Runway AIF 2026 Seoul Screening: AI Video Starts Becoming Real Cinema

All ten films shown at this screening were short pieces under 10 minutes. Being short was in some ways a weakness, but it also made clear exactly what each film chose to include and what it left out. According to the official AIF 2026 page, the Film category finalists are as follows. https://aif.runwayml.com/

• A Face Only A Mother Could Love — Robert Gaudette, 7 min 50 sec
• THE WELL — Dorian & Daniel, 5 min 29 sec
• Where Knights Fall — Mathery
• Costa Verde — Léo Cannone & New Forest Film
• TAIRELL ISN'T REAL — Dave Clark, 5 min 4 sec
• Little Mes — Lucas Levitan & Fabián Jiménez, 4 min 19 sec
• Between Before and After — Sifang Chen & Yang Wang, 8 min 21 sec
• POSTMAN — YUUUKI, 7 min
• Sal (煞) - Divine Retribution — Minju Choi, 7 min 39 sec
• Simiyu, Lies and Happiness — John Kirk Kelly

Character consistency and production itself are no longer the question

The first thing I noticed was that maintaining character consistency or finishing a piece from start to finish is no longer the audience's first concern. Just a year or two ago, faces, costumes, and spaces shifting from shot to shot in generative video got in the way of simply watching the work. This time, instead of counting technical flaws, I found myself watching how one scene led into the next, what the characters were feeling, and why the camera was positioned where it was.

This shift doesn't simply mean the visual quality improved. It means the video has started to feel like cinema. Generative video has moved past the stage of grabbing viewers with the sheer novelty of the output, and has started pulling them along through the timing and rhythm of scenes.

Editing is now what matters most

What stood out most in this screening was the importance of editing. What determined the impression a film left wasn't whether AI could generate a single scene, but which scenes were chosen among many, in what order they were arranged, and where they cut. The under-10-minute limit made this even clearer. A string of beautiful images alone doesn't make a film. It's the gaps between images, the length of the cuts, sound and silence that build a story.

Filmmakers now need to be not just people writing prompts for a generative model, but people designing rhythm in the editing room. Between having many good shots and having a good film, there is still a huge threshold called editing.

The most beautiful works caught the eye, but what stayed longest were works with their own worldview

Because of the 10-minute limit, visually beautiful works drew my eye first. Having to convince the audience of a world in a short span means that films with strong color, texture, lighting, and movement grab attention quickly. But looking back after the screening ended, as a cinephile I also felt some disappointment in works that skillfully followed the paradigms and mise-en-scène established by existing film auteurs.

Reproducing a familiar cinematic grammar well is different from making a film that is truly one's own. What was more welcome in this lineup were the moments when a work pushed a distinctive sensibility and worldview instead of copying the surface of legacy works. The next stage of AI video will likely open not from the question "who does this resemble?" but from "is this a world only this director could have made?"

A summer carried by the memory of a camcorder: Costa Verde

Among the screened works, one that anyone who has actually used a camcorder would especially recognize was Costa Verde. According to Runway's official description, this film is set in Corsica in the 1990s and follows a boy who gets his first camera and uses it to document a summer with his grandparents. The moment the screen showed the grandfather and grandmother standing side by side, it became clear in a single shot what this film was trying to do. The texture reminiscent of VHS and 16mm home movies, the whirring sound suggesting a zoom lens turning, the moment the focal plane shifts — these went beyond a simple retro filter and functioned like an actual memory captured on a camcorder.

That effect wasn't decoration imitating the past. It turned the very act of holding a camera and watching someone's life into a kind of fantasy. Even though it was an AI-generated scene, it created the sensation of memory — as if someone really had been holding a camera that summer.

A prince who slips outside the fairy tale: Where Knights Fall

Another film that stuck with me strongly was Where Knights Fall. The worldbuilding and mise-en-scène ideas in this piece were especially good. Runway's official description explains that Rapunzel waits at the top of a tower, while at the base lies a pile of bodies — knights, princesses, and princes who came before and fell — and finally a man arrives, but instead of climbing the tower, he crawls and slips.

