Model or dataset
freestylefly/director_ai avatar
freestylefly/director_ai

director_ai: a Flutter comic-drama app wearing a starter template's README

AI 漫剧制作 APP,能一键生成剧本、分镜及合成视频,让你在手机上也能快速制作漫剧。

1,761 stars368 forksPythonLicense varies

At a glance

What is it?
The repository ships a Flutter skeleton for an AI comic-drama generator, a sponsorship line pointing at a hosted model platform, and a README that describes React and Vite instead.
Who is it for?
director_ai is a mobile-first Flutter project for turning a prompt into a comic-drama clip, and the repository is genuinely interesting to read even though the documentation is not. The tree tells you it targets Android and web from one Dart codebase, the sponsorship line tells you the generation runs against a hosted model platform rather than on the phone, and the file list tells you there is a docs/ directory that a landing page never mentions.
Can I use it commercially?
Not without permission. GitHub finds no licence file in the repository, and without a licence all rights are reserved by default: you may read the code but not reuse it. Check the README, or ask the authors, before using it.
Is it still maintained?
Yes. The repository last received commits 158 days ago.
What is it written in?
Mainly Python, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on September 20, 2026, and from our analysis. They are not legal advice.

Editorial analysis

A Flutter project behind a Vite README

The clearest thing to say about freestylefly/director_ai is that its documentation and its source tree describe two different projects. The repository description calls it an AI comic-drama production app that generates scripts, storyboards and a composited video, with the phone as the target device. The tree agrees with that description. `lib/`, `android/`, `web/`, `pubspec.yaml` and `analysis_options.yaml` are the standard shape of a Flutter application compiled for mobile and web from one Dart codebase, and there is a `docs/` directory beside them.

The README file disagrees with all of it. Its first line is `# React + TypeScript + Vite`, and everything after that is the boilerplate that ships with the Vite starter: a note about two official Fast Refresh plugins, a section on expanding the ESLint configuration, and two JavaScript fences showing how to enable type-aware lint rules. Nothing in it mentions Dart, Flutter, comic-drama tooling, or any of the features in the repository description.

For anyone deciding whether to fork this, that gap is the first thing to reconcile. What the repository contains looks like a real application. What the README describes is an unmodified JavaScript web starter, so the most likely explanation is that a template was used to bootstrap the project and the documentation was never replaced before publishing.

What the file tree tells you about the app

With the README out of the picture, the tree becomes the useful document. `pubspec.lock` sitting next to `pubspec.yaml` is the marker of a Dart package that pins its dependency versions, which is how you know this is Flutter rather than a JavaScript or Python project even even though the language field on the repository is recorded as Python. `android/` and `web/` are the two platform folders a Flutter build targets, which fits the claim that the app is meant to be used on a phone.

Two other entries are worth pausing on. There is a `test_api.py` at the repository root, a single Python file sitting outside the Dart structure, and the language field on the repository says Python. Some mobile app backends ship a small Python service alongside the client, so this file may be the seam where the Flutter app hands off generation work to a hosted endpoint, though nothing in the repository confirms that.

The `.claudeignore` file is a smaller signal but a telling one. It suggests the project was developed with an AI coding assistant in the loop, which fits the category the repository describes itself as being in. The `images/` directory is the only place where visual material lives, so any screenshots or sample output from the app are there rather than in a documentation page.

The sponsorship note is the most informative line in the README

Buried between the Vite title and the plugin list is one line that is actually about this project. The README carries a sponsorship credit for a Chinese-language AI aggregation platform, linked at `ciyuan.today`, described as a place to use top international models at low cost, with GPT Image 2 and nano banana named as examples.

That single line tells you more about the architecture than anything else in the repository. An image generation model large enough to draw comic-drama frames does not run on a phone, so the generation is happening somewhere else. The sponsorship arrangement suggests the app calls a hosted endpoint for both text and image work, with the Flutter client handling input, sequencing and playback. The alternative, a model bundled into the app, is not plausible at that model size.

It is also a caution about how to read the project. A repository that credits a paid model aggregator in its README is telling you the interesting engineering is in orchestration: turning a script into a set of shots, then a set of shots into a video file. The model itself is somebody else's, and the quality ceiling of the output is set by which endpoint the app is configured to call.

