LibTV vs Tiangong Short Drama Studio: Two Bets on AI Filmmaking in 2026

Two of China’s biggest AI companies shipped short drama platforms within two weeks of each other in spring 2026, and they could not have made more opposite bets. One handed you an infinite whiteboard covered in nodes and said build your own pipeline. The other handed you a script box and said press the button, the director agent takes it from here. Same goal — turn a script into a finished vertical series — two philosophies that barely speak the same language.

If you’re a creator or a studio trying to decide where to spend your production budget, the choice isn’t about which has the better video model. Both wire in the same handful of frontier engines. The choice is about how much of the pipeline you want to control by hand versus how much you want an agent to decide for you.

Most coverage of these launches just reprints the feature lists. That’s not a decision. Below we put LibTV and Tiangong Short Drama Studio side by side on the axis that actually matters — control versus automation — and then get honest about who each one is for, including the creators neither one is really built to serve yet.

Two answers to the same problem

The reason two heavyweights launched short drama tooling in the same month is simple: the money arrived before the tooling did. Omdia forecast microdramas to generate $11 billion in global revenue in 2025, and Sensor Tower tracked short-drama app revenue climbing from $178 million in Q1 2024 to nearly $700 million in Q1 2025 — roughly a 4× jump in a year. When demand scales that fast, whoever industrializes production wins.

Both platforms are attacking the same three pain points every studio complains about: production takes too long, costs too much, and characters drift from episode to episode. Where they split is the interface philosophy for solving it.

One tool gives you a canvas of nodes; the other gives you a director agent. Same output, opposite control model.

Practical rule: Before comparing features, decide one thing — do you want to drive the pipeline or supervise it? That single answer picks your platform faster than any spec sheet.

LibTV: the infinite canvas and the model marketplace

A quick sketch to put the idea on paper — the diagram below maps the key flow:

Illustration: drawn by the BibiGPT team

LibTV comes from LiblibAI and launched on 18 March 2026 as a video platform built for both humans and AI agents. Its defining idea is an infinite canvas with a node-based workflow: scriptwriting, storyboarding, generation and editing all live on one board, and any node can be modified and reused without redoing the stages downstream of it.

The capability set is broad — over 20 built-in functions including character three-view sheets, a multi-camera nine-grid, and a “plot deduction” four-grid for working out how a scene resolves. Crucially, LibTV is a model marketplace, not a single engine: it brings 30-plus video models into one workspace, including Kling 3.0, Seedance 2.0 and HappyHorse 1.0, so you can route each shot to whichever model renders it best.

There’s a second, quieter innovation: a dual entry system. Human creators use the graphical canvas, while agents reach the same capabilities through a Skill interface — meaning the platform is designed to be operated by software as well as people. That bet is landing commercially: LibTV cleared 100,000 visits on launch day, and by the time its Team Edition arrived in May 2026, more than 300 short drama companies and film studios had become customers.

Practical rule: Choose LibTV if you think in nodes — if the phrase “reuse this storyboard node across three episodes” sounds like a feature, not a chore.

Tiangong Short Drama Studio: the agent-driven director

Tiangong Short Drama Studio (天工短剧工作台) comes from Kunlun Tech’s Skywork team and launched at the end of March 2026 with the opposite instinct. Its headline is “agent-driven automation of the content pipeline” — the promise of input a script, get a finished film.

The mechanism is a team of cooperating agents rather than a canvas of manual nodes. An asset-extraction agent reads your script and pulls out characters and settings; a director agent decomposes the script into shots and generates director-level video, using multi-view reasoning to hold character consistency across scenes. Kunlun frames this as giving AGI short drama production “director’s thinking” for the first time — opinionated structure instead of a blank board.

It’s not fully hands-off, though. Tiangong offers a dual track — an agent storyboard mode and an infinite canvas mode — with six intervenable stages: script upload, asset extraction, blocking/staging diagrams, prompt templates, storyboard script, and video generation. So you can let the agents run or step in at any stage. On models, it curates rather than sprawls, integrating four frontier engines — Seedance, SkyReels, Kling and Vidu.

The catch for most readers: access is application-based. At the time of writing you request access by emailing the team rather than signing up instantly, which makes it more of an enterprise/studio onboarding than a self-serve tool.

Tiangong’s pitch: input a script, let the director agent carry it to a finished film.

Practical rule: Choose Tiangong if you want the tool to make the directorial decisions and you’ll step in only to correct — supervision, not construction.

Node canvas vs one-click agent: how to choose

Strip away the marketing and the two products differ on four axes that actually change your day.

