After the Sora Hype Subsides, Reason Emerges: What 5 Top Creators Really Think

Hi, I'm Koji Yang, and Crossing is my personal newsletter. The name comes from something Steve Jobs said about standing at the intersection of technology and the humanities.

I'm the co-founder of sleep environment brand Tangdao and content company The Fair. I previously founded the Jiepang app and was part of the founding teams at Fanfou and Jumei International Holding Limited.

As a quintessential cross-disciplinary entrepreneur, I've let curiosity carry me across all kinds of intersections — from mobile internet to e-commerce, from content to consumer goods.

If you'd like to join the Crossing community chat, feel free to add me on WeChat: YuanchengYang.

HOLLYWOOD has become SORAWOOD

Two weeks after Sora's release, the noise has died down, temperatures have cooled, and reason has emerged.

The sweet spot for re-examining its impact, controversies, and value is right after the peak, while the embers are still warm.

At this moment, we can more objectively assess the real impact of the hype without being swayed by excessive promotion. We can also think deeply about the debates and discussions it sparked, understanding the perspectives and positions of all sides. And it helps us identify the value and opportunities the hype has created, providing reference for future decisions and actions.

Sora's official website

I've read the massive amount of online discussion about Sora, but most of it comes from AI media or tech geeks — they're watching from the sidelines, theorizing from a distance.

What's missing online are perspectives from people actually creating on the front lines, especially those carrying real weight — with representative works, with influence.

I have quite a few such friends in my circle. This weekend, I interviewed them intensively to get their honest thoughts on Sora.

❶

Zou Fei — Advertising Director, Founder of "Zhizuo"

*Zou Fei (personal website: https://zoufei.me/), one of China's top commercial advertising directors with countless iconic works — you've definitely seen his ads.

He's also a super geek who was training his own GAN models back in 2020.

And he's the founder of Zhizuo, a next-generation AI creative tool for the film and television industry.*

In April 2023, Zou Fei shot an ad for Taobao — possibly the first fully AI-produced short film.

I think Sora's emergence is a good thing. Technological democratization: everyone can create, everyone can have more time to create.

Giving time back to creation.

If we're talking downsides, I really can't think of any. I don't think there will be fatal impact for now. I'm not even sure if Sora will be available to the general public, or if it's just a demonstration of OpenAI's world model capabilities. Initially, it might only threaten stock footage sites, since most training data probably came from them.

Many people worry that AI will cause human unemployment. I think it will just give people more time to do what they love. The world's output stays the same, but we have more time.

But if a world model or AGI actually emerges, the impact on human society won't be something the advertising industry alone has to face.

For my daily work as a commercial director, Sora essentially helps me with storyboards — quickly realizing what I imagine a film to look like, quickly showing clients. If the client buys it and the footage meets their standards, you could basically deliver the final cut.

By my standards, based on the videos released so far, Sora hasn't reached high-quality film and television level yet. But once it does, there will be numerous safety concerns, and I believe OpenAI won't open it to the public. So for now, Sora can't help me increase my income.

Among Sora's current demos, the butterfly video moved me most.

Compared to its physical property perception, frame detail, etc., what amazed me most was maintaining near-perfect 3D continuity across such large camera movements, complex scheduling, and extended duration!

I'm very curious about how OpenAI specifically processed its training set. This is the embryonic form of a world model. If it can achieve real-time rendering on top of this, future virtual reality won't be virtual at all.

👦🏻 Koji:

Similar to Zou Fei's view, in my previous article on Sora "After Reading 100,000 Words of Sora Discussion, Here Are 6 Points I Want to Share", I mentioned that seeing Sora made me believe video creation capability will become something everyone can possess.

This significance is no less than the evolution from Oracle bone script to the printing press.

In the Oracle bone script era, only those who mastered the most expensive materials (oracle bones) and the most complex content technology (carving) could create content that could spread. What about everyone else? They had only their own voices, and even shouting themselves hoarse could only reach the neighbor across the mountain.

