Could Writing a Script Become the New Way to Edit AI Films?

Imagine writing a story and watching it become a film, complete with characters, locations, dialogue and music, without needing to master complicated video editing software.

That is the idea behind Doki, an experimental AI filmmaking tool developed by Adobe Research that allows creators to generate and edit videos directly within a text document.

Unlike conventional AI video generators that produce individual clips from prompts, Doki brings the entire creative process into one workspace. Writers can develop their stories, establish characters, generate scenes and make adjustments without constantly switching between different applications.

The technology was presented at the ACM Symposium on User Interface Software and Technology (UIST), held in Detroit from 2 to 5 November 2026, where its research paper received a Best Paper Honourable Mention.

Adobe research scientist Bruce Liu described the concept simply: "You write your story the way you'd write a document, and it becomes a film."

How Does Adobe's Doki Turn Writing Into Video?

Most AI video generators create short clips from written instructions, leaving creators to assemble the results using separate editing programmes.

Doki takes a different approach by making the written document the centre of the entire video project.

Users begin by defining their characters, locations, objects and visual styles using tags. These details can then be reused throughout the story, helping characters and settings maintain a consistent appearance across different scenes.

For example, a writer creating an animated adventure could describe their main character once before developing the story through a series of paragraphs. Doki uses those descriptions to generate individual shots without requiring the same character details to be entered repeatedly.

Small video previews appear directly within the document, allowing creators to review scenes, regenerate clips or make adjustments as they work.

Dialogue and music instructions can also be incorporated into the text, while an AI assistant helps develop ideas, rewrite sections and restructure stories.

The experimental version used Google's Veo 3 Fast as its default video generation model, with still-image previews allowing users to check how scenes might look before generating the more demanding video clips.

What Happened When Filmmakers Tested Doki?

Adobe researchers Bruce Liu, Mira Dontcheva and Ding Li carried out a week-long study involving ten participants with different levels of filmmaking experience.

The group included professional filmmakers and animators alongside people who created videos only occasionally. After an introduction to the software, participants spent five days independently producing videos using Doki.

Together, they created 46 videos, most lasting between 30 and 90 seconds. Their projects ranged from animated fairy tales and cooking demonstrations to advertisements, music videos and experimental films.

The results were encouraging. Using a standard questionnaire designed to measure how easy software is to use, Doki achieved an average System Usability Scale score of 81.2 out of 100, placing it within the excellent range for usability.

Eight participants reported that the system helped them move more quickly from an initial idea to a usable video draft, while four said the document-based approach made it easier to understand and organise their stories.

One particularly popular feature was the ability to maintain consistent characters without repeatedly writing lengthy descriptions. For independent creators, that could make developing longer AI-generated videos considerably more straightforward.

Could AI Make Filmmaking More Accessible?

Doki offers an interesting possibility for writers, independent filmmakers and content creators who have creative ideas but limited experience with traditional editing software.

Instead of learning complex editing timelines before experimenting with a story, creators could begin with something familiar: writing.

The technology could eventually support short films, animations, promotional videos and visual storytelling projects, allowing people to spend more time developing their ideas.

The research also identified areas requiring improvement. Nine participants encountered occasions when the generated videos did not accurately match their descriptions, sometimes requiring several attempts to achieve the intended result.

Some also experienced visual errors and difficulties controlling precise camera compositions, audio extending across multiple scenes and transitions between shots.

The study involved only ten people, including four Adobe employees, and did not directly compare Doki with conventional editing software. Its findings therefore demonstrate early promise rather than establishing that the tool is faster or more effective than existing alternatives.

Nevertheless, the positive usability results suggest that writing-based video creation could become an appealing option alongside traditional filmmaking methods.

When Will Adobe Doki Be Available?

Doki remains an experimental research prototype, and Adobe has not announced a public release date or confirmed whether it will become a commercial product.

The project forms part of the company's wider exploration of how generative AI can support creative professionals and simplify production workflows. Adobe has already introduced a growing range of AI-powered features through its existing software, including tools designed to make video editing more accessible.

SoapLandTV previously reported on how Adobe Firefly gained the ability to remove video backgrounds without frame-by-frame editing.