---
title: "The Pain Only Creators Understand: A Real Guide to Filling the Gaps from AIGC Video Concept to Delivery"
author: "Unique Research"
sourcePublication: "Unique Research Substack"
originalPublishedAt: "2025-12-17T08:01:13+00:00"
canonical: "https://ffcap.cn/en/research/src-20251217-01html"
source: "https://uniqueresearch.substack.com/p/src-20251217-01html"
language: "en"
---

# The Pain Only Creators Understand: A Real Guide to Filling the Gaps from AIGC Video Concept to Delivery

_Original · Unique Research · 2025-12-17_

_Historical edition: This complete rendition preserves the December 2025 article and its roundtable transcript. Rankings, company descriptions, model-release claims—including the statement about 30 Kling iterations in less than six months—and predictions are the source’s or speakers’ statements at that time, not independently verified current facts. AIGC means AI-generated content. Film titles such as Space Opera House, Monet’s Garden, Daisy, and The Fox-Tomb Bride are descriptive English translations; studio names are rendered from the source, not verified official English names. “Lightstorm” remains the source’s phonetic transcription, not a confirmed company identification. The source predominantly writes Long Yitong’s given name as 以童, while one spoken reference writes 以彤; both are rendered Yitong here._

From What AI Can Do to How AI Creates Viral Hits: What Happened in Video Creation in 2025?

At a super-individual workshop during the 2025 Unique Awards Beijing, moderator Zhang Yuanyuan, partner at Yilun AI, led an in-depth discussion with Long Yitong, still studying at the Central Academy of Fine Arts, whose team supplied materials for The Fox-Tomb Bride, now number one on multiple platforms; Wen Qin, founder of Wenweisi Studio, who left his role as an AIGC product manager at Meitu, joined Kling's director program, and made Daisy; Sen Hai, who left his position as a visual director at CCTV to establish Senhai Yingguang Studio; director Hebi, who made online films, writes science fiction, and is now preparing an AI feature; and Zhou Xiangu, who previously worked at a leading Hollywood production company.

With Kling iterating 30 times in half a year, were they anxious or excited? When democratized technology puts a pen in everyone's hand, does a professional director still have a moat? From the chaos of team collaboration to the deep pit of long-form narrative, how did they use unconventional methods to solve real problems?

I. Different Starting Lines, Similar Moments of Revelation

Place their entry points on a timeline and an interesting pattern appears: some took root as soon as Midjourney V5 launched, others experimented intensely with Stable Diffusion nodes, and still others did not truly pivot until last year.

Long Yitong's story is a classic case of legs moving faster than the brain.

While still at university, she became one of Jimeng's first Super Creators and made the sample Monet's Garden for the platform. Later, her studio served as the original-art team for Shanhai Qijing: Breaking the Waves, directed by Chen Kun. Her latest AI short drama, The Fox-Tomb Bride, gained traction across Hongguo, Douyin, and Fanqie Novel and fought its way to number one on two rankings.

For her, AIGC was never a question of whether to enter, but of having entered early and now deciding how fast to run.

Wen Qin began with one image—Space Opera House.

As an AIGC product manager at Meitu in 2022, seeing that image quietly shifted his career path: perhaps product managers could face not only event tracking, retention, and DAU, but also cinematic language and film. He became a Kling AI Super Creator, participated in the Daisy director program, and now independently runs a studio creating AIGC shorts for brands.

Sen Hai's turning point came from moving between an institutional role and entrepreneurship.

After years as a visual director at CCTV, he knew the rhythm of traditional processes well: one major project per year, one film every six months.

When Midjourney appeared, he realized that the foundation of the visual industry was being rewritten. He left the institution and took his team into more experimental and commercial AIGC video, hoping to establish AI production standards that genuinely work.

Hebi and Zhou Xiangu represent two kinds of creators pushed forward by their era: one is a director whose science-fiction dreams were repeatedly rejected by budgets. In traditional film, science fiction starts at 100 million (currency unspecified in the source); in the AIGC era, he suddenly saw that a small team might genuinely complete its own science-fiction feature.