Then came the printing press, then the typewriter, then the internet and social media — and written creation and distribution were democratized. Today, right now, the article I'm typing here — if it's good enough — can reach thousands, millions, tens of millions of people overnight.

This is democratization brought by technology.

Sora's breakthrough for video creation is another such leap, another democratization. This world will evolve from everyone being able to write articles, to everyone being able to make videos.

I very much hope this era arrives quickly. Any idea or inspiration can more freely become a finished work. Something to look forward to.

---

Zou Fei's Zhizuo is a next-generation AI creative tool for the film and television industry, comprehensive in function and excellent in experience.

Honestly, I can hardly imagine how Zou Fei has been allocating his time and energy this past year — managing to continuously shoot top-tier commercial ads while also founding a non-trivial AI Native App. His daily information throughput and creative energy are a small universe unto themselves.

Zhizuo official website

Technological democratization, everyone can create — this aligns with Zhizuo's philosophy.

Zhizuo keeps pace with cutting-edge technology trends, continuously integrating the latest and best AI tools and linking them together into complete creative workflows for our practical application scenarios.

For Zhizuo, if Sora opens its API, we'll be the first to integrate it into "Zhizuo Storyboard," helping everyone generate shot-by-shot video from storyboard content, paired with audio generation models, exporting finished films with one click! Zhizuo already has these capabilities — everything is ready, we just need the east wind.

The more anxious the industry gets, the more interesting Zhizuo becomes. Good technology needs to land — that's what Zhizuo is doing, bringing the latest and best AI into our practical application scenarios. I hope to see the entire industry transition smoothly through this technological revolution, using new technology well to reduce costs and increase efficiency. When everyone's doing well, that's when things are really good.

Welcome to try Zhizuo: https://zizo.pro/

❷

Zhang Wei — Founder/CEO of The Fair

*Zhang Wei is my entrepreneurial partner. For eight years at The Fair, we've fought side by side, working to continuously create cultural phenomena.

The Fair serves top domestic and international brands, shooting over a hundred brand ads and brand story videos every year.

Zhang Wei is also the most talented creator I know. He often sits across from me, casually chatting, and writes several paragraphs that make me want to cry.*

The Sora demo that impressed me most was undoubtedly the dolphin riding a bicycle. When I chatted with director Chen Kun, he mentioned this same clip — he felt similarly. The rules of the physical world were extended so simply, and it was so real — it's exactly what a dolphin would look like if it could ride a bicycle. Scenes that previously required extensive model design and calculation to achieve are now realized effortlessly.

Human imagination has truly become its own boundary — or perhaps there are boundaries beyond human imagination.

1）

I'm involved in quite a few industries, but they can all be defined as "using the best tools to communicate with people."

Sora represents AI tools that are good tools — they'll lower technical barriers and give more people the ability to express themselves.

So I think the biggest impact will be the emergence of new creators, new content idols completely off people's radar, with excellent taste, ways of thinking, and worldviews, but who in the past lacked even basic video production capabilities due to technical limitations.

2）

The new tool throws into relief how terrifyingly inefficient this industry has been.

In the past, expressing uniqueness had several prerequisite constraints. Creating a world in video required over $100 million in investment. In the future, that likely won't be necessary.

Creative freedom will be amplified by orders of magnitude. High-quality expression will become the capability of more people.

3）

The "bad" I can think of is all short-term. For a period, the industry will experience standard failure — good standards built in the past, many of which were technology-based, will become standards everyone can reach. I think past evaluation criteria, including pricing standards, will suddenly be erased. Much creative labor will become worthless.

The industry may be flooded with large amounts of similar, indistinguishable content. The sudden emergence of content that's hard to rank will kill many people's creative motivation.

I think after tools like Sora become universalized, this state will last about a year.

But eventually, people will realize that what's truly valuable is story and aesthetics, and uniqueness will still stand out.

4）

I've always had an outline for a world-building fantasy novel in my head, with humans and demons, involving the expression of an entire world and non-existent species. In the past, this would have required a very large professional film team working for several years to produce.

