AI technology is in a phase of rapid development, with most AI film technologies having primary update cycles of just 3 months or even shorter. This means that the technology mastered today will likely be obsolete within a few months. Films produced today may become unwatchable within a matter of months.
This is even faster than the visual quality updates in the gaming industry. I remember the art director of Guild Wars 2 once said something similar: “Graphics will become outdated with technological advances, but good artistic design will not become outdated.” However, Guild Wars 2 has still become visually outdated. Good games might last ten years, and some, like World of Warcraft, can persist for twenty years, but in later stages, their appeal to new players drops dramatically.
Newly released games often become unplayable after a few years, as their graphics can no longer be supported. Newly released AI-generated films become unwatchable after a few months, as their visuals can no longer be viewed. Art in the era of rapid technological development may not be eternal. Art in the technological plateau phase gradually loses its vitality.
The rapid pace and cyclical nature of technological updates is a problem every AI creator must face. Some people, therefore, argue that technology isn’t essential – that art and aesthetics are what truly matter. Art is also a form of technology to some extent, requiring extensive practice to develop proficiency and find the right feel. As for aesthetics, it’s shaped by a person’s lifetime experiences, requiring both money and time investment.
Moreover, AI technology is evolving in a trend of becoming simpler. Previously, writing prompts for image generation required special formatting; now, it’s increasingly moving toward natural language. Previously, image editing relied entirely on Photoshop; now, you can control multimodal models to edit images with a single sentence. Previously, video generation had very low success rates; now, the understanding of prompts is becoming increasingly sophisticated. It provides more opportunities for people to participate in AI film production.
Based on my hands-on experience, no matter how simple the technology becomes, there are still many details throughout the entire process. These details cannot be fully mastered without dedicating extensive time to researching and applying them. In other words, getting started is easy, but becoming an expert still requires extensive practice to familiarize oneself with each tool’s characteristics and overall control. Thinking you can skip technology in AI work is just wishful thinking.
The best approach is to have a team with dedicated people responsible for discovering new technologies. Since I’m making AI films alone, I have no choice but to watch Bilibili and check groups daily to stay up-to-date with the latest technologies. If I fall behind for a few days, I need to find opportunities to catch up; otherwise, I don’t feel secure… This might be an exaggerated fear, but I know it’s necessary.
Speaking of technology, we must mention ComfyUI, which many people in AI filmmaking can’t avoid. Some people think ComfyUI will become obsolete because it’s too complex. I disagree. ComfyUI won’t become obsolete precisely because of its node-based programming. Once something becomes programmable, it has exceptionally high development value. Node-based programming is used in the Unreal Engine series and hasn’t been eliminated either.
Moreover, using ComfyUI is currently not difficult – simply install an integrated package or click on a workflow at RunningHub for out-of-the-box functionality. The real problem is that the current version of Flux has a significantly larger aesthetic gap compared to Midjourney. If Flux 2 can be updated during Flux’s first anniversary in July this year, that would be another significant technological milestone.
After staying in the AI industry for a while, people develop a sense of technological update cycles. Preparing scripts, personnel, and funding during the end of a version cycle and producing heavily during the early stages of a version – this sounds perfect. However, in actual operation, you need to create content on a continuous basis. So, whoever catches the optimal technology cycle has some element of luck.
For example, the AI film technologies updated in May and June this year have already freed everyone’s AI films from the PowerPoint-like viewing experience. Companies or projects starting at this time have the advantage of being in the early stages of development. As far as I know, several friends caught this technological node. Therefore, several productions are expected to emerge in the second half of this year.
After saying all this, I actually think what we should be wary of is the “technology is useless” argument. Many people argue that technology isn’t essential, claiming that art and aesthetics are what truly matter. Although what most people think is usually correct, when too many people say this, we should instead consider whether it’s accurate. Some people believe technology is useless simply because they’re not the ones mastering the technology. Some people think technology is useful also because they are the ones mastering the technology.
I still maintain that viewpoint: AI filmmaking is about striking a balance between the excellence of technology, art, and literature. All three are indispensable here; lacking any one of them will result in a work that won’t be remembered. Compared to technology, more people will overlook literature.
It’s not humans using tools; it’s tools using humans. If you look at the history of human civilization, you can see the process by which humans have been changed by the tools they have developed. I’ve always believed that science is not just a tool but also an ideology – meaning it’s a culture as well.
Follow me and explore AI filmmaking together.