The other is a producer who quotes B for Busy to joke, "My career is on a downward slope, and it is going very smoothly."

When she found that many projects could not advance after years in the traditional film environment, AIGC became an exit—not a retreat, but a side door.

She positioned her studio directly for overseas markets, using AI short dramas to compete with Netflix and global platforms instead of gambling on another lengthy domestic-drama cycle.

These completely different backgrounds ultimately point to the same judgment: AI is not a list of tools, but a new creative path.

Some were pulled in by curiosity; others were pushed out by the old world.

II. Before a Viral Hit, Fix the Most Painful Link

Everyone acknowledged that viral hits never come as gifts from models; they result from people and AI filling every pit together.

1\. Team Collaboration: AI Liberates Technology but Amplifies Aesthetic Disagreement

Long Yitong's greatest obstacle was not technology, but the team. AI lowers the entry barrier dramatically, yet stable long-form content requires consistent aesthetics, unified cinematic language, and continuous emotional control—all highly personal and experience-based.

When several people relay work on a project, even small differences in their understanding of beauty make the finished film look assembled by different people as the timeline lengthens.

They encountered a typical failure: chaotic version management caused the director to use old materials in the final cut, throwing the tone of the entire film out of alignment.

The lesson forced her to redefine the talent needed in the AI era: foundations in art, cinematic language, and AI understanding are all indispensable.

2\. Audiovisual Language: No Matter How Strong the Model, It Cannot Replace a Camera That Knows How to Speak

Wen Qin's pain point is that the real threshold has never been knowing how to use models, but knowing how to speak the language of the camera.

Give a person who cannot tell stories ten additional models, and you merely obtain ten more mediocre films.

His solution is simple: repeatedly watch large numbers of excellent traditional short films and advertisements, break down their narrative structures and shot design, and then combine those lessons with AI tools.

He uses large language models to expand ideas, but does not treat a prompt as an omnipotent screenplay machine. He treats it as a collaborative partner that can be trained.

When you know what you want, a prompt is only a tool. When you do not know what you want, a prompt leads you somewhere even more ambiguous than your own thinking.

3\. Long-Form Narrative: Alchemy Is Not Mysticism but an Empirical Science

Hebi said people enjoy spectacular samples on social platforms: a few dozen seconds of visuals accompanied by comments unanimously declaring that AI is incredible.

But when they must produce a complete long-form work, many teams become completely stuck.

His interpretation is that AI gives everyone a pen with ideas of its own. It cannot create shots you consider simple, yet occasionally surprises you with images you thought impossible.

The real solution is only four Chinese characters: practice at scale.

His team rarely discusses theory. Instead, newcomers generate 100,000 images and discover each model's boundaries through failure.

Eventually this is no longer artistic intuition, but empirical science—like a photographer learning every temperament of a lens.

4\. Cognitive Gaps: Creators Understand the Boundaries, but Clients and Audiences Do Not

Zhou Xiangu's obstacle comes from her background as a producer. She knows the limits of current models and actively avoids effects technology cannot deliver. Clients and audiences do not know those limits.

They request things beyond current technical reality, or deliver a final verdict on six months of work with a comment saying, "This AI is terrible."

She says this information asymmetry is the largest hidden cost in AIGC creation today.

Creators must understand technology and narrative while also learning to communicate with the market: what can be done now, and what belongs to the next generation of models?

III. Technological Anxiety: What Remains After Skills Are Reset to Zero?

Kling iterated 30 times within six months of release. New models arrive one wave after another, from ComfyUI to cloud nodes, from Stable Diffusion to Nano Banana, and from inpainting to infinite canvases.

Creators experience the same emotion every day: the thing they learned yesterday appears unnecessary today.

Long Yitong said she was deeply anxious when she entered in 2023. Configuring local environments, graphics cards, and strange errors made every step feel like walking on broken glass. Later, she realized that instead of fearing it, she should accept one fact: technology will inevitably become easier to use.