The lint configuration belongs to the template

The two JavaScript fences in the README deserve a note, because they are the kind of detail that sends people down the wrong path. Both show an `eslint.config.js` file in flat-config style. The README's version wires up React-specific lint plugins from the Rel1cx project. Removing only the placeholder comment lines leaves this:

javascript
import reactX from 'eslint-plugin-react-x'
import reactDom from 'eslint-plugin-react-dom'

export default tseslint.config({
  extends: [
    reactX.configs['recommended-typescript'],
    reactDom.configs.recommended,
  ],
  languageOptions: {
    parserOptions: {
      project: ['./tsconfig.node.json', './tsconfig.app.json'],
      tsconfigRootDir: import.meta.dirname,
    },
  },
})

Every line of that comes from the README, but none of it applies to this repository. There is no `package.json`, no `eslint.config.js`, and no `tsconfig.app.json` in the tree, and a Dart project lints through `analysis_options.yaml` instead. The other fence points at `tsconfig.node.json` and `tsconfig.app.json`, which are Vite's TypeScript project files. Both fences are starter scaffolding, so treat the JavaScript in this README as belonging to a different project than the one hosted here.

Stars without releases, homepage or topics

The repository metadata is where the picture gets interesting. It has 1,761 stars and 368 forks, which is a substantial audience, against 8 open issues. There is no homepage URL, no list of releases, no topics set, and the default branch is `master` rather than `main`, which is a small indicator of when the repository was first created. The repository is not archived, and the most recent push is dated 2026-05-01.

A fork count that is a fifth of the star count suggests people are not just watching. This is the profile of a project people copy and adapt for their own use, which fits an application template where the interesting part is the workflow rather than the codebase. It is also the profile of a project whose users are not being asked to report problems, since 8 open issues across that many stars is a low number for anything with real daily use.

The absence of releases is the part that affects a decision most concretely. There is no version to pin, no changelog to read before upgrading, and no published artifact to download. The `pubspec.lock` file is the closest thing to a version marker, and it describes dependencies rather than the app itself.

How to evaluate a project whose documentation is unfinished

The practical route for this repository is to read the tree instead of the landing page. `pubspec.yaml` is where the Flutter package name, the Dart SDK constraint and every dependency live, and it is the single most informative file in the project. If the app talks to a hosted model endpoint, its HTTP client and any streaming logic will be in `lib/`, and the screen structure will be visible in the widget tree.

`docs/` is the most interesting directory to open first, because the presence of a docs folder in a project whose README never mentions it usually means the author intended to document something and did not finish. That directory may hold the real usage notes, or it may hold only scaffolding.

The `images/` directory tells you what the output is supposed to look like, which for a comic-drama generator is more informative than a feature list would be. A reader trying to judge whether this app is worth trying will learn more from a few sample frames than from any description of its settings.

What the README does not do, and what no external documentation page is linked from here to fix, is state the model provider, the cost structure, or whether generated content can be exported in a usable format. Those are the questions that decide adoption for a tool like this, and they are unanswered.

Editorial conclusion

director_ai is a mobile-first Flutter project for turning a prompt into a comic-drama clip, and the repository is genuinely interesting to read even though the documentation is not. The tree tells you it targets Android and web from one Dart codebase, the sponsorship line tells you the generation runs against a hosted model platform rather than on the phone, and the file list tells you there is a docs/ directory that a landing page never mentions. The last push was on 2026-05-01 and no tagged release exists, so the honest starting point is pubspec.yaml and lib/, not the README.

Frequently asked questions

What is director_ai and what technology is it built with?

It is an AI comic-drama production app that generates scripts, storyboards and assembled video for phone use. The repository tree points to Flutter and Dart: pubspec.yaml, analysis_options.yaml, lib/, and the android/ and web/ platform folders. The README is misleading here, because it is still the React and Vite starter template.

Does director_ai run AI models locally or call a hosted service?

The README credits a Chinese AI aggregation platform for sponsorship and names GPT Image 2 and nano banana as the models it provides access to. Combined with the phone-first design, that points to generation happening through a hosted endpoint rather than on the device. No model code appears in the repository tree.

How do you build and run director_ai?

The README contains no build or install command, which is a direct consequence of it still being the Vite template. The files that describe how this project builds are pubspec.yaml and the android/ and web/ folders, so the Flutter toolchain is what applies here. Reading pubspec.yaml is the fastest way to confirm the package name and dependency set.

How active is the director_ai project?

It is not archived and the last push is dated 2026-05-01. The repository has 1,761 stars, 368 forks and 8 open issues, but it publishes no releases, sets no homepage and carries no topics, so there is no version to pin and no changelog to check. Fork traffic is high relative to issues, which reads more like adaptation than active upstream maintenance.

Official sources

  1. freestylefly/director_ai on GitHub
  2. Issues
  3. README
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/freestylefly-director-ai.svg)](https://hysenlabs.com/projects/freestylefly-director-ai)