Guide > Chat: requests and actions Skills

Video Understanding: picture and sound cues

Inspect selected video in REAPER, answer picture questions, and find actions and sound cues. Write the results as Markers, or explicitly request library-based sound design and import.

01

Prepare video and local models

Connect REAPER Bridge and select the video Item to inspect. In Tools, check that the required local vision models are installed; follow the app’s preparation prompts if they are missing. Sound design also requires scanned, enabled local sound libraries.

02

Inspect picture and find sound positions

  1. 01Select a video Item in REAPER, then choose Video Understanding in the Chat Skill menu.
  2. 02Specify what to inspect: “Find the door closing and object impacts, and add descriptive Markers.”
  3. 03Check the answers and timestamps. To understand the picture only, ask a question without requesting a project change.
  4. 04For sound, explicitly add “find local assets for these cues, compose them, and import into REAPER.” Play the video to check synchronization, duration, and level.
03

Three common requests

Understand picture
What happens in this video? At what moment does the door fully close?
Mark sound cues
Find footsteps, impacts, and lock movements, and write descriptive Markers.
Design sound for video
Compose effects for the identified actions using local libraries and import into REAPER.