As tools become more foolproof, only two genuine barriers remain: aesthetics and expression.

From an engineering perspective, Wen Qin offers an anxiety-reducing principle: providers' underlying model structures and tagging systems may differ, but the fundamental logic is broadly similar.

Since the second half of this year, both infinite canvases and workflows have fundamentally pursued the same goal—solving efficiency problems so people have more mental capacity for content.

Hebi's attitude is the most direct: I actually think it updates too slowly.

In his view, updates invalidate node techniques he studied for months and allow beginners to start easily, which naturally disappoints technical experts.

But if the goal is a genuine science-fiction feature, every technological iteration that shortens cycles and lowers costs is beneficial.

Until AGI arrives, AI remains an amplifier of human taste, understanding, and courage.

Zhou Xiangu simply transformed the anxiety of being unable to learn everything into relief:

She intended to master ComfyUI and hired an expert to teach her, but before she understood it, Nano Banana launched.

At that moment she realized that instead of chasing engineers' skills, she should reconfirm her own position. Do you want to be an engineer, an artist, or a commercial creator?

Once you know who you are, technological updates stop being examinations and become only a few more buttons in the toolbar.

IV. AI Is Not Here to Replace You, but to Force You to Become a Complete Person

Will AI replace video creators? This tired question produced a consensus: no, but it will ruthlessly eliminate people who know only how to press buttons.

Sen Hai sees AI as an extension of capability boundaries, not a replacement.

What audiences truly want is never a model's capability list, but a creator's view, an emotion, and a cross-section of an era.

AI compresses production cycles, allowing you to make videos as you write and shortening, for the first time, the closed loop between wanting to say something and showing it to others.

Long Yitong is more specific: her imagined future AI film-production line divides work between a director and a professional production team. The director controls ideas, aesthetics, and overall tone; the team runs processes for scene generation, asset management, and technical implementation.

She is already bringing traditional film scheduling, asset management, and version control into AI workflows—not to make "AI works," but to make long-running IP financially possible.

Wen Qin has brought his product-management experience completely into AIGC creation. In his view, client phrases such as "creative," "designed," and "cooler" create only waste unless translated into concrete requirements.

He uses Gantt charts and Feishu spreadsheets for project management, breaking shooting, generation, retouching, compositing, and delivery into fine-grained nodes so the team and client can see exactly where a film is stuck.

Current tools are not intelligent enough: AI provides computing power, while people provide alignment.

These creators are all doing the same thing: pulling AI video creation away from a pile of dazzling samples and back into a real, manageable, deliverable industry.

V. New Battlegrounds Created by AIGC: Animation, Short Dramas, and Global Expansion

When the discussion turned to future opportunities, it suddenly became highly specific.

Hebi believes animation may be among the industries advanced fastest by AI. Technically, animation better fits AI generation logic; commercially, it inherently offers IP and room for ongoing monetization. That is why he remains committed to an AI science-fiction feature: in a traditional industrial process, every frame carries enormous cost; in the AI era, more budget can go into worldbuilding, character design, and emotional design.

Zhou Xiangu focuses directly on global expansion.

In her experience, a huge gap once separated domestic film from Hollywood. No one dared invest heavily in a project that looked like a US drama but remained only an experiment.

Now AIGC shows her another path: produce content with high audiovisual-language standards at extremely low cost, benchmark Netflix and international platforms, and use short-drama pacing to find a position in specialized niches.

AI did not immediately open the global market, but it gave Chinese creators a voice in global storytelling.

VI. Five Things Most Worth Doubling Down On in This Era

At the roundtable's end, Zhang Yuanyuan raised an excellent and difficult question: if you could give one sentence to creators entering or already working in the field, what is most worth investing twice as much in today?

They offered five answers that fit together like a clear assignment sheet:

Long Yitong said action. "My legs move faster than my brain." In her view, this era's greatest risk is not experimentation, but seeing a great deal and doing nothing.

