Night Street Race: Two Cars VFX Prompt with Depth Map Refs

Output ONE single full-frame 16:9 shot only - never reproduce the stacked two-panel layout of the reference video; no split screen, no duplicated frames, no panels of any kind. Creatives: [video1] is a VFX control reference for one shot, repeated as a loop - TOP half is the live-action night plate with BOTH race cars masked as flat color silhouettes: the NEAR car is solid RED, the FAR car is solid GREEN; BOTTOM half is the synchronized depth map of the same frames (warm = near, cool = far) - it defines both cars' 3D form and relative depth, honor it but never show its colors in the output. [image1] shows the car for the RED silhouette. [image2] shows the car for the GREEN silhouette. Night street race, two cars neck-and-neck sprinting left through a blue-lit city street, backgrounds motion-blurred, cinematic 2000s action-film look. Recreate the TOP panel's background, camera motion and timing exactly. Replace the RED silhouette with a metallic gold 1990s sports coupe with a tall rear wing and white-and-red flame graphics - strictly maintain the consistency of the body shape, flame livery, wheels and gold paint of [image 1]. Replace the GREEN silhouette with a silver late-1990s performance coupe with angular blue vinyl graphics and dark multi-spoke wheels - strictly maintain the consistency of the silver paint, blue graphic pattern, front bumper and wheel design of [image 2]. Each car matches its silhouette's position, scale, overlap and motion frame-for-frame - the gold car partially occludes the silver car exactly where the red silhouette overlaps the green one. Headlights on, blue underglow beneath the silver car, wheels blurred at race speed. 0–4s: both cars charge left in tight formation, gold coupe nearest camera, silver coupe half-hidden beyond it. The camera holds its plate motion. Hold. No red or green silhouettes visible in the output, no depth-map or heat-map colors anywhere in frame. No subtitles, no BGM, no on-screen text. Prohibit objects appearing out of thin air, prohibit either car changing shape, color, or livery mid-shot, prohibit extra cars entering frame, no mixing of the two cars' designs, accurate quantity: exactly two cars. The entire video should have no sound.