Each of these is a real operator chain. The same four run as live graphs on the Image home. Place the operators from the Image view of the Voro menu and wire a TOP into the first one.

Depth from one camera

Video Device In, Depth Anything, Composite

One frame in, per-pixel depth out. The depth lands as a TOP next to the image it came from. Mask fog with it, drive a depth of field, displace geometry.

• Depth Anything

Segment and follow

Video Device In, Segment SAM, Composite

Every match gets a mask with an id that holds from frame to frame. Key one instance out of the id map and the effect follows that subject.

• Segment SAM
• Detect RF-DETR when you want boxes instead of masks
• Track Anything to follow a point or a region you picked

Flow into a blur

Video Device In, Optical Flow, Blur

Optical Flow turns motion into a two-channel field. Blur’s directional and warp presets take a field and smear or push the frame along it, so the picture moves with what moved.

• Optical Flow
• Blur

Latent to image

Latent Seed, Latent Transform, StyleGAN, Upscale

No camera. Latent Seed makes a starting point, Latent Transform walks it, and StyleGAN turns the walk into pictures. Upscale gives the generator output-resolution detail on the way out. Latent Compose mixes two latents when you want a blend.

• Latent Seed
• Latent Transform
• Latent Compose
• StyleGAN
• Upscale

When a chain needs your own kernel, script or model, the three authoring operators take the same wires.