How to Make an AI Runway Walk Video in Idoli

A runway clip needs more than a glamorous model. The feet need to meet the floor, the camera needs to leave space for the walk, and the ending needs enough time for a balanced pose. Defining those relationships gives the video a clear sequence to follow.

This tutorial uses Idoli’s AI walking effect template to create an original adult model from text. You can optionally replace the appearance description with a saved character if you want a recurring model.

1. Start with the Runway Template

Open the AI walking effect runway template for the complete prompt. Its default model is a beautiful 26-year-old white European woman with fair skin, dark-brown waves, and a tall, naturally curvy figure. The look uses black satin and lace lingerie with opaque cups and opaque front and back panels, paired with black heels.

Soft overhead spotlights, a dark audience area, and occasional blurred photographic flashes establish the fashion-show setting. The default prompt generates the model from text; creating or uploading a character first is optional.

2. Frame the Whole Walk

The template targets ten seconds in landscape 16:9. Match those settings in the interface if your selected model provides them.

Keep the model visible from head to shoes, with space below her feet. Position the camera at eye level, facing her, and have it track backward along the runway centerline at her walking speed. This keeps the model’s size relatively consistent while showing the steps.

Avoid combining a backward tracking shot with instructions to zoom in on the face or crop at the waist. Those changes would remove the footwork this example is intended to show.

3. Divide the Action into Three Phases

Use these time blocks together with the full template’s appearance, clothing, scene, and camera descriptions:

0:00 - 0:03 | Begin: The model walks toward the camera along the runway centerline, shoulders relaxed, arms swinging naturally. Show controlled heel-to-toe steps and smooth weight transfer.

0:03 - 0:07 | Continue: Keep an even pace, subtle shoulder and hip counter-rotation, and light hair movement. Maintain clear shoe contact with the floor and steady full-body framing.

0:07 - 0:10 | Present: She shortens the final two steps and stops at the presentation mark. The camera slows with her. She rests one hand lightly on her hip and holds the final pose as her hair settles.

These are action phases in one continuous shot, not three cuts. The ending is a held pose rather than a seamless return to the first frame.

4. Check Footwork, Camera, and Clothing

Watch the entire clip before refining the styling. Check whether the heels slide, the legs cross unnaturally, or the knees bend abruptly. Confirm that the head and shoes stay inside the frame and that the camera slows as the model stops.

Inspect the satin panels, lace trim, and straps for abrupt changes. The garment should remain securely fitted, with its opaque panels intact. Generated motion and clothing details may vary, so check the result rather than assuming the prompt guarantees every detail.

If the final pose feels rushed, change the timing in the full prompt:

Begin slowing at six seconds, stop by eight seconds, and hold a balanced final pose until the end. Keep the camera steady during the hold.

When using that variation, replace the original timing rather than leaving both versions in the prompt.

5. Adjust the Look or Reuse a Saved Model

For a color variation, replace black satin with deep burgundy and keep the garment construction consistent. For a brighter presentation, lighten the runway background while retaining enough contrast to separate the model from the audience.

Video generation in Idoli does not currently have conversation memory. Submit the complete revised prompt for each run, including the appearance, outfit, runway, camera, and action. A short follow-up such as “make the lights brighter” is not a substitute for those instructions.

For a series featuring the same model, create and save a character, then select and reference it in video generation. Replace the appearance description with the actual reference, using it once. Keep the wardrobe and motion requirements and ensure the interface selection matches the reference. A saved character does not automatically carry clothing from a previous generation.

Use the full runway walking prompt as your starting point, then change one aspect at a time.