How to Generate YouTube B-Roll With AI

TL;DR

Create a reusable B-roll design system, annotate a script with visual directions, and submit every requested animation in one large Claude prompt. Giving each animation its own page and limiting it to around 20 seconds makes the results easier to download, trim, and place on the editing timeline while the generation process runs unattended.

Transcript

Everything you saw, every animation, every bit of on-screen text, I didn't make any of it. Not by hand anyway. It was generated for me while I was having my dinner. And here's the biggest issue I find with editing. The cuts are not what's using most of your time. It's the B-rolls, the text, the little motion graphics that stop a video from feeling ... Read More

Key Insights

• B-roll is the main editing bottleneck in the creator’s workflow, accounting for about 80% of the edit because finding stock footage, animating text, downloading assets, and repeatedly placing more than 20 visuals on a timeline requires substantial manual effort.
• AI-generated animations can preserve detailed motion without manual keyframing. One example included text bouncing from a YouTube logo, shaking slightly, and absorbing another text element, creating a livelier result than a static slideshow while taking under five minutes to generate.
• A B-roll design system is a set of rules governing how footage should look, feel, and move. Giving Claude these rules before requesting animations helps produce consistent, channel-specific visuals instead of relying on the same generic visual style for every output.
• Claude can help define a visual style through a short question-and-answer exchange. The creator spent about five minutes answering design questions and reported that the resulting system reached roughly 90% of the intended style, with small adjustments needed to complete it.
• Batch generation works by combining every animation request into one large prompt. This removes the need to submit an individual request, wait for its completion, and repeat the process for every visual needed throughout a talking-head video.
• Selective B-roll placement is important because excessive motion can distract viewers just as insufficient visual variety can bore them. The creator recommends reading the script like a director and inserting concise animation notes only beneath lines that genuinely benefit from supporting visuals.
• Separate pages make generated animations easier to use because each asset can be downloaded independently and placed directly into an editor. Without this instruction, Claude may combine the requested animations into one large clip that becomes difficult to divide later.
• The workflow shifts editing time toward higher-value channel work, including writing sharper scripts, studying successful content, and producing more videos. Its main benefit is not merely faster asset creation, but the ability to redirect reclaimed hours toward activities associated with channel growth.
• More videos from this channel:
  • How to Grow on YouTube by Tracking What Works
  • How to Write YouTube Titles That Earn More Clicks
  • How Does YouTube Decide Which Videos to Push?
• More from this channel: Kevin Chee

Questions & Answers

Q: How can AI generate B-roll for a YouTube video?

Start by creating a design system that tells Claude how the visuals should look, feel, and move. Review the script line by line, identify passages that need visual support, and add a short animation direction beneath each selected passage. Submit the complete annotated script as one large prompt, then download the generated pages individually and place them into the editing timeline.

Q: What is a B-roll design system for AI visuals?

A B-roll design system is a reusable set of visual rules that defines the appearance, feeling, and movement of generated footage. It gives Claude a consistent style to follow before any animations are requested. The creator built the system by answering Claude’s questions for about five minutes, reaching roughly 90% of the desired result before applying small finishing adjustments.

Q: How do you batch-generate multiple animations with Claude?

Write every required animation direction into one annotated script and paste the entire collection into Claude as a single prompt. This lets the generation run as one batch instead of requiring repeated prompting and waiting. The creator recommends explicitly requesting a separate page for each animation and limiting each asset to around 20 seconds for easier handling during editing.

Q: Why should each generated animation have its own page?

Each animation should have its own page so it can be downloaded separately and dropped directly into the appropriate place on the editing timeline. If this requirement is omitted, Claude may attempt to stitch all requested animations into one large clip. Dividing that combined clip afterward creates additional cutting work and undermines the time-saving purpose of batch generation.

Q: How long should an AI-generated B-roll clip be?

The creator recommends capping each generated animation at around 20 seconds. That duration provides enough material to trim and adjust within the editor without producing an unnecessarily long asset. The limit should be included directly in the batch prompt alongside the instruction to place every animation on its own page, making the resulting files easier to manage.

Q: Where should B-roll be added in a talking-head script?

Read the script line by line as if directing the finished video, then mark only the passages that need visual support. Add a concise description of the desired animation beneath each chosen line. B-roll should not appear on every line because excessive motion can be as distracting as too little movement. Spacing the visuals allows the presentation to breathe.

Q: How much editing time can AI-generated visuals save?

The creator says B-roll and related visual work previously represented about 80% of the edit, with a five-hour edit later reduced to one hour. A manually produced animation could require 15 to 20 minutes, while one demonstrated asset generated in under five minutes. With around 22 animations per video, the creator estimated recovering more than 400 minutes per project.

Q: What are the limitations of this AI B-roll workflow?

The system should not be expected to produce perfect results on its first attempt. Initial setup requires back-and-forth refinement, similar to teaching a new team member how to work. A large batch prompt also consumed about five hours of the creator’s Claude Pro usage, so the process was started before sleep and allowed to run unattended overnight.

Summary & Key Takeaways

• B-roll, animated text, and motion graphics consumed most of the creator’s editing time because every visual required searching, downloading, importing, and adjusting. Generating those assets with Claude reduced the repetitive work while preserving the small movements that make talking-head footage feel more engaging than a sequence of static slides.

• A reusable B-roll design system defines how generated visuals should look, feel, and move. The creator developed this system by answering Claude’s questions for about five minutes, then making small refinements. These rules help every animation remain visually consistent with the channel instead of resembling an unmodified default template.

• The batch workflow begins by reviewing the script line by line and adding visual directions only where B-roll is useful. All directions are submitted together in one large prompt. Each animation should receive a separate page and an approximately 20-second limit, allowing easy downloading, trimming, and placement in the final timeline.