AI Car Outfit Transition: A 7-Step Tutorial for Reels and TikTok

You know the outfit is about to change. You still wait for the steering wheel to swing back and reveal it: a white T-shirt becomes a deep-purple velvet dress, and a daylight car interior turns into a purple-lit night scene. A brief sweep of motion blur hides the switch.

To build that effect with AI, split the job into manageable pieces. Create two looks for the same character and two matching car interiors, generate each video with a depth-based motion reference, then join them as the wheel passes in front of the face.

This tutorial follows a 10.6-second example from character design through editing. Open the car outfit transition canvas alongside the article to inspect the original generation history, prompts and node settings.

Before you start: give each reference a clear job

The character image establishes identity and styling; the scene image establishes perspective and lighting; the video reference supplies movement, timing and foreground occlusion. References guide the result, so check each generated clip before moving on.

The recorded workflow uses GPT Image 2 for the character and styling, GPT Image 2.5 Sunburst for the daylight interior, and Seedance 2.5 for the videos. The two generations are set to 5 and 6 seconds, both in vertical 720 × 1280.

The prompts below are reusable adaptations of the example, not newly tested generation prompts. Check the canvas for the original wording and settings.

Step 1: create a recognizable original character

The outfit should change while the person stays recognizable. Start with a clean studio portrait: a plain background and soft frontal light make the eyes, nose, lips and jawline easy to compare later.

This is the reference the subsequent looks will build on. Generate a few options if needed, then zoom in to check the eyelids, mouth edges and proportions; avoid excessively smoothed skin.

Character prompt

Create an original adult female character in a realistic close-up studio portrait. White background, soft frontal light, short black hair with wispy bangs, clear eyes, light makeup and natural lip color. Keep visible pores and fine flyaway hairs. Give her distinct, recognizable facial features without imitating a particular real person. Avoid excessive skin smoothing or exaggerated facial proportions.

Step 2: give the same character two contrasting looks

Clothing is only part of the reveal. Hair, makeup, jewelry and nails help the change read instantly: an understated everyday look first, then a dressed-up evening look.

The everyday look

Use the original portrait as the image reference and change the styling to a white T-shirt and long, straight black hair. Bring the right hand near the collarbone so the nude-pink almond nails are established before they appear close to the camera on the wheel.

Everyday styling prompt

Use the same adult woman from the reference. Preserve her face shape, facial proportions and skin tone. Give her a white crew-neck T-shirt, long straight black hair, small pearl studs and a fine silver bracelet. Use light, natural makeup and pale-pink lips, keeping real skin texture. Raise her right hand gently toward her collarbone to show nude-pink almond-shaped nails. White background, soft lighting, sharp facial detail.

The evening look

Use the everyday portrait as the next reference. Specify what may change and explicitly preserve the spacing of the eyes, nose shape and jawline.

Evening styling prompt

Use the same adult woman from the reference. Keep her face shape, eye spacing, nose, lips and skin tone unchanged. Change only her makeup, hair, clothing and accessories: a deep-purple velvet dress, long black loose waves, long gold earrings and a purple French manicure. Add stronger eyeliner and a berry-purple lip color. Preserve natural skin texture and a clear front-facing portrait composition.

Compare the two portraits side by side. Makeup can change the mood, but a different nose tip or longer chin needs fixing in the reference images before you generate the videos.

Step 3: build the empty daylight interior

Generating the character and car separately makes each easier to adjust. Start with an empty driver's-seat scene that establishes the phone position, the wheel's size in the foreground and the seat perspective.

Place the camera low on the dashboard, close to the steering wheel, looking slightly upward. Let the near edges of the wheel blur naturally while leaving enough space in the middle for the character's face and shoulders.

Daylight interior prompt

Vertical 9:16, realistic phone photography inside an empty ordinary car. The phone is fixed on the dashboard near the steering wheel, with its lens slightly below the wheel's center, looking through it toward the driver's seat at a slight upward angle. The upper and lower wheel edges are close to the lens and softly out of focus. Show the gray leather driver's seat and headrest clearly through the middle. Overcast trees and houses outside, soft natural light. No people, hands or human silhouettes inside.

Step 4: keep the framing and change the lighting

Use the daylight interior as your reference for the night scene. Preserve the 9:16 frame, camera height, wheel size and seat position; change the materials and lighting to black and purple to complement the dress.

