Send us the script
A simple three-point script is enough: from which second to which second / what is on screen / what text appears. If you already have live-action footage, send that too.
AI Video Production | Turn an idea into a film, even without footage
You provide the script, we generate the visuals with AI, then cut them into a finished film through the normal post-production process. What AI does is obtain the footage; everything after that is still editing work.
Some shots simply cannot be filmed, or are not worth filming for the sake of a few seconds: abstract ideas, products that do not exist yet, locations you cannot get into. AI-generated imagery has made those shots possible for the first time.
But one thing needs saying up front: what AI generates is not a finished film, it is material. The generated clips are usually very short, inconsistent in pace and often flawed in the details, and they still have to be selected, cut, graded, subtitled and mixed. Our job is to turn those clips into a video you can actually use.
The two things we offer
1. Script → AI-generated visuals → edited into a finished film
- Working out from your script what each section needs to show
- Generating and sifting through the usable clips — getting two or three usable clips out of ten is normal, and this stage takes up a larger share of the hours than people expect
- Cutting, pacing, transitions and grading, so that clips from different generations look like one film
- Title cards, subtitles, music and mixing
2. AI voice-over, narration and multilingual subtitles
- AI voice-over for Chinese and English narration
- Adjusting pace, pauses and the rhythm of each section, to avoid that “machine reading a script” feeling
- Subtitling, plus output in multiple language versions
- Timing the narration to the picture
These two can be commissioned separately. If you already have the visuals and only need voice-over and subtitles, you are welcome to ask about that on its own.
When AI production is a good fit
- The product does not physically exist yet: you need a video at the crowdfunding, teaser or proof-of-concept stage
- Abstract ideas that are hard to film: data moving, time passing, a mood or atmosphere
- Locations that are hard to get or too expensive: overseas, restricted sites, historical settings
- Short-form social video that needs multiple versions for testing: one script extended into several visual angles
- Courses or presentations that need illustrative visuals: supporting images for explanation, where a real record is not required
- You have live-action footage but are missing a few shots: use AI to fill the gaps and mix the two
How to send the script
The rules for materials are the same as for any other video — follow whichever matches what you have:
1. You do not have footage yet (the usual case for AI video)
A simple script is enough. You only need to write down three things:
- From which second to which second
- What is on screen
- What text appears
For example:
0–5 sec | Opening shot (for example an aerial of the setting, the entrance, the product itself) | Text: the brand or topic name
5–12 sec | Main content, camera pushing in slowly | Text: the one line you most want people to remember
12–20 sec | Closing shot | Text: contact details or a call to action
It does not need to be beautifully written, just clear. The more specific the description of the picture, the higher the hit rate when generating it.
2. You already have some live-action footage
- Upload the original files to Google Drive, then share the folder link with us
- Please put the files in order first (01, 02, 03…), and mark which sections need to be filled in with AI
For anything else, just ask on our official LINE.
About quotes and scheduling: AI generation takes repeated attempts before it produces usable footage, so the hours cannot be worked out directly from “how many seconds the video runs”. We read the script first and establish how demanding the visuals are and how many sections there are; only once we have confirmed what can realistically be done do we provide a quotation and a production schedule.
If we conclude that some sections would be better shot live or made from footage you already have, we will say so plainly — we will not force a project into AI just to make it an AI project.
Process
Production process
Every project is bespoke, so the process starts with “reading the script first”, not with a price.
Assessment and quotation
Once we have read the script we establish how demanding the visuals are and how many sections there are, and provide a quotation and production schedule.
Generating and selecting
We generate the visuals section by section and pick out the usable clips. We will check the direction with you as we go, rather than leaving you to see it only at the very end.
Editing and delivery
Cutting, grading, title cards, subtitles and voice-over mixing, delivered as a finished film you can use straight away.
FAQ
Frequently asked questions
Can Chinese text appear in the picture?
Chinese characters inside generated footage almost always come out wrong, so every piece of text that needs to appear is handled as a post-production title card rather than generated by AI. That way the text can be controlled precisely, and the typeface and colours can be matched to your brand.
Can you use the exact look of our product?
If you have live-action footage of the product, it can be mixed with AI-generated visuals. AI alone cannot correctly reproduce the fine details of a specific product — we make that clear at the assessment stage.
Does AI voice-over sound natural?
For Chinese and English, AI voice-over is already fairly natural on informational content (tutorials, product explanations, walkthroughs of a process).
For narration that needs emotional range, a human voice is still clearly better. We will give you an honest recommendation based on the content, rather than pushing to close a sale.
Are there copyright issues with the generated footage?
We use generation tools that carry a commercial licence. That said, the law on who owns the copyright in AI-generated content is still developing in every country; if your intended use has strict requirements about ownership of rights (registering a trade mark or building a long-term brand identity, for example), please say so when you enquire and we will work out a suitable approach together.
How long does production take?
It depends on the number of sections and how demanding the visuals are; a 30–60 second video generally takes about 7–12 working days. Generating takes repeated attempts, and the actual schedule is set out along with the quote.
Got an idea you would like to try?
Send us a three-point script, or just describe what you have in mind on LINE. We will assess how feasible it is and how demanding the visuals are, then provide a quotation and a schedule.