were developed. These prompts extended beyond simple keywords, incorporating environmental conditions, entity behaviors, social rules, event structures, lighting, and camera composition. The generated scenes were subsequently edited and arranged according to stages of civilization, resulting in a unified narrative of approximately three minutes.

Key Challenges
A primary challenge was the tendency of generative AI to produce human-like characters or unnecessary objects. Elements such as faces, eyes, limbs, anthropomorphic features, and textual objects (e.g., books or documents) conflicted with the conceptual framework and therefore required strict exclusion.

Another challenge was ensuring that the scenes conveyed a sense of a functioning society rather than remaining as abstract motion. To avoid purely decorative or background imagery, the structure was designed to clearly articulate stages such as the formation of tribes, the emergence of boundaries, the establishment of taboo and law, the birth of myth, the accumulation of cities, exploration, conflict, collapse, and cyclical regeneration.

Key Decisions in the Experimental Process

• A global prompt was applied across all scenes to ensure that only pre-existing entities could move
• All anthropomorphic features (faces, eyes, limbs, hands, feet) were excluded
• Linguistic objects (books, documents, text) were excluded
• Textures were maintained at a hyper-real level, while spatial and event structures were composed in a surreal manner
• Prompts were written as extended narrative structures to ensure that each scene represents a stage of civilization
• Sound was designed not as background music, but as an auditory translation of material friction, pressure, fracture, and flow

Technical Description

Tools, Programs, Models, and Workflow
This work was produced using generative AI tools for image and video creation, followed by post-production in video editing software. The workflow is as follows:

• Composition of environmental images based on the textures and material surfaces of Rokkaku’s paintings
• Development of global prompts, negative prompts, and stage-specific narrative prompts
• Input of reference images and prompts into generative AI video tools to produce scenes
• Rearrangement and editing of generated scenes
• Adjustment of transitions, duration, color grading, and credits
• Output in both 9:16 (original) and 16:9 formats, in 4K and FHD resolutions

Image / Video / Text Generation Method
Image generation was based on the creation of composite environmental images, combining multiple material and painterly surfaces into a unified setting. These images functioned as reference models for all subsequent video generation.

Video generation employed both reference image-based and prompt-based methods. Rather than relying solely on text prompts, pre-constructed environment images were combined with textual conditions to control events and movement within each scene.

Text functioned not as descriptive language, but as a form of design script—simultaneously defining environmental laws, entity behavior, material qualities, and constraints.