Wen Qin said curiosity. Curiosity drove him from product manager to director and studio leader, kept him trying again through countless failed draws, and made his first response to every new model not anxiety but a desire to understand exactly what it can do.

Sen Hai said awakening. Once technology is set aside, you must answer two questions: who do you want to become, and what do you truly want to express? If those questions remain unclear, stronger AI only amplifies your confusion.

Hebi said passion. Only when creation itself brings joy can a work connect with others. When moats are washed flat, only this remains: are you willing to stay awake one more night for an image or an idea?

Zhou Xiangu added one final word: attitude. AI knows more than people, but people have attitudes. Good content is not merely beautiful; it carries a creator's position toward the world. If every technical barrier disappears one day, that may be the only thing that still distinguishes works.

Final Note: A Viral Hit Is a Byproduct, Not a KPI

If this conversation about AIGC video creation were compressed into one sentence, it would be:

A viral hit is not a model's gift, but the byproduct of a person repeatedly questioning themselves amid the waves of an era.

Technology will continue updating, new platforms will continue appearing, and tutorials for viral hits will be cut into short video after short video and published.

But how far you travel will probably still be determined by several things that look very un-AI: whether you dare to begin before overthinking, whether your curiosity persists, whether you maintain clear self-knowledge in complexity, whether passion is why you remain, and whether you are willing to take responsibility for yourself and your work.

AI has picked up the camera, but the moment the record button is pressed remains a human choice.

The credit you place at the end is the real copyright declaration of this era.

More Details from the Conversation

01\. Opening and Guest Introductions

Zhang Yuanyuan: Hello, everyone. Thank you very much, including the many friends standing at the back, for joining today's special roundtable on AIGC video creation.

Last year or the year before, we were still discussing what AIGC could do. Today, however, we have already seen many viral products created with AIGC.

We are honored to welcome these pioneers in AIGC video creation to discuss genuine lessons from making viral hits. First, please briefly introduce yourselves and explain what opportunity or moment of realization made you go all-in on this field.

Long Yitong: Hello, everyone. I differ from the other guests because I am still an undergraduate at the Central Academy of Fine Arts. I began experimenting with AIGC relatively early. When Midjourney V5 had just launched two years ago, I was already rooted in the field.

I was also among China's earliest AIGC video creators and one of Jimeng's first nine Super Creators, producing the sample Monet's Garden for Jimeng. Later, our studio served as the original-art team for Shanhai Qijing: Breaking the Waves, led by director Chen Kun. Today we focus more on long-form, high-quality exploration.

A short drama for which we recently helped produce materials, The Fox-Tomb Bride, launched on Hongguo, Douyin, and Fanqie Novel and now ranks number one on both Hongguo and Douyin. Please watch it if you have an opportunity.

Wen Qin: Hello, everyone. I am Wen Qin, known online as Wenweisi. I previously served as an AIGC product manager at Meitu.

I began working with AI in 2022, when Midjourney was still at a very early version. The first image that stunned me was Space Opera House. As a product manager, I found it extraordinary and began studying the field from then on.

I am also a Kling Super Creator and participated last year in Kling's director program, collaborating with director Wang Zichuan on the film Daisy. I have now founded a studio that primarily produces AIGC creative shorts, advertisements, and films.

Sen Hai: Hello, everyone. I am Sen Hai. I previously worked as a visual director at China Media Group, CCTV, and have now founded Senhai Yingguang Studio.

I also began researching AI the year before last, when Midjourney V5 appeared, and started experimenting with video late last year, entering competitions and winning awards. This year's goal is to gradually formalize and commercialize AI production.

Hebi: Hello, everyone. I am Hebi. I previously worked in traditional film and television, shooting online feature films and series.

I encountered AI relatively early and began exploring image and video generation through Stable Diffusion in 2023.

I entered because I love science fiction, write science-fiction novels, and especially wanted to make my own science-fiction film. In traditional production, however, science fiction is so expensive that it cannot be done without 100 million (currency unspecified in the source). AI showed me a possibility. I am also a Kling Super Creator, and I felt hope that I could create ambitious science fiction, so I entered the field. I now primarily make vertical short dramas and high-quality AI shorts while preparing my own AI film and animation.

