My WeChat article “In the AI Era, One Person is an Entire Film Crew” has encountered some questioning voices these past few days. Therefore, although I’ve already written “Current Limits of AI Filmmaking,” I still want to discuss today what AI can and cannot do.
First, some people questioned that since I’m not proficient in everything—screenwriting, cinematography, music—and couldn’t possibly master all of these, I shouldn’t be so confident. First, I’m a New York University graduate, and NYU students typically have a kind of mysterious confidence, which I’m proud of. My confidence doesn’t need to be granted by others.
Secondly, while one person might be unable to master everything, AI can. Even though AI currently has flaws in music composition, its comprehensive capabilities in film are usually greater than those of ordinary individuals. AI is at least at a doctoral level in each field, if not legendary. Using AI allows you to achieve interdisciplinary integration and become a polymath.
This is what current AI can already do—make you a polymath. Before the AI era, becoming a polymath required the times to allow it, meaning it needed an era where knowledge wasn’t overly abundant. Otherwise, with modern disciplines so detailed and specialized, cross-professional content becomes incomprehensible, making it challenging to produce polymaths. A polymath is someone proficient in multiple disciplines.
However, with AI and substantial language model assistance, a person can quickly break through knowledge barriers and obtain structured knowledge input in specific disciplines. Since ChatGPT appeared, my reading and learning habits have changed because I can directly chat with it to learn the knowledge I need, and it can be customized.
So even if one person cannot master everything, their combat capability has been infinitely enhanced. I wrote “In the AI Era, One Person is an Entire Film Crew” to record my previous mental journey of producing AI films alone, not to show that I understand everything.
Another questioning voice says that AI filmmaking currently cannot achieve continuity and that I was exaggerating AI’s role in my writing. They also asked if I’ve used Veo3. I’ve only used the educational version of Veo3 and haven’t used Veo3 to make my films yet, mainly because I’m a fanatic about character and scene consistency. Veo3 doesn’t surpass Kling by much in image-to-video generation.
Can AI video achieve continuity? Actually, people proficient in editing already made it coherent last year. For someone like me who isn’t a master editor, I can only use color grading, character consistency, and other methods to give audiences a feeling that the story is continuing. Although AI cannot handle overly long shots, and the first and last frames with Midjourney are still lacking, I’m confident I can edit for atmospheric effects.
This raises the question of whether AI can currently produce long-form narratives. I can. Several AI short dramas that have already gone online have achieved considerable long-form continuity. So, we shouldn’t avoid breakthrough attempts just because something is difficult.
Next, let me mention several things AI currently cannot do. First, 100% character consistency cannot be achieved; if needed, it must be combined with 3D. Scene consistency must also be combined with 3D. Then, many camera movements cannot be completed. The solution is still mixed with 3D. We can see that AI and traditional 3D production will merge.
Can fight scenes be done now? Partially, the kind of multi-angle switching with intense emotional fine-tuned fight scenes I need is still impossible. Of course, I’ve always been weak at fight scenes, and there are many methods I don’t know that await my development.
I write on WeChat to let everyone see the Computational Universe novel and participate in my creative process. I’ve been thrilled about AI because it’s one of my true loves. I’m an AI filmmaker, so I understand AI’s limitations very well. Some of my adrenaline-pumping WeChat articles mainly talk about my experiences. The usual content is about my fears, confusion, and anxiety, but ultimately, how I continue on the path.
The production of the Computational Universe short film “The Resolve” has reached the final stage and will soon meet everyone. This is the first of my 25 short story ideas, and the novel has already been written—you can read it on my WeChat account.
Follow me and explore AI filmmaking together.