Dimension	LibTV (LiblibAI)	Tiangong Short Drama Studio (Kunlun)
Core philosophy	Node-based infinite canvas — you build the pipeline	Agent-driven — the pipeline builds itself
Control model	Manual, node-by-node, reusable	Automated with 6 intervenable stages
Video models	30+ marketplace (Kling 3.0, Seedance 2.0, HappyHorse)	4 curated (Seedance, SkyReels, Kling, Vidu)
Access	Self-serve, incl. an agent Skill interface	Application-based (email request)
Best for	Studios that want maximum control and model choice	Teams that want director-level output fast

The honest read: LibTV rewards people who like controlling every stage and want the widest model choice; Tiangong rewards people who want a director agent to carry the load and are fine applying for access. Neither wins outright — they’re optimized for opposite temperaments.

Decision filter: Count how many shots you actually want to hand-tune per episode. High number → LibTV’s canvas. Near zero → Tiangong’s agents.

What both leave on the table

Both platforms are built for the professional end of the market — studios, film teams, the 300-plus companies already buying LibTV seats. That focus creates two real gaps for everyone else.

First, the solo creator. If you’re one person with a premise and no studio, a 30-node canvas or an enterprise application form is a lot of friction before your first episode exists. The demand is enormous — Chinese-origin apps alone now hold over 90% of the short drama market in Japan, and the audience for individual creators is wide open — but the newest pro tools aren’t aimed at you.

Second, the series as the native unit. Both tools are excellent at turning a script into a film. Serialized drama is a different problem: episode 41 needs the same lead, wardrobe and voice as episode 1, and a season needs to be planned as a season. That’s a data-model question, not a rendering one, and it’s the lane where a serialization-first tool earns its keep.

This is where a self-serve, episode-native option fits between the two extremes. If you want the season — not the single film — to be the unit you work in, and you’d rather start now than wire nodes or wait on an access email, that’s the niche DramaSo is built for.

Picking your platform

Here’s the shortest path to a decision:

1. You run a studio and love controlling every stage → LibTV. The node canvas and 30-model marketplace are made for you.
2. You want a director agent to do the heavy lifting → Tiangong Short Drama Studio, assuming you’re fine applying for access.
3. You’re a solo creator or small team who wants to ship serialized episodes today → start with a self-serve tool built around the season. Write one premise with an AI script generator, then let an AI episode generator carry your cast forward across the whole run.

If you’re still deciding what to make before you pick a tool, our comparison of AI short drama generators covers the wider field with verified prices and limits.

The category is young enough that the “right” platform is really just the one that matches how you like to work. Node-builders, agent-supervisors and solo serializers are three different creators wearing the same “AI short drama” label — and in 2026 there’s finally a tool tuned for each.

FAQ

What is LibTV?

LibTV is an AI video creation platform from LiblibAI, launched in March 2026. Its defining feature is an infinite canvas with a node-based workflow that unifies scriptwriting, storyboarding, generation and editing, and it brings 30-plus video models (including Kling 3.0 and Seedance 2.0) into one workspace. It also exposes an agent Skill interface so software, not just humans, can operate it.

What is Tiangong Short Drama Studio?

Tiangong Short Drama Studio (天工短剧工作台) is an AI short drama platform from Kunlun Tech’s Skywork team, launched in late March 2026. It uses cooperating agents — an asset-extraction agent and a director agent — to turn a script into a finished, director-level video with consistent characters, while letting you intervene at six stages. Access is currently application-based.

LibTV vs Tiangong: which is better?

Neither is universally better — they optimize for opposite workflows. LibTV suits studios that want manual, node-by-node control and the widest model choice. Tiangong suits teams that want an agent to make directorial decisions and generate a film from a script with minimal input. Pick by how much of the pipeline you want to control yourself.

Are LibTV and Tiangong free?

LibTV is self-serve with a Team Edition for studios that buys resources based on team size and project needs. Tiangong Short Drama Studio is currently application-based — you request access rather than signing up instantly — which positions it toward enterprise and studio users rather than casual solo creators.

Which AI short drama tool is best for a solo creator?

The pro platforms above are built for studios. A solo creator with one premise is usually better served by a self-serve tool where the series is the native unit — one that plans a season, locks each character once, and exports captioned 9:16 episodes without a node canvas or an access application.

---

The bottom line: LibTV and Tiangong Short Drama Studio are two coherent, opposite bets on the same future — one says give creators an infinite canvas, the other says give them a director agent. Both are aimed squarely at studios. If you’re a solo creator or small team who just wants to ship serialized episodes this week, pick the tool built around the season and start your first episode free.