Zhou Xiangu: Hello, everyone. I am Zhou Xiangu and also come from traditional film. I previously worked as a producer at a US company transcribed phonetically in the source as “Lightstorm”, then returned to China to make web dramas and write screenplays and novels.

I encountered AIGC very late, around this time last year. To explain why I turned toward AIGC, I will borrow a line from director Shao Yihui's B for Busy: "My career is on a downhill slope, and it is going very smoothly."

I entered AIGC as the traditional film industry cooled. When many projects remained unable to advance after years, AIGC became an excellent outlet and gave me creative freedom. As models mature and commercialization milestones arrive, my studio, URSA Studio, now primarily creates AIGC short dramas for the United States and other overseas markets.

02\. Obstacles and Solutions in Creation

Zhang Yuanyuan: The path to a viral hit is certainly difficult. What has been your greatest obstacle in AI video creation? Do you wait quietly for models to evolve, or are there other solutions?

Long Yitong: Our obstacle is team collaboration. AI technology has liberated the technical layer and returned creation to fundamentally human aesthetics, cinematic language, and ideas. But these are highly personal and experience-based.

When stable long-form content must be created, unifying several people's experiences is extremely difficult. If their understandings of beauty differ, then as the timeline lengthens, the film stops looking as though one person made it.

Solution: we eventually understood what talent AI creation requires. When recruiting, we seek partners with matching professional backgrounds—foundations in art and cinematic language, plus an understanding of AI technology. When everyone shares a consensus, they can agree on what is good and complete high-quality work.

Wen Qin: I strongly agree with Yitong. Audiovisual language is extremely important. Partners without relevant experience struggle to imagine how to design shots, what constitutes a good idea, and how to be distinctive.

Solution: our method is to watch more outstanding traditional shorts and advertisements. Although today's AI tools such as GPT can help us expand early ideas, I still combine them with the logic of excellent traditional short films. Creating images with AI is no longer difficult; ideas remain the hard part.

Sen Hai: The obstacle is deciding what I want to make. Technology changes every day and future barriers will be extremely low. At that point, ideas and content alone will no longer be enough.

Today's creative process is not linear like a traditional one; it resembles solving a Rubik's Cube. Previously, you thought first and made later. Now you think while making, interacting with AI, and the outline of the work becomes clearer through repeated refinement.

Hebi: The largest obstacle is long-form narrative. Everyone finds AI samples spectacular and thinks dialogue alone can generate them, but that is not true.

AI democratizes technology and gives everyone paper and pen, but the pen has ideas of its own. It cannot create a shot you think is simple, yet can create one you think is difficult.

Solution: it is an interactive process. You must understand AI's properties to dance well with it. The core is practice at scale, or model "alchemy."

Theory is useless for a new teammate; ask them to generate 100,000 images and they will naturally understand model boundaries. This is an empirical science.

Zhou Xiangu: Coming from production, I habitually build the framework before planning. I also encounter obstacles, but I focus more on cognitive gaps.

I know where model boundaries lie and avoid them while creating; clients and the market do not. They demand things current technology cannot achieve.

Viewers in comments say, "Bah, AI is garbage," without knowing that this is already the highest current expression of accumulated human intelligence. This information asymmetry and difference in understanding AI's capability boundaries is my greatest obstacle.

03\. Anxiety Amid Rapid Technological Iteration

Zhang Yuanyuan: Kling iterated 30 times within six months of launch. Is such rapid iteration honey or a trap for you? Does it cause anxiety?

Long Yitong: I was anxious when I began in 2023. Using Stable Diffusion and ComfyUI required configuring local environments and reading code, which was extremely painful.

Later I realized that technology inevitably becomes easier to use. Large companies such as Kling will continue improving tools. Our core attention should return to literacy, aesthetics, and expression. Only when liberated from the tool layer and returned to emotion and expression can AIGC works genuinely enter people's hearts.

