AI Creation & Media ·
OiiOii 2.0: I Finally Took AI Video Creation Seriously
English edition based on the Chinese original.
English edition based on the Chinese original.
Date: 2026-06-15
I have played with AI for a long time, but I had never seriously worked through an AI video creation flow.
It is not that I was not interested. It is that video work is exhausting the moment you think about it: script writing, image generation, image-to-video, revisions, editing, voiceover, and final assembly.
Those steps at least have a recognizable workflow. AI video is harder because the output is sensitive to motion, pacing, shots, characters, sound, and continuity.
I had seen many creators recommend OiiOii for ACG, short drama, animation, media accounts, and commercial video. After the OiiOii 2.0 upgrade, I finally tested it seriously.

My conclusion is simple: the most attractive part is that OiiOii breaks video creation into something closer to a real creative team workflow.
When you enter an idea, it is not just one generic agent behind the input box. The process has roles that resemble an art director, screenwriter, character designer, scene designer, storyboard artist, and sound director. The idea is pushed toward a finished video step by step.
OiiOii 2.0 has three core upgrades:
- Shot reconstruction: upload a video, let the system analyze its shot language across professional dimensions, then replace subjects, characters, and scenes.
- Smart canvas: the agent understands the whole canvas, and you can revise characters, scenes, storyboards, music, and local details directly.
- Skill library: built-in creation kits for media accounts, advertising, marketing, merchandise design, and related scenarios.

I tested two workflows. The first was creating a video from scratch with the smart canvas. The second was using shot reconstruction to remix a famous World Cup moment into an AI-industry parody.
For the reconstruction test, I chose Cristiano Ronaldo's free-kick moment against Spain in the 2018 World Cup.

The scene has a clear dramatic structure: Portugal is behind, the pressure is high, everyone is watching Ronaldo, and one shot changes the result.
That makes it a good source for short-form adaptation.
So I turned Portugal versus Spain into OpenAI versus Anthropic, with Sam Altman and Dario Amodei standing in for the two sides. The football match became a symbolic AI-company showdown.
The useful part of OiiOii is not that it magically solves every video problem. It still needs direction. You still need to decide the story, inspect the output, and make revisions.
But it gives video creation a more workable shape. Instead of throwing a prompt into a black box, you get a process with roles, assets, scenes, and revision points.
That is why OiiOii 2.0 is worth watching. AI video will not become serious only because models can generate prettier clips. It becomes serious when the workflow is controllable enough for creators to keep iterating.