Manus goes from prompt to a true multi-layer timeline

Manus 2.0 adds Video Editor to its video generation workflow. The agent can research a topic, write the script, produce the various assets, and create an initial cut, before delivering each component on a multi-layer timeline editable manually or using natural language.

Generation No Longer Stops at the First Cut

Changing a music track, shortening a shot by half a second, or replacing an image without regenerating the rest of the video: this is the problem addressed by Video Editor, integrated into Manus Studio on macOS and Windows.

Manus can produce a complete video from an instruction of just a few sentences. The agent conducts web research, builds a script, selects or generates shots, and then assembles music, voiceover, sound effects, subtitles, and animations. It can also analyze a YouTube video, Instagram content, or other references to study their pacing and structure.

The resulting edit is not delivered solely as a finalized video file. Each element remains accessible separately within Video Editor. Every Element Keeps Its Own Track

Shots, images, captions, motion graphics, voiceovers, music, and sound effects appear on separate tracks of the timeline.

An edit can then be made directly with the mouse or requested from Manus in natural language. "Lower the music" or "tighten this part" acts on the same project that is being manipulated manually. When the agent resumes work, it preserves previous modifications rather than entirely rebuilding the sequence.

This continuity addresses a common limitation of entirely chat-driven video workflows, where a local correction can trigger a new generation and alter elements that were already approved. Video Generation, Code, and External Media in a Single Project

Manus combines several production methods. Models like Seedance 2.5 can generate certain shots, while images, music, voices, and sound effects are produced separately.

The agent also has its own computing environment to program animations, graphics, interfaces, or map routes. The post published by Manus describes, for example, using Python to analyze an image or FFmpeg to convert a sequence from 24 to 120 fps before a slow-motion effect.

Teams can also connect video work to the context already present in Manus. For a film dedicated to Cue, the agent leveraged information from meeting notes, Notion, Slack, Figma, and code repositories before proposing several directions and a 16-shot storyboard. According to Manus, the first storyboard proposal arrived after 76 minutes. Two Hours of Raw Footage Transformed into a Timeline

Video Editor also accepts local files stored on the computer. Videos can be dragged into the chat or retrieved from a folder indicated to Manus.

During an internal test detailed by the company, 125 clips representing nearly two hours of raw footage were imported. Manus transcribed, analyzed, and then organized them chronologically. An initial version of 12 minutes and 48 seconds was then reduced to 10 minutes and 50 seconds with 195 cuts.

The final timeline featured 325 subtitles, 112 text effects and stickers, 195 video segments, seven music tracks, and 118 sound effects, each remaining separately editable. From the Web to the Desktop Editor

Creation can begin from Manus on the web or in its application. The user selects the video mode and describes the desired result, then can transfer the project to Video Editor to work on the timeline.

The full editor remains linked to Manus Studio, available on macOS and Windows. From the web version, the "Edit in Video Editor" option redirects to the desktop application to continue editing.