I'm very much looking forward to trying to create it alone after Sora emerges.

Some may know that seven years ago, we worked with Baidu to design an online virtual space called "The City of Eternal Rain" using 3D, open for people to explore.

In 2017, we made "Rain City" based on H5, explorable in 360°. Back then, searching "Rain City" in Baidu Maps, you could actually find and enter this virtual city.

At the time, we asked The Fair users to contribute ideas about what this eternally rainy city would look like, what physical rules it would follow, what living spaces it would have. 4,345 rain enthusiasts participated, and we received over 10,000 urban construction proposals — all excellent.

But the implementation process was terrifying. It took us nearly a year, multiple design and modeling teams, and over 1 million RMB to produce a product we considered very unsatisfactory. Few ideas were realized, and those that were came out crude.

I think now those users' ideas could be simply realized, and would no longer be let down. Combined with Apple Vision Pro, the effect would certainly be excellent. This is what I see as AI liberating people's creative capabilities.

❸

Qian Ninghuang — Sci-Fi Director

Qian Ninghuang, a new-generation sci-fi director.

Representative work: The Egg Man. On the variety show Directors, Please Advise, he made a comeback all the way, and his Nebula Award short film Everyone Loves Charles received great acclaim.

Qian Ninghuang's work montage from Directors, Please Advise

With the emergence of Sora and generative AI, art and VFX R&D costs will drop dramatically. This is a very important change.

Now some art departments can already use AI to preview various possibilities in advance, investing at scale only when the direction is relatively precise, thereby saving massive amounts of time and resources.

For example, designing a lunar base — we can use MidJourney to brainstorm and realize the base's internal structure, seeing various styles. Two computers running MidJourney 24 hours a day, generating hundreds of images daily. Though most may not be suitable, the art director can definitely pick out one or two that are relatively appropriate and very complete. For this volume and completion level, even a very good draftsman might need 3-4 days per image.

I used to think visual references could already be largely solved by MidJourney. Now if Sora can solve video too, then in the future it's likely not just art design — animation, film, VFX and other visual fields' R&D costs will all drop significantly.

Current Sora still has some imprecisions and small bugs that don't follow physical rules, but I believe these will be solved in the future.

Although AI technology can help us improve efficiency and reduce costs, it may also cause some people to lose their jobs.

From what I know, many game companies and film/TV projects' art departments have already reduced budgets significantly through AI use.

👦🏻 Koji:

Whenever we talk about AI, the word "unemployment" inevitably pops up. This is indeed real anxiety that concrete individuals are facing. Even Qian Ninghuang worries that many of his friends may face unemployment risk.

What do I think? Forget about unemployment — just make yourself the person in your circle who understands AI best!

Embrace change. This isn't motivational fluff, it's an action guide.

AI may eliminate some people, but only those who don't know how to use it. Just as white-collar workers today need to know how to use computers, or they'll be eliminated. Photographers too — they need to know Photoshop, or they'll be eliminated.

❹

Liang Zi — Photographer

*Liang Zi (personal website http://liangzidop.com/), one of China's most expensive commercial advertising photographers (because the work is just that good) — I'd love to drop the "one of."

With countless iconic works that have gone viral again and again, he possesses outstanding art direction and superb visual narrative ability.

Among Liang Zi's works, my favorites are his BV ad quartet: Love, On the Road, All the Way to You, All the Way Home, and The First Light of the New Year.*

I've watched all of Sora's demos. Very impressive. Not open yet, but we'll use it eventually, and it'll be amazing.

We're already using MidJourney. I think Sora will be even more amazing, capable of much more.

Under the generative AI wave, art departments will probably be the first to lose their jobs.

The dragon dance demo left a deep impression. Just those few seconds — previously we'd have spent hundreds of thousands to shoot: location, personnel, set dressing, scheduling, post-production, all costs.

In the future, I think similar large-scale shots serving to elevate atmosphere and mood, plus the aerial landscape shots from the demos, are what we'll consider using Sora for.

