← Back to stories Vibrant futuristic 3D render featuring abstract shapes and AI themes.
Photo by Google DeepMind on Pexels
虎嗅 2026-04-08

In 23 Days, He Made One Million with an AI Short Film

The film and the creator

A 7‑minute AI short called 《牌子》 earned its maker a 1 million yuan prize and a place on the stage at Bilibili (哔哩哔哩)’s first AI Creation Contest. The filmmaker, who posts as DiDi_OK and works as an animation artist at WPP’s London arm, reportedly completed the piece from script to final render in 23 days; the video broke 10 million views within a week and, reportedly, even The Wandering Earth director Guo Fan left an enthusiastic comment. How did one animator turn a solo experiment into a mainstream hit so quickly?

DiDi_OK treats generative models as a high‑end camera. A standout 20‑second, single‑take intersection shot was created after thousands of iterations; he says the creative work was mostly framing, prompting and curation, not low‑level engineering. The turning point for him was Runway’s Gen‑4: he went from resisting code to evangelising “text‑to‑video” workflows after seeing plausible physics and motion generated by text prompts. His process—write the story, strip form to be easily understood, then use AI for execution—reflects a broader shift from technical gatekeeping to expressive selection.

A wider shift in China’s creator economy

Bilibili data shows China’s appetite for AI content is surging: in Q1 2026 nearly 24 million users watched AI videos daily, with over 120 million monthly views; the contest itself drew more than 8,300 valid submissions, total plays topping 700 million and 143 pieces surpassing one million views. Winners included people with non‑traditional backgrounds—a biology PhD known as 半吊子Bill_ took a top prize in another track—highlighting that the bottleneck is moving from tools to taste, not access. Models such as Seedance, 可灵 and Veo are making “type‑to‑film” feasible; creators who can marry clear structure, voice and a handful of cinematic conventions are winning attention.

Geopolitics and what’s next

This creative opening sits against a tense technology backdrop. Many leading generative tools still rely on Western model releases and high‑end GPUs, and it has been reported that export controls and broader U.S.–China technology rivalry complicate access to top compute. Chinese platforms are responding: Bilibili has launched long‑term programs—including compute subsidies, flow support and an in‑house tool called updream that reportedly remembers individual creators’ styles—to lower friction and keep the workflow local. If anyone can now “make a film,” the real question becomes: who can turn life experience, curation and repeatable taste into sustained audience trust?

AI
View original source →