A solo creator spent $2,800 to make a 94-minute AI sci-fi movie

Imagine producing a full-length, 94-minute photorealistic science fiction movie entirely by yourself. There is no studio lighting crew, no cast of professional actors, no visual effects department, and a production budget of just 20,000 RMB (around $2,800 USD). That is exactly what a digital creator recently accomplished, capturing national attention after spending 18 months creating a complete film adaptation of a story by renowned sci-fi author Liu Cixin.

The feature-length project, based on Liu Cixin's short story Mountain (《山》), rose quickly up the video charts on BilibiliA video platform that grew out of anime, comics, and gaming fandom into China's defining Gen Z video community, famous for its bullet comments. (a major Chinese video-sharing and streaming platform favored by young creators and anime fans). Built using photorealistic AI digital human technology, the 94-minute film represents one of the most ambitious solo AI filmmaking endeavors to emerge from China's internet landscape. It offers an intriguing window into how low-cost artificial intelligence tools are reshaping independent cinema and fandom culture.

The background

To understand the magnitude of this project, it helps to start with the source material and the creator's process. The film is adapted from Mountain (《山》), a science fiction work written by Liu Cixin, the Hugo Award-winning author best known globally for The Three-Body Problem. Liu's stories are famous for grand cosmic concepts, high-concept physics, and intricate world-building—elements that typically require tens of millions of dollars in visual effects budgets to bring to the big screen.

Instead of waiting for a major film studio to option the rights and launch a massive production, a single creator took matters into their own hands. In Chinese internet culture, this process of building complex software, mechanical models, or art projects completely on one's own is often referred to as shoucuo (hand-crafted or solo-built from scratch).

Over the course of 18 months, the creator relied on shoucuo methods to complete the entire pipeline: writing the script adaptation, generating digital environments, generating character performances, and assembling 94 minutes of continuous footage. The total out-of-pocket financial cost was reported at just 20,000 RMB (approx. $2,800 USD), primarily spent on hardware resources and computing costs rather than traditional production overhead. Rather than using stylized 2D anime or simple motion graphics, the creator opted for AI fangzhenren (AI photorealistic digital human technology), aiming for realistic human facial structures, lighting, and textures across the feature-length runtime.

Why it is happening now

The arrival of a 94-minute AI feature film is not an accident; it is the direct outcome of several converging developments in technology, platform ecosystem, and consumer interest within China.

First, generative AI production tools have progressed from creating isolated five-second video clips to supporting long-form visual storytelling. Until recently, AI video generation was prone to heavy visual inconsistency, making character continuity across multi-minute scenes nearly impossible. By spending 18 months fine-tuning digital human models, the creator demonstrated that individual users can now maintain stylistic continuity for a full feature runtime.

Second, platforms like Bilibili provide a unique environment for long-form, user-generated experiments. On Bilibili, audiences actively seek out deep-dive video essays, full fan edits, and indie animation. When the film was uploaded, it quickly earned a spot on the B-zhan paixingbang (Bilibili trending rankings), demonstrating that internet users are willing to sit through an entire 94-minute cinematic experience created by a solo amateur author.

Third, Liu Cixin’s literature has inspired a massive ecosystem of fan creators across China. As official screen adaptations of sci-fi novels often take years to develop, tech-savvy fans are turning to AI tools to manifest their own visual interpretations of complex literary worlds.

Why it matters beyond China

For audiences and filmmakers outside China, this 94-minute AI adaptation serves as a striking proof-of-concept for the future of global indie film production.

Historically, feature-length sci-fi cinema was restricted to major Hollywood studios or heavily subsidized production houses due to the astronomical costs of CGI, set design, and hiring cast members. The fact that a single person, operating with 20,000 RMB and a laptop setup, produced a 94-minute photorealistic sci-fi feature signals a dramatic collapse in the barrier to entry for long-form cinematic storytelling.

It also highlights how Chinese internet creators are approaching generative AI tools. Rather than viewing AI strictly as an auxiliary tool for background art or pre-visualization, creators are leveraging AI fangzhenren (photorealistic digital humans) to take on the roles of full cast ensembles and virtual set builders. This DIY ethos shows how independent directors worldwide might soon bypass traditional film distributors altogether, delivering feature films straight to online audiences.

Furthermore, it demonstrates the evolving relationship between literature and fan-made visual adaptations. Intellectual property that might otherwise sit in development hell can now be realized by dedicated solo creators willing to commit time and modest computing capital.

What to watch

Keep an eye on whether single-creator projects like this 94-minute Mountain adaptation spark a broader wave of indie AI feature films on global video platforms. As generative AI workflows become even more streamlined, the next question will be whether mainstream audiences accept full-length digital human performances, and how traditional film studios choose to respond to solo creators producing feature movies on micro-budgets.