I was equally deeply impressed by the astronaut on an alien planet demo. Sora actually produced a handheld camera effect — amazing. Handheld shooting enhances realism, makes audiences feel more present. I didn't expect Sora could achieve this level of handheld effect. Very impressive.

I'd consider first using Sora to generate large-scale shots, then doing art direction, set dressing, and lighting based on those shots, so that human-shot close-ups and details can match the Sora-generated footage, making the final cut fluid and complete.

I shoot a lot of car ads, often needing various wide shots and big scenery. If Sora can generate accurate models by car model in the future, that'll be amazing. For example, an aerial shot of a certain Audi model speeding across Chongqing's Yangtze River Bridge as night falls — doing this with Sora would be incredibly efficient. I don't think that day is far.

Directors and actors won't lose their jobs.

I think Sora and AI probably won't be suitable for shooting narrative films for a long time, especially those requiring human actors. Character consistency issues must be solved first. Beyond that, to shoot good narrative films with human actors, first you need to rely on actors' movements and expressions, second on directors' precise design and scheduling — countless details that let audiences enter the story, empathize with actors, be moved. AI can't micromanage every detail, so it's hard to truly replace directors and actors.

The impact on my income isn't visible yet. But I believe people who can use AI well will definitely produce better work and earn more income.

What I'm curious about is: with further computing power improvements, can Sora generate videos over 60 seconds that remain logically coherent throughout? If so, I can't think of a word more impressive than "amazing."

👦🏻 Koji:

Liang Zi came to my place last weekend, and friends watched the film 12.12: The Day together. The preceding content was my interview with Liang Zi before the movie started. Then, watching the film, I couldn't help thinking: "This shot could use Sora, that shot Sora still can't handle."

With this perspective, I felt like I had become a "video creator" too — an unprecedented feeling, a feeling of being empowered, a feeling like my toolbox had suddenly gained a superpower.

But honestly, over the two hours of 12.12: The Day, I'd estimate no more than 1/20 of the shots could be generated with Sora. Many other shots still require precise actor performances, complex on-site multi-person/multi-camera coordination.

❺

Fang Jianping — Co-founder of UDL Design Studio

Fang Jianping is co-founder of what I consider one of China's most impressive design studios, UDL (https://u-d-l.com/).

Unfortunately, Tangdao's visual identity wasn't done by UDL — they're a bit too expensive.

But I've always hoped Tangdao can quietly earn and save more, looking forward to the day we can afford Fang Jianping and UDL for a design upgrade.

UDL work showcase

Our late-night chat:

👦🏻 Koji I'm quite curious — if you weren't who you are today, but say a sophomore or junior in college, facing this new industrial revolution, what choices would you make?

👦 Fang Jianping Good question. At our age, there weren't many choices in college — that was actually easier. Sometimes I put myself in these post-95s, post-00s' shoes. It's tough, so many opportunities, so many possibilities, how to choose.

👦🏻 Koji Maybe we're also older and more afraid of choosing wrong? Hahahaha

👦 Fang Jianping That's part of it, definitely. At least subconsciously. You've already gained certain skills and achievements in some areas. When something new appears, you'll definitely tense up — what's this? Is it a threat to me? Can it help me? Will I be eliminated?

👦🏻 Koji If I went back to sophomore or junior year, I'd probably want to go to the United States more, to the eye of the storm pushing the wave forward. I've been a bit regretful lately — I got a CMU offer with scholarship back then, but gave it up to stay and start a business in China.

👦 Fang Jianping But that's the thing — we don't have the ability to return to the past. All judgments are based on the present.

👦 Fang Jianping When MidJourney came out, I considered a question. Of course this was about the technology. My biggest feeling was: damn, this prompt command logic — isn't it exactly the same as a client writing a brief? If the brief isn't clear, you don't get the answer you want.

👦🏻 Koji Hahahaha, so true. Actually being able to write a clear brief is already hard.

