Higgsfield AI Depth and space
Use the local higgsfield-use-after-effects MCP tools. Read ae_catalog for exact operation arguments before calling ae_do; availability depends on server policy and the installed AE version. Load ae-clean-rig as the shared construction and delivery standard. User instructions and existing project constraints take precedence over recipe defaults.
Choose depth representation from the task. Native 3D layers and cameras are available when the catalog and installed renderer support the needed operation; do not inherit the old bridge's blanket ban on cameras. Layered 2D is sufficient for many UI and illustration scenes. Do not introduce a camera merely to move flat text.
Assign coherent depth planes. Farther planes usually move less under parallax and may lose contrast, saturation and detail; keep these cues consistent. Example saturation reductions of 0/15/30/50 percent and haze increments around 15 percent are starting points, not physically universal values. Match the reference and avoid washing out readable text.
Use small contact shadows to anchor touching surfaces and broader restrained shadows for separation. Keep one light direction and check shadows after parenting or precomposition. Anchor rigid objects before rotations; inspect actual perspective rather than compensating with arbitrary skew.
Expose a depth/parallax strength control whose zero returns exactly to the baseline layout. Test foreground and background extremes for uncovered edges, excessive scale and clipping. Add camera drift or idle wiggle only when requested. Enable motion blur when motion benefits from it, then render at actual settings and check thin UI strokes and text.