The Director Studio Speaks Five Languages: Chinese, English, Spanish, Arabic and Japanese
Yes — 墨川导演台 (MiniMax H3 Director Studio) works natively in five languages: Chinese, English, Spanish, Arabic, and Japanese. That means you can brainstorm a script, generate a six-part storyboard draft, and manage your project without ever leaving your own language. If you're building AI short videos or short dramas for more than one market, this is the feature that quietly saves you the most time.
Below I'll walk through how the multilingual side actually works, where it helps most, and how to set it up so your scripts, storyboards, and voiceovers all line up.
Which languages does the studio actually support, and where do they apply?
The studio supports five working languages — Chinese, English, Spanish, Arabic, and Japanese — and they apply to the AI screenwriter and conversational features, not just the interface. In practice, you can chat with the AI screenwriter in any of these five languages, and it will reply in the same language while writing the draft into your script library.
Here's the practical breakdown:
- AI screenwriter (对话式编剧): Chat in Chinese, English, Spanish, Arabic, or Japanese. The AI responds in your language and produces a six-part script plus storyboard draft.
- Script library (剧本库): Stores scripts, episodes, and storyboard cards. You can import existing scripts or start new ones — useful if your team writes in one language and shoots in another.
- Project library (项目库): Organizes generation tasks and versions by project, with single-shot and batch video generation.
- Asset library (资产库): Manages characters, scenes, props, and reference videos — the reference images you upload here are what keep characters and scenes consistent across shots.
- Audio library (音频库): Voice cloning, lip-sync, background music, and sound effects.
- AI voiceover (AI 配音): IndexTTS 2.5 multilingual voiceover with emotion control.
The key thing to understand: the five languages are a *workflow* feature, not a translation gimmick. You're not writing in English and then pasting into a separate translator. You stay in one language from idea to storyboard, and the studio keeps the structure intact.
💡 One note on the conversational features: the AI screenwriter and chat functions can connect to your own LLM API key. So if you have a preferred model for a particular language, you can wire it in — check with the official site or customer service (WeChat: ymcandai) for the current setup.
How do I write a script in my own language and still get a clean storyboard?
Answer: start the conversation in your target language and let the AI screenwriter return a six-part script plus storyboard draft in that same language, which gets written directly into your script library. The trick is to be explicit about language and format in your very first message.
Step by step:
- Open the AI screenwriter and state your language up front. For example: "Write this in Spanish" or "请用中文写" — don't assume it will guess from context.
- Describe the format you want. The studio's screenwriter produces a six-part script structure with storyboard drafts. Ask for it explicitly so the draft lands in the right shape.
- Let it write to the library. The AI replies and actually writes the draft into your script library — you're not copy-pasting from a chat window.
- Review the storyboard cards in the script library, where scripts, episodes, and storyboard cards live together.
- Adjust per shot before you move to generation.
A prompt pattern that works well:
- Language: [Chinese / English / Spanish / Arabic / Japanese]
- Format: six-part script + storyboard draft
- Genre and length: e.g., 60-second short drama, three scenes
- Tone: e.g., warm, suspenseful, comedic
- Constraint: keep character names consistent with my asset library
Because the draft is written into the library rather than left in chat, your storyboard cards become the single source of truth — which matters a lot once you're generating multiple shots.
How do I keep characters and scenes consistent across languages and shots?
Answer: use the asset library and reference images — consistency comes from your uploaded references, not from the language you write in. This is the part most creators underestimate.
Here's the workflow:
- Upload reference images for characters, scenes, and props into the asset library. These references are what anchor a character's look across shots.
- Generate first and last frames with AI image tools (text-to-image and image-to-image) to lock the opening and closing look of each shot.
- Use reference videos where you need motion continuity.
- Generate single shots first, then move to batch generation once a look is approved.
