Depth map testing with TouchDesigner + Scope:

Did you know that in TouchDesigner you can generate depth maps directly from your own custom 3D geometry? As long as the scene is being rendered, TouchDesigner’s Depth TOP can output a high-quality depth map in real time.

With that in mind, I wanted to bring the same idea into Scope by using VACE to drive video generation from a depth input. By feeding in a depth map rendered from custom cube geometry, I was able to control the structure of the AI video, turning a simple cube scene into a jiggly, depth-aware animation. And honestly, it works surprisingly well.

Controlling VACE input Scale:

One final question: I never was able to get rid of that pesky cube at the front of the video. It doesn't exist in the depth map! Why is it still there?

Prompt: "glossy, cinematic 3D scene. Five translucent jello cubes wobble, slide, and bounce gently across a smooth reflective tabletop. Each cube is made of soft, semi-transparent gelatin, with rounded edges, subtle surface tension, and visible internal light scattering. The cubes jiggle and deform with realistic gelatin physics, compressing on impact, rippling through their interiors, and slowly settling back into perfect cube shapes. Soft studio lighting passes through the cubes, creating glowing color gradients, refractive highlights, and faint caustic light patterns on the surface below. Tiny air bubbles and micro-imperfections are suspended inside the gelatin, enhancing realism. As the cubes collide, they wobble, tremble, and ripple, sending gentle waves through their soft bodies. The camera performs a slow cinematic long take, smoothly tracking the jello cubes while preserving stable silhouettes, consistent cube proportions, and clean geometry. Motion is smooth, satisfying, and physically believable, with subtle motion blur, shallow depth of field, and high-end food-commercial aesthetics. The tone is playful, soothing, colorful, and visually delicious, with no warping, melting artifacts, or visual glitches. "