Travel vlog prompts for a character who looks the same everywhere

What makes travel prompts work

Decide the travel outfit once and put it in the references: a jacket, a bag, a hat if they wear one. Viewers recognize a vlogger by silhouette as much as face.

Changing clothes every clip is fine in real life but doubles the drift risk in AI video. If you want a new outfit, save a new version of the character for that leg of the trip.

Use vlog camera language: selfie stick at arm's length, walking toward the lens, over-the-shoulder views of a street, a wide shot with the traveler small in the frame. These framings tell the model the genre without long descriptions.

Describe places by what is visible, not by name alone. "A narrow cobbled alley with laundry lines and blue shutters" gives more control than a city name, and avoids depending on how well the model knows a specific landmark.

Length, shape and sound are in brackets before each prompt; they are settings, not prompt text. The prompts say "the traveler", and RefVid attaches the saved character.

Arrival and walking shots

Prompt (7 s, 16:9): The traveler steps off a train onto a busy platform, backpack on one shoulder, and looks up at the departure board. The camera follows from behind, then swings around to their face as they smile. Morning light through a glass roof, people moving in soft focus.

Why it works: arrivals are natural openers, and the swing ends on a front view the references cover. If the backpack changes, include a photo of it on its own.

Prompt (6 s, 9:16): Selfie-stick framing of the traveler walking slowly through a narrow cobbled alley with laundry lines and blue shutters. They glance up at the buildings and back to the lens. Handheld sway, warm late-afternoon sun bouncing off pale walls.

Why it works: selfie framing keeps the face large and stable throughout. Keep the walking pace slow; fast walking adds bounce that blurs features.

Prompt (8 s, 16:9): Wide shot of the traveler walking barefoot along the edge of a quiet beach at sunrise, shoes in one hand. Small in the frame, waves washing over their feet. The camera is static on the sand. Pale pink sky, soft wind.

Why it works: wide shots are forgiving because the face is small. Use them to vary pacing between close talking shots.

Talking to camera

Prompt (8 s, 9:16, sound on): The traveler stands at the rail of a small ferry, wind moving their hair, holding the camera at arm's length. They say: "Two hours from the city and the water is already this clear." Then they point the camera briefly toward the island ahead. Bright midday sun, sea spray. Sound: engine hum, wind.

Why it works: a single line with a gesture to the view is the core vlog beat. Wind on the hair adds motion without moving the face. In RefVid the line in quotes is said as written; English and Chinese speech sound best.

Prompt (7 s, 9:16): The traveler sits at a tiny café table on a sidewalk, a small coffee in front of them, talking quietly to a phone propped against the sugar jar. They laugh, look down at the street, then back to the lens. Soft shade from an awning, street sounds in the background.

Why it works: a static, propped camera keeps identity stable. With no line written, the clip works as a cutaway under narration you record yourself.

Scenic B-roll with the character

Prompt (8 s, 16:9): Aerial-style view slowly rising from behind the traveler as they stand at a mountain viewpoint, arms folded on a wooden rail, a valley of terraced fields below. Golden hour, long shadows, light mist in the distance.

Why it works: a back view with a reveal of the landscape needs only the outfit and hair to match. Include a back-view photo in the references.

Prompt (6 s, 16:9): The traveler browses a night market stall, picking up a paper cone of roasted nuts and smiling at the vendor. Over-the-shoulder camera, colorful hanging lamps, steam from nearby stalls.

Why it works: market scenes are busy, so the over-the-shoulder view keeps attention on the character. If the vendor's face pulls focus, keep them partly out of frame.

Planning a trip as a series

In RefVid one clip can hold up to 3 shots, with the clip's length split evenly between them: a 9-second clip could be shot 1, a wide of the traveler on the platform; shot 2, a selfie close-up as they leave the station; shot 3, the street outside. "Next shot" starts a new clip from the last frame of the previous one, so a walk can continue across clips.

On Veo 3.1, as of October 2026, according to Google's Veo docs, clips with reference images are 8 seconds, so plan one beat per clip. Either way, cut the clips together in your editor and keep the reference set identical for every clip of the trip.

Every video of a character is listed on its page in the RefVid studio, so a ten-city trip stays in one place and drift between cities is easy to spot. Prompts can be written in any language; RefVid translates them for the model.