Wen Qin: Anxiety certainly exists. But I come from software engineering and know code, so I learn slightly faster.

Since the second half of this year, every platform has been building infinite canvases and workflows, fundamentally to improve efficiency.

The underlying logic does not change. Although providers use different data labels, they are broadly similar. Technology can now generate content from plain language. What matters remains how you think, write a prompt, and understand the logic behind the model.

Sen Hai: Models iterate too quickly. Films once took years or months; now they take weeks or days. Often a new tool appears before the images finish generating, creating a dilemma over whether to discard everything and restart.

Response: at this stage I prefer short, fast-turnaround content lasting only a few minutes, suited to rapid completion by an individual or small team and reducing interpersonal communication costs.

Hebi: Anxiety comes from skills being reset to zero. You may study ComfyUI nodes for months, only for an online-model update to make the skill useless.

Now, however, I am not anxious; I think updates are too slow! Current models remain far from genuine long-form science-fiction narrative and contain many deep pitfalls.

I want them to update immediately, although that means beginners can start easily and my technical barrier disappears. Core competitiveness will always be ideas, user understanding, and an interesting soul. Until AGI is achieved, AI is only a tool.

Zhou Xiangu: The previous guests are top students; I am the struggling student. I once vowed to learn ComfyUI and hired an expert, but Nano Banana appeared before I had learned it.

Model iteration liberated me instead. I do not need to study that complex engineering. The key is how we position ourselves: engineer, artist, or commercial creator? Different positions produce different experiences.

04\. The Human-AI Relationship: Replacement or Partnership?

Zhang Yuanyuan: Will AI replace video creators?

Sen Hai: It is a partner and an extension of our capability boundaries; there is no replacement.

Creation is communication. Audiences want to understand the author's views and soul. If AI replaces expression, they lose interest. AI compresses the creative cycle, lets us make video as we write, and closes the expression loop.

Long Yitong: A future AI film-production line should divide work between a director who controls ideas and aesthetics and a professional production team handling batch production.

Our team needs foundations in aesthetics, cinematic language, and technology while introducing traditional asset management and project scheduling. We once suffered when chaotic version management caused a director to use old materials and damaged the broadcast result.

Wen Qin: After transitioning from product management, I found product thinking highly useful.

A product manager connects levels and translates an owner or client's abstract needs into concrete requirements, avoiding ineffective communication. AI creation is the same: when a client says "make it creative," I guide them to specify what they want.

Using Gantt charts and Feishu spreadsheets for project management, PM, is also critical. Current tools are not intelligent enough, so people must manage progress precisely to guarantee delivery.

05\. New Opportunities and Futures Created by AIGC

Hebi: Animation may be the next industry rapidly transformed by AI.

This is the best and worst era for creators: best because tools are democratized, worst because moats have shattered. Ten years of experience can become obsolete instantly, forcing you to compete naked in the market.

Zhou Xiangu: AIGC gives us an opportunity to create content for global markets.

Domestic film once remained disconnected from Hollywood, and no one dared invest heavily in work with the texture of a US drama. With AIGC, we can now create content with sophisticated audiovisual language at extremely low cost and compete with Netflix and international markets.

Markets are segmented, and AI helps us enter every field.

06\. Summary: Which Core Asset Should We Invest Twice as Much in Today?

Long Yitong: action. My legs move faster than my brain. In this flourishing era, rushing forward first and learning through action comes before everything else.

Wen Qin: curiosity. It drove my transition from product manager and sustained me through countless failed draws until I obtained magical positive feedback.

Sen Hai: awakening. Set technology aside and decide who you want to become and what you want to express.

Hebi: passion. Only finding joy in creation produces connection; that is the sole winning weapon.

Zhou Xiangu: attitude. AI knows more than people, but people have attitudes. Good content is not only beautiful; it has an attitude.

---

Original publication: https://uniqueresearch.substack.com/p/src-20251217-01html
On-site reading page: https://ffcap.cn/en/research/src-20251217-01html