《息壤纪·禁止生成西湖》前传概念版 Original

In the preliminary stage, we built the video’s world setting and completed prequel clips. Follow-up materials are under production. See the attachment for further introduction.
Title: EARTH AT REST: WEST LAKE, ELSEWHERE

Do Not Generate West Lake is an AIGC experimental moving-image work shaped by an Eastern vision of the future. It forms the first chapter of the Xi Rang Chronicle series.

The story takes place in 2226. Following the Earth Relief Agreement of 2186, industries with high energy consumption, heavy resource use and major ecological impact have gradually been relocated off Earth. Professional engineers now spend years working at off-world stations before returning home on leave. The protagonist, a Hangzhou-born engineer raised during the “Xi Rang era,” is reminded of home by a simple weather update: “Hangzhou today: cloudy, clearing later.” He enters an AI world model, hoping only to walk through his hometown for a while.

The AI first generates a spectacular futuristic Hangzhou stripped of ordinary life. It then produces a technologically enhanced West Lake assembled from standard landmarks, tourist viewpoints and familiar visual clichés. When the protagonist says, “Do not generate West Lake,” a nearly imperceptible fracture in time occurs. Without informing him—or the audience—the AI stops generating the city and connects the simulation to Hangzhou’s real-time digital twin.

Only later, inside a quiet teahouse, does the protagonist see the live digital presence of his ageing parents. He can see them, but they cannot see him.

The title does not reject the real West Lake. It rejects a hometown reduced to landmarks, labels and averaged visual data. “Xi Rang” refers to the self-growing earth found in ancient Chinese mythology and also suggests a future civilisation that has finally allowed the Earth to rest. Through the hidden transition from generated imagery to real-time imagery, the work asks whether AI can truly remember a city—or whether it can only reproduce the city that people collectively expect to see.

The film builds two contrasting visual systems. The generated Hangzhou is polished, monumental, standardised and visually spectacular. After the temporal fracture, the imagery gradually becomes more tactile and lived-in: humid air, natural grain, ambient sound, hand-rowed boats, teahouses and forest paths. Technology is not presented as an exhibition of devices. It remains embedded within space, time and everyday life. The temporal fracture functions both as a transition and as visual evidence of the narrative’s concealed truth.

AI image-generation and dynamic-visual tools are used to support environmental concept development, future-city construction and selected motion sequences. Compositing, editing, colour grading and sound design are then used to maintain continuity across characters, spaces and narrative layers. AI expands the range of future environments that cannot be filmed directly, while the final selection of images, narrative structure, pacing and aesthetic decisions remain under the control of the creator.

The work can be presented as a single-channel experimental film or expanded into a multi-screen installation, immersive projection, urban cultural exhibition or digital-twin experience. In an installation format, the generated city and the real-time city could appear simultaneously, allowing audiences to experience the distance between algorithmic imagination, urban reality and personal memory.

By using West Lake as a familiar public entry point into an unfamiliar future, the work responds to the Experimental Moving Image track’s emphasis on future visual languages, innovative image-based storytelling, technological aesthetics, communication potential and public engagement.

Production & Asset Notes:
Scene concepts for this work were refined using GPT large language model and MrcGeoStudio, our self-built script Skill. Images were generated via Image2 and Banana models. For audio-visual production, we primarily adopt the state-of-the-art SD2.5 model, with Kling O3 and Minimax H3 employed for image generation and animation. DaVinci Resolve was used for editing, compositing and color grading.