👦 Fang Jianping At the time I thought, clients who can write clear prompt commands must definitely give clear briefs. So looking at Sora today, I feel the same. It's a tool where you need to accurately describe what you're thinking to get maybe 50% accuracy, and you have to keep adding more precise descriptions. This process is very similar to our collaboration with clients.

👦🏻 Koji When you saw Sora, did you think of any projects where you could immediately use it?

👦 Fang Jianping Our service process is: design strategy — concept design — detailed design — final artwork. I think MidJourney and Sora might help at the first stage (design strategy).

Let me explain — the design strategy stage is actually when we align the brief with the client.

Take Tangdao as an example. Suppose Tangdao's positioning is: modern, relaxed, comfortable. You give me the brief wanting a logo and visual system. My first step is aligning with you: what is Tangdao's modern, relaxed, and comfortable? This requires a series of images and video materials to help calibrate these keywords. At this point, MidJourney and Sora might help — they can quickly get me some answers.

Like in Sora's case of the cat waking its owner — if that can help me explain modern, relaxed, and comfortable life for contemporary young people, I think that's great.

👦🏻 Koji Got it. Actually similar to what the two director friends said earlier.

👦 Fang Jianping The opposite of tools is people, things that can empathize. If you use AI to shoot Dune, I think it works. If it's Eat Drink Man Woman or Brokeback Mountain, I don't think so.

Actually my secret hope is that Ang Lee's next film uses AI. I think there will definitely be such a work.

👦🏻 Koji Interesting. What would Ang Lee shoot with AI? He can't shoot Dune either...

👦 Fang Jianping Don't know. I think Ang Lee's observation is so delicate — he has that superpower of telling your sexual orientation from one look.

👦🏻 Koji Well if I ever get to know Ang Lee, I'll have him assess me...

👦🏻 Koji Back to the point — can Sora help you make money?

👦 Fang Jianping Not for me. Our work is highly customized. Sora-type tools will have bigger impact helping companies like Taobao, Douyin, Pinduoduo — their users are mostly universal, without much professional experience. For them, I think efficiency will increase massively.

👦🏻 Koji But though your work is highly customized by client request, it's ultimately presented to ordinary users to see. So what's the difference in between?

👦 Fang Jianping Maybe uniqueness? No one wants to spend heavily on a piece of clothing only to find someone else wearing it. You're also a brand owner — how do you see this?

👦🏻 Koji First, from the end backwards, originality of the work is fundamental. Then you need to interpret the brief clearly and comprehensively, and match the brand tone. Whether it's AI-made or not, I don't really care — slowly people won't be able to tell if something is AI-made or human-made anyway, right?

👦 Fang Jianping Then if AI can solve everything, your reason for finding a design company is?

👦🏻 Koji AI can't do fine-tuning. And I think design companies provide design intelligence. Often my own brain is foggy, I don't know what I want, and need someone to help think it through, or "guide" thinking it through.

👦 Fang Jianping I think you gave the answer! For us: designing a McDonald's logo isn't hard. What's hard is how I make the choice and be confident it can accompany [the brand] for the long term. Our service process is giving 3 options: McDonald's M, Starbucks' mermaid, and Samsung's SUMSANG.

Then asking: "Boss, what do you actually want?"

👦🏻 Koji Hahahaha, the boss might say: ALL IN! Please help me develop all three directions. Deadly.

👦 Fang Jianping Right!? People are way more complex than tools — greedy, selfish, confident, and the arrogance that comes from overconfidence. Sora can't understand that.

👦🏻 Koji And people get sleepy... I'm going to sleep...

👦 Fang Jianping Yeah, good night. Really enjoyed chatting like this.

---

Afterword

India has a genius advertising director, Tarsem Singh. His work is super creative, and his rates for shooting ads are also scary. Once, a client asked him — you're charging us millions of dollars for a 30-second ad, how is that reasonable?

Tarsem Singh said:

The price you paid isn't just for the time I spend working for you. You're buying the crystallized essence of my entire life's experience: every sip of wine I've drunk, every cup of coffee I've tasted, every book I've read, every place I've been, every love I've had.

So, is that price expensive?