Explainer presenter prompts for tutorials, courses and how-tos

What makes explainer prompts work

Plan the frame around the graphics. If a chart will appear on the right, place the presenter on the left third and ask for a plain wall on the right. Empty space is easier to fill in editing than a busy background is to clean up.

Use a neutral, quiet set: a light wall, a single plant, a desk. The more detailed the background, the more it changes between lessons. If the set matters, add a photo of it as RefVid's scene photo, which sets the location and look.

Keep each spoken line to one idea and write it in quotation marks so the model treats it as speech. With sound on, the model generates the voice along with the picture, and short lines keep the voice and mouth in step. RefVid has no voice cloning, so the voice can differ between clips; for a course with one consistent voice, record the narration yourself and keep sound off.

The examples call the presenters Lena and Marco; use the name you saved yours under. Clip length and aspect ratio are settings, shown here in brackets.

Lesson openings

Prompt (8 s, 16:9): Lena stands in front of a plain light-grey wall, positioned on the left third of the frame. She smiles, gives a small nod and says: "In this lesson, you will set up your first project in under five minutes." Medium shot, static camera at eye level, soft even light from a large window on the right.

Why it works: the empty right side leaves room for a title card. Keep the same placement for every lesson so the course looks consistent.

Prompt (6 s, 16:9): Marco sits on the edge of a desk with a laptop beside him, looks up from the screen to the camera and says: "Let's start with the one setting most people miss." Medium-wide shot, warm desk lamp and cool window light, bookshelf softly blurred.

Why it works: looking up from the laptop motivates the cut to a screen recording. If the laptop screen draws attention, angle it away from camera.

Pointing and gesturing toward graphics

Prompt (5 s, 16:9): Lena stands on the left of the frame and points with an open hand toward the empty wall on the right, glancing at it and then back to the camera. Static medium shot, plain off-white wall, even light.

Why it works: an open-hand point toward empty space is the standard setup for a pop-up graphic. Add the graphic in editing exactly where the hand points.

Prompt (6 s, 16:9): Marco holds both hands apart at chest height as if showing the size of something, then brings them together while speaking calmly. Medium close-up, light background, soft key light.

Why it works: simple size and comparison gestures read well and pair with almost any diagram. Avoid detailed counting on fingers; extra fingers still appear.

Prompt (5 s, 16:9): Lena stands beside a large blank whiteboard and taps it twice with a marker, then turns to the camera with a slight smile. Static medium-wide shot, bright classroom light.

Why it works: a blank board avoids invented writing; you add the diagram afterward. Writing on the board in the clip usually produces nonsense text.

Walk-and-talk segments

Prompt (7 s, 16:9): Lena walks slowly down a bright office corridor toward the camera, speaking naturally, one hand gesturing. The camera moves backward in front of her at the same pace. Daylight from glass walls, colleagues out of focus.

Why it works: walk-and-talks break the rhythm between screen recordings. Keep the pace slow so the face stays sharp.

Prompt (6 s, 16:9): Marco stands outdoors next to a small solar panel installation, turns toward it and gestures along its edge, then looks back to the camera. Medium shot, slight handheld feel, clear midday light.

Why it works: on-location shots make a lesson feel real. The object should be large and simple; small detailed equipment is easily misdrawn.

Planning a course

In RefVid, clips run 3 to 15 seconds on Kling 3.0 Omni, and one clip can hold up to three shots, so an opening line and a pointing shot can come from one generation. The length you pick is split evenly across the shots.

Veo 3.1, for comparison, generates 4, 6 or 8 seconds, and its reference clips are always 8 seconds, according to Google's documentation as of October 2026. With any model, keep the presenter's references and the set fixed for every lesson; the only things that change should be the line and the gesture.

RefVid keeps the presenter as a saved character, so a course of thirty lessons uses the same references in every clip.