On the cloud side, inference runs through ComfyUI workflows that support reference-image consistency, storyboard continuity, and batch queuing. That's the technical reason your references matter: they're feeding the workflow that keeps your character recognizable from shot one to shot twenty.
A practical habit: keep your character names identical across all five languages in your asset library. If your Spanish script calls a character "Lucía" and your English storyboard calls her "Lucy," you've just created a consistency problem for yourself. Pick one name per character and stick to it everywhere.
How do I handle voiceover and lip-sync for multiple languages?
Answer: use IndexTTS 2.5 multilingual voiceover with emotion control, combined with the audio library's voice cloning and lip-sync tools. This is where a multilingual workflow really pays off — one character, multiple language versions.
The steps:
- Clone the voice you want in the audio library.
- Generate voiceover with IndexTTS 2.5, setting emotion to match the scene.
- Apply lip-sync so mouth movements match the audio.
- Add background music and sound effects from the audio library.
- Version it per language — same character, same emotion, different language track.
For AI music videos, there's a dedicated AI MV feature. And if you're producing a short drama for multiple markets, this is the moment to think about versioning: one visual master, several language tracks.
Do I need a cloud GPU, and how do I connect one?
Answer: not necessarily — you can run locally on your own GPU machine, connect to a cloud GPU server via SSH, or call the studio's capabilities through an API. Multilingual work doesn't change your deployment choice, but cloud deployment makes multi-user collaboration easier.
Your three modes:
- Local deployment: runs on your own GPU computer, inference stays local.
- Cloud deployment: connects to a GPU cloud server you rent, with inference over an SSH tunnel.
- API calls: access the director studio's capabilities through an interface.
If you go cloud, the studio supports general cloud GPU deployment across AutoDL, Alibaba Cloud, Tencent Cloud, Xiangongyun (仙宫云), and custom SSH servers. It connects to a cloud ComfyUI over SSH, with local tunnel ports isolated automatically so multiple users don't collide.
For the Xiangongyun test flow specifically:
- Connect to the Xiangongyun private image (not public).
- Register at https://www.xiangongyun.com/register/B1H9M6 — you must register fresh and enter invite code B1H9M6.
- After registration and verification, send your Xiangongyun account ID to customer service (e.g., ac401472-db02-5df3-xxxx-xxxxxxxxxxx).
- Once your test account is opened, you'll see the director studio image — deploy an instance with it.
- When the instance is running, click SSH on the right side to get the host, port, and password. Enter these into the cloud deployment page's SSH Host / SSH Port / SSH password fields.
- Click "Verify SSH," then "Start Cloud ComfyUI," and wait for the green success message.
- You're ready to write a script and test video generation.
If any of this gets murky, the fastest path is usually the customer service WeChat: ymcandai. You can also find the team by searching 「杨墨川AI」 on Taobao, or 「杨墨川」 on Douyin, Xiaohongshu, and WeChat Channels. Support email is info@ymcdirector.com.
FAQ
Q: Can I write the script in Arabic and still generate video with English voiceover?
A: Yes — the five languages cover the AI screenwriter and conversational features, and voiceover is handled separately through IndexTTS 2.5 multilingual voiceover in the audio library. Write in the language you think best in, then produce the voice track you need.
Q: Does the studio translate my script automatically between the five languages?
A: The supported languages are the working languages of the AI screenwriter and conversational features — you chat and receive drafts in your chosen language. Treat it as a multilingual writing and production workflow rather than a dedicated translation tool, and confirm specifics with customer service if you have a particular use case.
Q: Do I need a cloud GPU to use the multilingual features?
A: No. You can run locally on your own GPU machine, connect to a rented cloud GPU via SSH, or use API calls. The language features are part of the studio itself, not tied to a specific deployment mode.
---
If you're producing for more than one language market, the smartest move is to set up your asset library and character names *before* you start writing — that one habit prevents most consistency headaches later. Then pick your deployment mode, write your first six-part script in your own language, and test a single shot before committing to a batch.