This is a parody that inverts the familiar fairy-tale structure of "the long-haired princess waits in the tower, and the prince goes to save her." What makes this piece interesting is that it turns the grand heroism of the original narrative into a ridiculous, physical chain of failure. It also shows that AI comes closer to finding its own voice not by reproducing the surface of classic fantasy, but by twisting its grammar.

Next comes subtle emotional performance and deliberate camera intent

Going forward, results seem likely to improve the more specifically prompts incorporate subtle emotional performance from actors and deliberate camera effects. It's not enough for a face to simply not fall apart. Tiny shifts — hesitation right before speaking, the speed of a gaze breaking away, an expression that tries and fails to hide emotion — change the meaning of a scene.

The same goes for the camera. Whether the same character is viewed from a distance with a long lens, followed with a handheld camera, or isolated in a fixed wide shot changes the distance the audience feels. Generative prompts now need to move beyond ordering up a "beautiful scene" and become closer to a director's notes, dictating the relationship between actor, camera, and edit.

A new chapter opening for independent filmmakers

Overall, this is clearly different from a year or two ago. After watching this screening, I felt that AI video is no longer a genre defined merely by "the fact that it was made with AI." It has already started to feel like cinema.

That shift could open a new chapter for independent filmmakers. Worlds that used to exist only in someone's head, out of reach because of production budgets, equipment, or location scouting, can now actually be tested on screen. Of course, as the barrier to production drops, the bar for direction rises. Filmmakers now have to explain not what can be made, but why it should be made in this particular shot and this particular rhythm.

Today's conclusion is simple. AI video has moved from the stage of proving that production is possible to the stage of proving cinematic choices. The works that will endure won't be the ones that most convincingly copy an auteur's mise-en-scène, but the ones that leave audiences with new memories through their own sensibility and worldview.

What became visible only after watching all ten together

The ten films used different genres and visual grammars, but they shared a common direction: hiding AI within the scene rather than putting it front and center. Instead of continually reminding the audience "this was made with AI," they tried first to convey a character's actions, the mood of a space, and anticipation for the next cut. The technology was becoming part of the production process rather than the headline subject.

This point matters a great deal. In early AI video works, the visible traces of the generative tool were themselves the spectacle — the way the screen distorted, unrealistic movement, moments where an image suddenly morphed into another form drew the audience's attention. This time, the opposite was true: works lasted longer in memory when those devices didn't get in the way of the story. Audiences stopped checking every frame for signs of generation and instead followed what the scene was trying to say.

Shorter runtimes reveal a director's attitude

The under-10-minute length wasn't merely a screening condition. It was an editorial condition that revealed what each director chose to give up. They had to decide whether to explain a character's past through a long line of dialogue or leave it in a single expression, whether to lay out the worldview across several scenes or imply it through the texture of one scene.

In a short film, a single scene has to do more than one job. It has to introduce a space while showing a character, and convey a character's emotion while setting up the next event. So even a work full of beautiful images can feel disjointed if the relationship between scenes is weak. Conversely, one or two strong images tied to the direction of the story leave a much longer afterimage.

This is also why I felt editing mattered most in this screening. The ability to generate good scenes with AI is becoming basic craft that many directors can now share. The difference lies in when to show a scene and when to withhold it, how long to hold it, and what sound to pair it with.

Different cinematic approaches across the films

A Face Only A Mother Could Love and THE WELL stood out for establishing characters and events clearly within a short time. One centers on appearance, gaze, and mismatched relationships, while the other builds its world through a fairy-tale-like event and mystery. Both leaned toward making the audience curious about what comes next, rather than simply listing the details of generated imagery.

TAIRELL ISN'T REAL and Little Mes deal more directly with questions of identity and choice. A character whose very existence is doubted in the online world, and a character who wonders who they might have become had they made different choices — both pair naturally with "generated images" as a subject. But what matters here isn't the use of AI as a topic, but the attempt to turn that topic into a drama of a character's anxiety and regret.

