# How to make AI B-roll for your video (without paying per credit) > Leave a gap on the timeline where the B-roll should go, select it and describe the shot in the AI video panel. The clip comes back cut to the gap’s length and your video’s shape. Check the price in dollars before each run, try a still picture with a camera move first, and on Pro or Business use your own fal.ai key to pay the model provider directly. - Page: https://cutluma.com/guides/how-to-make-ai-b-roll - By the CutLuma team · Published 2026-10-11 · Updated 2026-10-11 · 5 min read - Try it: https://cutluma.com/editor?guest=1&tool=aigen&mode=t2v (Generate a B-roll shot) ## Step by step 1. **Cut your main video first** — Edit the talking head, voiceover or screen recording until the story works. B-roll is easier to plan once you know exactly where the words land. 2. **Leave gaps where B-roll goes** — Make room on the track above your main video, or open a gap where a shot should cover the speaker. A gap shows its length, so you know how long the shot must be. 3. **Select the gap and describe the shot** — Open the [AI video panel](https://cutluma.com/tools/ai-video-generator) with the gap selected and describe what the viewer should see: “close-up of hands typing on a laptop, soft morning light”. The shot is sized to the gap’s length and the project’s aspect ratio. 4. **Pick a model by price** — Each model shows its price in dollars before you run it. Start with the cheapest one that fits the shot and only move up if it falls short. 5. **Try the cheaper path** — For many shots, a still picture with a slow camera move, a stock clip or a template looks just as good. The panel suggests these before you spend on a video model. 6. **Finish and export** — Trim the shot, add captions, music and your brand kit, and export the whole video, up to 4K on Pro. ## What B-roll is for B-roll is the footage that plays over your main shot: the product while you talk about it, the city while you mention it, a chart while you quote a number. It hides jump cuts, gives the viewer something new to look at every few seconds and shows what you are saying instead of only telling it. Good B-roll is short and specific. Two to five seconds per shot is usually enough, and a shot that matches the exact sentence beats a pretty but generic one. Write the prompt from the line it covers: if the voiceover says “we pack every order by hand”, ask for hands packing a box, not “a warehouse”. ## Generate straight into a timeline gap Generating into a gap saves the fiddly part. The shot comes back at the gap’s length and in your project’s shape (9:16, 1:1, 4:5 or 16:9), so there is nothing to crop or stretch. If the gap is longer than the model can make in one run, generate a shorter shot and fill the rest with a second angle, or [extend the clip](https://cutluma.com/tools/extend-video-with-ai). Have a photo of the real product or place? Use [image to video](https://cutluma.com/tools/image-to-video) instead of text: the model starts from your picture, so the B-roll matches the rest of the video. ## Pick a model by cost, not by hype The models differ in look, length, sound and price. CutLuma shows the price in dollars before every run — the model provider’s price plus a small service fee — so you can compare them on the shot in front of you rather than converting credits in your head. - **Start cheap.** Run the shot on a lower-cost model first. For a two-second cutaway that sits under a voice, it is often good enough. - **Pay for sound only when you need it.** Veo 3.1 Lite makes audio with the picture. If music or a voiceover will cover the shot anyway, a silent model is fine. - **Keep shots short.** Video models charge by length. Generate the seconds you will use, not a long take you will trim. - **Upscale at the end.** Generate at a modest size, then [upscale](https://cutluma.com/tools/ai-video-upscaler) only the shots that make the final cut. ## The still picture and camera move trick A lot of B-roll is a slow push-in or pan across a nice picture, and that does not need a video model at all. Make one picture with [AI images](https://cutluma.com/tools/ai-image-generator) (or use your own photo), drop it into the gap and give it a gentle zoom or pan. It costs a fraction of a generated video clip, or nothing with your own photo, and viewers rarely notice the difference under a voiceover. The [stock library](https://cutluma.com/tools/free-stock-videos-and-music) and the templates are the other free path: real footage of common scenes (offices, cities, nature, food) and animated title or number cards that do the job of B-roll in explainers. The AI video panel offers all three before it runs a video model. ## Bring your own fal.ai key Paid plans include a monthly generation credit, and top-up packs never expire. If you make a lot of B-roll, Pro and Business let you add your own fal.ai API key instead: generation is billed by fal directly, at fal’s price, with no CutLuma markup. You still get the same panel, the price before each run and the shot dropped into your timeline. AI generation is not part of the free plan. Editing, templates, the stock library and 1080p export (with a small watermark) are free; see the [plans](https://cutluma.com/#pricing) for what each one includes. ## FAQ **Can I make AI B-roll for free?** Not with the video models: they cost money to run, so generation is part of the paid plans. The free path is a photo or stock clip with a camera move, which costs nothing. **Which AI video model is cheapest?** Prices change with the model and the length. CutLuma shows each model’s price in dollars before you run it, so compare them on your own shot. **What is a bring-your-own-key plan?** On Pro and Business you can add your own fal.ai API key. fal bills you directly for generation at its price, and CutLuma adds no markup. **How long should a B-roll shot be?** Usually two to five seconds. Cut away for one idea, then come back to the speaker or move to the next shot. **Can I use AI B-roll commercially?** Generally yes, in your own videos, under each model provider’s terms. Don’t generate real people, logos or brands you don’t have the rights to. ## Tools used in this guide - [AI video generator](https://cutluma.com/tools/ai-video-generator) — Describe a shot and generate it with Seedance 2.5, Kling 3 Turbo, Veo 3.1 Lite or Wan 3.0. The price shows in dollars before you run it, and the clip lands on your timeline. - [Image to video](https://cutluma.com/tools/image-to-video) — Turn a still photo into a moving shot: drop in the picture, describe the motion and generate it with Seedance, Kling, Veo or Wan. Or use a free camera move instead. - [AI image generator](https://cutluma.com/tools/ai-image-generator) — Describe a picture and AI makes it in the shape of your video — photo, 3D, illustration, anime, watercolour and more. Drop it straight into your edit. - [Stock library](https://cutluma.com/tools/free-stock-videos-and-music) — Search free-to-use photos, videos, music and sound effects without leaving the editor. One click puts them on your timeline. No credit needed. - [Extend a video](https://cutluma.com/tools/extend-video-with-ai) — A shot ends too soon? Extend it with Veo 3.1 Fast: the AI continues the motion from the last frame so the clip runs longer. Price shown before you run it.