AI video editing glossary: 25 terms explained
Last updated 2026-10-20
AI video editing terms fall into two groups: classic editing words such as B-roll, jump cut and keyframe, and newer AI words such as AI video agent, text-based editing and MCP server. This glossary defines 25 of them in one to three plain sentences each, and says where Woven does or does not do the thing.
Editing basics: footage and cuts
Six terms for shots and cuts.
- A-roll. The main footage that carries your message, such as you talking to camera.
- B-roll. Supporting footage shown while the main audio keeps playing, like the product in use.
- Cutaway. A shot that cuts away from the main action to show something else in the scene, such as an object or a place.
- Jump cut. A cut inside one continuous shot that makes the action leap ahead, such as removing a pause.
- J-cut. A split edit where the next scene's sound starts before its picture does.
- L-cut. The reverse: the previous scene's sound carries on while the picture moves to the next shot.
Editing basics: sound, text and motion
Six terms for sound, text and motion.
- Audio ducking. Lowering music while someone speaks, then bringing it back. Woven has no automatic ducking. Chat can set a duck, but it silences the track fully.
- Lower third. A graphic placed in the lower part of the frame, often showing the name of the person on screen. Woven has no templates; you build one from a text box.
- Keyframe. A marker that stores a setting, such as position or opacity, at one moment, with the software animating between markers. In Woven, press ⌥K.
- Ken Burns effect. A slow pan and zoom over a still image. Woven offers a Slow Zoom loop, a Drift effect and keyframes.
- Hook. The opening seconds of a video, meant to give people a reason to keep watching.
- Open captions. Captions burned into the picture, so viewers cannot turn them off. Closed captions can be hidden or shown by each viewer. Woven's captions are open: drawn into the MP4, with no SRT or VTT export.
AI terms: the tools
Five kinds of software.
- AI video editor. An editing app where AI does part of the work, such as captions or edits requested in chat, and you can still fix the result by hand. Woven is one, for Mac.
- AI video generator. A tool that creates new clips from a prompt or a picture instead of editing your footage. Ask Woven's AI agent in chat and it can make new images, clips and audio too.
- AI video agent. AI that works toward an editing goal by itself: it runs an editor's tools, inspects the output and tells you what it did.
- MCP server. A small program that offers an app's actions to AI apps through the Model Context Protocol, an open standard. Woven's connector is one, for AI apps on your Mac.
- Text to speech. Software that reads text aloud in a synthetic voice. Woven can generate a voiceover from text by chat, but cannot record your voice.
AI terms: ways of working
Eight terms for instructing software and what it makes.
- Text-based editing. Cutting video by editing its transcript. Woven has no such command, but its AI agent can cut on word timings.
- Prompt-based editing. Typing a request instead of dragging clips. In Woven that is chat.
- Vibe editing. An informal name for prompt-based editing where you describe the feel, not the steps. It has no standard definition, so check each change.
- Auto-captioning. Turning speech into timed captions automatically. Woven does this from the Add menu with Generate captions.
- Auto-clipping. Software that picks short clips from a long recording by itself. Woven does not. You name the moments, or ask the agent to propose some from the transcript.
- Text-to-video. Generating a clip from a written prompt alone.
- Image-to-video. Generating a clip from a still image you supply. Woven's video models can start from a picture.
- Reference-to-video. Generating a clip guided by reference images, such as a character or product photo, where the model allows it. It is not Woven's reference analysis, which studies an existing video from a link.
| Term | You give it | Example |
|---|---|---|
| Text-based editing | Edits to a transcript | Delete a sentence and its footage goes |
| Prompt-based editing | A specific request in words | Cut the pauses and add captions |
| Vibe editing | The feel you want, loosely | Make it punchier |
Try the terms in Woven
Most terms above are one chat request or a few clicks in Woven.
Ask in chat
- Cut the long pauses from my talking clip and cover the jumps with product shots.
- Caption my talking clip one word at a time.
- Generate a short clip that starts from this product photo and place it after the intro.
Or do it by hand
- 1Layer B-roll. Drag the cutaway onto the Picture track at the moment it should appear. If it sits behind the main video, press ⇧⌘] to bring it to the front.
- 2Add a keyframe. Select a clip and press ⌥K to add a keyframe at the playhead. Add a second one later and change the clip's size or position there.
- 3Caption speech. Under Add, click Generate captions, and Woven turns the reel's speech into timed captions.
- 4Build a lower third. Choose Add, then Text, set Text style to Subtitle and drag the box near the bottom.
- 5Export. Pick 1080p from the Export menu. The MP4 lands in the workspace's exports folder.
Common questions
Text-based editing means cutting video by editing its transcript. Delete a sentence in the text and the matching footage goes. Woven has no such command, but its AI agent can cut on word timings if you ask.
Vibe editing is an informal name for describing the feel you want, such as faster or warmer, and letting AI make the edit. There is no standard definition, so check each change.
In a J-cut, the next scene's sound starts before its picture appears. In an L-cut, the previous scene's sound keeps playing after the picture moves on. Both are split edits.
You type a request such as cut the pauses and add captions, and the software makes the changes instead of you dragging clips. In Woven that is chat.
A cutaway is a shot that cuts away from the main action to show something else in the scene, such as an object or a place.
Sources
- What is B-roll footage - Adobe, checked October 20, 2026
- Using cutaway shots for film and television - Adobe, checked October 20, 2026
- What is a jump cut, and how to edit one - Adobe, checked October 20, 2026
- L cut and J cut in film - Adobe, checked October 20, 2026
- Auto Duck - Audacity Manual, checked October 20, 2026
- Lower third - Wikipedia, checked October 20, 2026
- What is keyframing - Adobe, checked October 20, 2026
- Ken Burns effect - Wikipedia, checked October 20, 2026
- Captions - W3C Web Accessibility Initiative, checked October 20, 2026
- What is the Model Context Protocol - modelcontextprotocol.io, checked October 20, 2026
Try Woven free for 3 days
Then $25/mo billed yearly ($299/yr) or $39/month. Native Mac app for short-form video.
Paste into Claude Code to install and connect Woven.