Night interior prompt

Use the reference interior as the composition guide. Keep the 9:16 frame, low camera position, foreground wheel size and driver's-seat position unchanged. Change the interior to black and deep purple, with purple sport seats and purple ambient lighting along the doors and roof. Make it dark outside. Preserve realistic leather texture and natural lens blur. Empty car, no change in camera angle.

When generating the night video, keep a soft, neutral key light on the face. Let the purple light color the cabin, hair edges and shoulders; flooding the face with purple can hide the makeup you just designed.

Step 5: generate the first clip with a depth reference

The first clip has four beats: look at the camera, raise both hands, grip the wheel, then turn it quickly across the face. Text alone is a difficult way to coordinate exactly when the hands move and how much of the face they cover.

You can convert footage you own or have permission to use with the AI Depth Video tool. In this example, the first reference combines about five seconds of depth video with a skeleton overlay: grayscale describes relative distance and occlusion, while the colored lines provide pose cues.

The tool produces depth video; adding a skeleton overlay is a separate step. The canvas does not document which tool created that overlay, so you can start with the supplied reference below.

Attach the everyday portrait, daylight interior and first motion reference to the video node. Set Seedance 2.5 to 9:16, 720p and 5 seconds.

Before-change video prompt

Take the character's appearance from the everyday styling image and the cabin and camera position from the daylight interior image. Use the motion reference only for pose, action order, timing and spatial relationships. Do not reproduce its grayscale appearance or skeleton lines.

Realistic 9:16 car interior video, fixed camera, one continuous shot, parked vehicle. The woman wears the white T-shirt; preserve her face, earrings and pale nails from the reference. Begin with her looking at the camera, then raise each hand onto the upper edge of the steering wheel following the motion reference. Near the end, turn the wheel quickly so the near hand and wheel pass in front of her face, creating foreground occlusion and motion blur for an edit point. No zoom, camera movement or extra effects.

Use this as a target timing guide, then judge the actual result:

• 0–1.5 seconds: look toward the camera with a relaxed expression.
• 1.5–2.5 seconds: raise the hands onto the top of the wheel.
• 2.5–4.3 seconds: hold the wheel with subtle head and eye movements.
• 4.3–5.0 seconds: turn quickly so the hand and spokes cross the face with visible motion blur.

Step 6: start the second clip already in motion

Swap in the purple evening portrait, night interior and second depth-only reference, which runs about 5.58 seconds. Set the generated clip to 6 seconds.

The beginning needs to continue the previous movement. Ask for the wheel to be turning already, so the new look appears as it clears the face; a pause followed by another hand raise will break the transition.

After-change video prompt

Take the character's appearance from the purple styling reference and the cabin and lighting from the purple interior reference. Use the motion video only for timing and spatial relationships, not its grayscale appearance.

Realistic 9:16 car interior video, fixed camera. Begin with the steering wheel already turning quickly. A spoke passes across the face before the wheel straightens, revealing the same woman in her new look. Follow the reference as her hands rest on the wheel, briefly display the purple French manicure, slide outward and return in rhythm. Keep the purple dress, wavy hair and long earrings consistent. Purple ambient light illuminates the cabin and her outline, with a soft neutral key light on her face.

Loop the first clip's ending against the second clip's beginning. Check that the wheel's movement flows across the cut and the face stays a similar size; small offsets may be fixable in editing, but a different camera angle or action needs a revised reference or regeneration.

Step 7: find the cut, then check the details

Place both clips on a 9:16 editing timeline. The canvas does not include the editing project, so the following is a suggested method; choose the precise cut and adjustments from your own footage.

1. Find the outgoing frame. Step through the end of the first clip and look for strong foreground coverage and motion blur. The face does not have to disappear completely if the movement carries across naturally.
2. Match the incoming frame. Find a frame near the start of the second clip with a compatible movement and a similar foreground shape. Trim any pause before it.
3. Try a hard cut first. Check the action without a dissolve or flash. A long dissolve can show both faces and wheels at once; adjust the cut before reaching for another effect.
4. Align the framing and sound. Use a small scale or position adjustment if the face sizes differ slightly, checking the resulting crop. If music carries the edit, mute or lower the generated audio and place the clear reveal on the beat you want to emphasize.
5. Export and inspect. Start with the source settings of 720 × 1280 at 24 fps. Check the quick turn frame by frame for warping or ghosting.

Save the character references, matched interiors and depth clips together once the workflow works. For the next outfit, reuse the proven camera position and movement, change the styling and scene colors, then adjust the edit to the new results.