LocalMiniDrama: An Open-Source Free Local AI Short Drama Generator 196

LocalMiniDrama is an open-source, local ai short drama and comic drama generation tool developed by xuanyustudio. It integrates the entire workflow from scriptwriting to video synthesis into a single desktop application, and runs entirely on your own API keys. The whole project is built from scratch in pure JavaScript, with the main selling points of "data stays local, no subscriptions required, fully open source." The latest version is v1.2.8, released on July 1, 2026.

Github address:https://github.com/xuanyustudio/LocalMiniDrama

LocalMiniDrama Main Features

The entire production process is divided into eight stages. You can go through them step by step, or run them all automatically with one click:

1. Input a story outline and style, and AI will produce a multi-episode script;

2. Manually edit or modify the script in the episode view;

3. Automatically identify characters in the drama, and generate a portrait image for each one;

4. Extract scenes from the script and generate corresponding background images;

5. Similarly extract or manually add props, and generate prop images;

6. Generate a storyboard script by episode, including shot size, camera movement, dialogue, and other information;

7. Generate a still frame image and a short video clip for each storyboard shot;

8. Finally, automatically merge all clips into a complete episode.

In addition, there are several practical features:

• The entire pipeline supports automatic retry on failure (up to three times), so you don't have to start over when hitting API rate limits;

• You can export the entire project as a ZIP archive, making it easy to back up or migrate to another computer;

• Supports three aspect ratios: 16:9, 9:16, and 1:1;

• Two storyboard modes are available: classic storyboard and full-featured storyboard;

• The prompt supports the @imageN syntax to reference multiple reference images;

• Video clips can link the first and last frames for more natural transitions;

• You can export the storyboard table as an HTML page for easy viewing or printing;

• The prompts for all images and videos support manual editing—tweak whatever you want.

Canvas mode is a highlight enhanced in v1.2.8. It provides an infinitely large canvas where each storyboard shot is arranged like cards in a column. You can drag and drop to adjust positions, and edit scripts or pull out characters, scenes, and props directly on the canvas. After selecting multiple shots, you can create a "workflow group" to regenerate the selected images, videos, or voiceover all at once. The canvas layout is automatically saved, so it looks the same as the last time you opened it.

On the model side, it supports multiple providers including Tongyi (Alibaba Cloud DashScope), Volcano Engine (Doubao), Kling (including Omni), Vidu, and Gemini. The text, image, and video models can be configured separately and don't affect each other. For example, you can use Tongyi Qwen for scriptwriting, Doubao for character image generation, and Seedance 2.0 for video.

Who is LocalMiniDrama For?

• Creators who want to make short dramas with AI on their own, with no programming knowledge required—just download and get started;

• People making comic dramas or dynamic comics, as this workflow supports those as well;

• Self-media creators or small studios that need to produce short dramas in bulk and manage multiple projects separately;

• People who care about data privacy—every asset and project file lives only on your own computer;

• Tech enthusiasts who want to study them or do further development, with the code fully open.

LocalMiniDrama Pros and Cons

Pros

• Data is stored entirely locally (SQLite plus local files), with nothing sent to the cloud;

• The software itself is free, with no subscription fee—you only pay for the API usage;

• Just download the Release package and run it, no environment setup needed;

• Covers every step from story to finished film, so there's no need to jump between different tools;

• Supports multiple AI service providers, which you can mix and match based on cost-effectiveness;

• List view is great for fine-tuning, while canvas view is suited for batch operations—and the two modes share the same data;

• The code is open source, so users with the skills can modify it on their own.

Cons

• Video and image generation must rely on third-party APIs, so it's not fully offline;

• API calls incur costs, especially video generation which is billed by duration, so watch how much you use;

• Locally it only does orchestration and composition, so no high-performance GPU is needed, but network and API stability will affect the experience;

• For newcomers, configuring APIs for multiple models and understanding the canvas workflow takes a bit of time;

• The community ecosystem is still in an early stage, with relatively few third-party tutorials and plugins.

How to Use LocalMiniDrama

1. Go to the GitHub Releases page and download the latest version's archive, then run the exe or app file directly after extracting.

2. After opening the software, click "AI Config" in the top right corner, and set up the provider and API Key for the three tabs—text generation, image generation, and video generation.

3. Taking Alibaba Cloud DashScope as an example: set Base URL to https://dashscope.aliyuncs.com/compatible-mode/v1, enter your own sk-xxx for the Key, use qwen-plus as the recommended text model, wanx2.1-t2i-turbo for images, and wan2.1-t2v-turbo for video. New users usually get a free quota, so you can test with that first.

4. Create a new project, enter a story outline and style, click "Generate Script," and the AI will produce multi-episode content that you can edit directly.

5. Generate characters, scenes, and props in order (you can also skip these and add them later).

6. Enter storyboard generation, pick a storyboard mode, and the AI will write out the details of each shot by episode.

7. Click "Generate Image/Video," and each storyboard shot will produce a still frame and a video clip.

8. Finally, click "Compose Video," wait for the merge to complete, and you'll find the finished film in the output directory.

If you'd like to use canvas mode, just click "Switch on Canvas" on the production page or the episode details page. On the canvas you can freely drag nodes, and after selecting shots right-click to create a workflow, check the items to be regenerated (images, videos, or voiceover), and rerun the whole group with one click.

LocalMiniDrama vs Similar Tools

Why Choose LocalMiniDrama

If you care about keeping your data off other people's servers and don't want to be locked into a specific platform, LocalMiniDrama is a very practical choice. There's no subscription fee, and you can switch to a cheaper API provider at any time. From writing the story to producing the final video, everything is done within a single tool, saving you the hassle of moving files between multiple applications. The code is fully open source, so individuals or teams can easily customize their own workflow. The canvas mode offers an editing experience similar to professional video editing software, rather than being a tool limited to just clicking a "Generate" button.

FAQ

• Q: Is the software free?

  Yes, it's completely free and open source. You only pay for the AI models you call, such as API fees for image generation and video generation.

• Q: What computer specs are required?

  Because it's a desktop application, it doesn't consume many resources itself, and no dedicated GPU is needed. Video generation runs through cloud APIs, so there are no special requirements for local hardware—just an internet connection is sufficient.

• Q: Which AI providers are supported?

  Currently supported are Alibaba Cloud DashScope (Tongyi), Volcano Engine (Doubao), Kling, Vidu, Gemini, Agnes AI, etc. More may be added in the future.

• Q: Is my data secure?

  All project files, images, and videos are stored in a local folder on your computer. The database is in SQLite format, and nothing is actively uploaded anywhere.

• Q: What's the difference between Canvas mode and List mode?

  List mode is suited for finely editing parameters on a shot-by-shot basis. Canvas mode provides a visual interface where you can drag nodes, select areas to batch-rerun, and conveniently adjust an entire segment at once. Data is synced between the two, so you can switch anytime.

• Q: How do I get an API Key?

  For Alibaba Cloud, log in to the Bailian Console, enable the model service, and create an API Key. Volcano Engine is similar—just generate one in the Ark Console.

• Q: How is the quality of the generated videos?

  It depends on which specific model you choose. The project examples include videos generated with Jimeng 1.0, as well as Seedance 2.0, Tongyi Wanxiang, Kling, etc. Newer models generally produce better results, but also come with higher costs.