Non-Linear Editing of Text-Based Screencasts
summary
Presents a non-linear editing algorithm and a working prototype that edits text-based screencasts by transforming later operations and asking users to resolve conflicts via a three-way merge style Ambiguity Resolver. Uses operational transformation (OT) internally and is open-sourced on GitHub. Six participants complete all tasks; roughly half of edits create ambiguities, each ambiguity needs about 1.45 decisions and takes ~19 seconds to resolve, and users report the resolver is hard to interpret without more context.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:31Text-based screencasts benefits
- 2:59Editing challenges explained
- 5:37Prototype demonstration
- 6:01Algorithm overview
- 8:15Ambiguity detection process
- 11:07Interface walkthrough
- 12:57User study results
- 16:43Future improvements
- 17:23Q&A session
related talks
18mRestructuring Structure Editing
09mMicro-Versioning Tool to Support Experimentation in Exploratory Programming
21mQuickCut: An Interactive Tool for Editing Narrated Video
28mDocument-oriented programming based on edit history
39mCursorless: A spoken language for editing code
14mRescribe: Authoring and Automatically Editing Audio Descriptions
04mLT Erich Studer
27mDefault Disambiguation in Online Parsers