SpiritFruit
SpiritFruit

End-to-end AI short-drama production — from scriptwriting to storyboard breakdown, character and scene extraction, and visual description generation, outputting a structured production plan ready for AI image generation and video compositing

Upload files

Upload a script document or story outline

The AI short-drama workspace that turns script outlines into characters, storyboards, and finished films

Input a script or a story outline and get character, scene, and prop assets, shot-by-shot frame prompts, and storyboard images and videos ready to composite into a finished film.

No sign-in needed until you send something
About this skill

What this AI short-drama production tool can do for you

SpiritFruit (spiritFruit) is an open-source AI short-drama production workspace for teams that need to produce short dramas, anime, and short-form video in a systematic way, keeping the entire pipeline from script to finished film within a single project.

Usually all you have is a script or a story outline, and from there you need to break out characters, define scenes, list props, split storyboards, write visual descriptions, and then send each off to generate images and videos — many steps and scattered tools, where missing any one of them stalls the whole line. SpiritFruit strings these steps into a fixed seven-stage flow: create a short-drama project; enter a script or have AI generate one and keep writing it chapter by chapter; extract characters, scenes, and props from the script; generate the corresponding character, scene, and prop images; split the storyboard and generate frame prompts for each shot; generate storyboard images and storyboard videos; and finally merge the clips and export a finished film.

What you get along the way are structured intermediate outputs, not a chat log: characters, scenes, and props each become their own entries with matching images, the storyboard is split and carries frame prompts, shot images and videos are archived by shot, and everything finally merges into a film you can export directly. Generation jobs run asynchronously, each carrying status, progress, results, and error records; long jobs can be cancelled, and repeat submissions are not dispatched twice.

The right fit is clear: small teams making their own short dramas or anime shorts; creators who want to turn a script into storyboards quickly and feed them to image and video models; and technical teams that want to build on top of a ready-made pipeline. It is not suited to one-line-to-video use cases, nor to material that doesn't need storyboard breakdown — it is a workspace, and you need to set up the project, script, and assets according to the flow before it ships reliably.

How it differs from other AI short-drama production tools

General-purpose AI chat tools can handle anything, and tweaking a line of dialogue or checking a setting is easy. But they have no concept of a project: keeping characters, scenes, and storyboards consistent within the same drama means repeating constraints in every conversation, and a new round of generation can come out inconsistent with what came before. SpiritFruit makes characters, scenes, props, and storyboards fixed entries in a project, with extraction and generation happening in separate steps and each step's outputs staying in place, so when something goes wrong you can rerun just that one stage instead of asking all over again from scratch.

Point-solution AI image or video tools can get a single shot right, but you have to wire together the links from script to finished film yourself: who appears in which shot and which character image that shot should use is all manual cross-checking. SpiritFruit connects them into one chain from script to finished film — first extract characters, scenes, and props, then generate assets, then split shots, write frame prompts, and produce images and videos, and finally merge and export. For teams that keep updating episode by episode and chapter by chapter, what it mainly saves is the repeated cross-checking and copying.

Another difference is that the jobs themselves are visible. With batch generation, the real headache is often not slow generation but not knowing which item is stuck. Here every generation job has status, progress, results, and error records; failures can be retried, long jobs can be cancelled, and repeat submissions are not dispatched twice.

How to choose: if you produce systematic short dramas or anime shorts and need assets and storyboards you can go back and edit or regenerate, choose SpiritFruit; if you just want to turn a one-liner into a video on the fly, or your content doesn't need storyboard breakdown, a general-purpose AI tool will be less trouble.

Frequently asked questions about AI short-drama production

9 questions

What is SpiritFruit?

SpiritFruit (spiritFruit) is an open-source AI short-drama production workspace that manages scripts, characters, scenes, props, storyboards, images, videos, and finished films in a single flow. It targets short-drama, anime, and short-form video production, and can serve both as a ready-made production backend and as a foundation for further development.

What kind of material can I give it?

You can enter a script yourself or have AI generate one and keep writing it chapter by chapter; images, videos, and other assets are uploaded through the backend. Uploaded assets go through the same asset management as AI-generated images and videos, while characters, scenes, and props are extracted from the script.

What results will I get?

You get four kinds of output: entries and matching images for characters, scenes, and props; a split storyboard with frame prompts for each shot; storyboard images and storyboard videos generated per shot; and a merged, exported finished film. These outputs all stay in the project and can be viewed and rerun stage by stage.

Can the generated storyboard clips be composited into a single finished film?

Yes. Once the storyboard videos are generated, you can merge the clips and export a single finished film in the workspace. The finished film, storyboard videos, storyboard images, and assets are all saved under the project, making it easy to swap shots or recomposite later.

Can I change the models used for generation?

Yes. Providers and keys for text, image, and video are configured separately, and whichever vendor's key you enter is the one that gets called — aggregator services or your own service both work. The project does not lock generation capability to any single model.

What if a job gets stuck halfway through?

Generation jobs run asynchronously. Every job has status, progress, results, and error records; failures can be retried and long jobs can be cancelled. Identical jobs that are queued or running reuse the same one, so repeated clicks are not dispatched twice.

What environment do I need to run it?

The project provides a containerized deployment: a single command brings up the full environment, including the backend API, asynchronous job processing, database, cache, and front-end pages. On first launch it automatically creates tables and initializes an admin account, and the deployment docs cover backup, update, and troubleshooting commands.

I only have a rough idea, not a complete script — is that okay?

Yes. The script can be entered by you or generated by AI, and supports continued writing chapter by chapter. It's great for turning a one-line idea or an outline into a story structure that can be broken down and then sent off for image generation and video creation.

How is this different from just asking an AI chat?

General chat tools can answer anything, but their output structure is unstable each time, and character settings and storyboard numbering easily end up inconsistent. SpiritFruit fixes extraction, storyboarding, and image generation into steps, with outputs staying in the project, so when something goes wrong you can pinpoint the exact stage and rerun only that step.