Between Before and After and Simiyu, Lies and Happiness read as works dealing with time, memory, and the distance between people. Notably, within their short runtimes, they tried to capture a single moment after much has already happened, rather than explaining the entire history of a relationship. In works like these, an actor's small expressions, the direction of a gaze, and the timing of a scene's end become crucial.

POSTMAN made an impressive connection between the vanishing culture of letter-writing and emotional recovery, while Sal (煞) - Divine Retribution brought a distinctly Korean subject into the world of generative video, through a Joseon-era village, an uncanny presence, and the fear shared among a mother, a shaman, and the villagers. What made director Minju Choi's work especially meaningful wasn't simply that it included a Korean setting. It was that it pushed how fear and belief spread within a community through visual atmosphere.

Beauty is the entry point, worldview is the afterimage

Honestly, right after the screening, the visually beautiful works were what came to mind first. Color, lighting, the scale of space, and the smoothness of movement can win an audience over in a short time. In works under 10 minutes especially, visual polish becomes the first impression.

But thinking it over again after some time, beauty alone isn't enough. No matter how refined an image is, if the rules of the world that image belongs to aren't visible, the impression fades quickly. Conversely, even if the texture of a scene is a bit rough, if there are actions and emotions possible only within that particular world, the work lingers.

That's why, in this lineup, I felt greater potential in works that pushed their own sensibility rather than skillfully reproducing an established auteur's mise-en-scène. A work whose reference points are recognizable tells you its starting point, but what matters more is what was newly built from there. For AI to avoid becoming just a tool for redrawing past cinema, a director's personal memories, odd tastes, and hard-to-explain worldview need to remain visible on screen.

What's needed now are prompts that direct the camera

Going forward, generative video prompts need to go beyond simply describing the subject and background. Even with the same character standing in the same room, the emotional tone of a scene changes entirely depending on how the camera looks at that character.

A long lens creates a compressed sense of distance between character and viewer, while a wide lens shows both the structure of a space and a character's isolation at once. Handheld creates instability and immediacy, while a fixed camera emphasizes that a character has nowhere to run. A slow push gradually tightens emotion, while a sudden pan forcibly redirects the viewer's gaze toward an event.

The same applies to performance. Rather than writing "a sad expression," it's more cinematic to direct a sequence of actions — closing the lips and looking away to hold back tears, then finally looking back at the other person. Emotion arises not from a single expression but across time.

What independent filmmakers need is not more images

The idea that AI will open a new chapter for independent filmmakers became even clearer after this screening. But that chapter doesn't simply mean an opportunity to make more scenes at lower cost. It's closer to meaning that creators can now quickly test scenes that used to exist only in their heads, discard directions that fail, and try different editing structures.

In the past, shooting even a single hypothesis required renting a location, gathering cast and crew, and preparing equipment. Now that hypothesis can be checked on video first. In exchange, there's more to judge: whether to accept a generated scene as is, whether to redesign it as if it were an actual shoot, whether to use a scene's flaws as style.

In the end, the competitive edge an independent filmmaker needs isn't knowing generative tools better than anyone else — it's the sense of judgment for choosing what to make. Generative video lowers the barrier to production, but it does not lower the bar for the work itself. In fact, as audiences care less about technical flaws, directors are held to stricter questions about story and direction.

The conclusion I'm left with after seeing it in Seoul

This screening wasn't a place to ask whether AI video can become cinema. It was a place to ask what works that already look like cinema still need to acquire. The answer was editing, emotional performance, deliberate camera intent, and a worldview of one's own.

In AI video from a year or two ago, the fact of having been made at all was itself the event. Now the event comes after that. The audience starts to ask which scene a director chose out of countless generated results, which scene was discarded, why this character falls silent at this particular moment, and why the camera stops here.

The works that can withstand that question are the ones that will survive. Technological progress is already giving many people a similar starting line. The way to get far from that starting line is not to stack up flashier images, but to hold on, to the very end, to one's own rhythm, sensibility, and worldview.

This article is a firsthand account written after attending the Runway AIF 2026: Seoul special screening held on August 7, 2026, at the Seoul Film Center. Film titles, directors, runtimes, and official synopses were cross-checked against the official Runway AIF 2026 page.