higgsfield-image-shots
Installation
SKILL.md
Higgsfield Cinematic Image Prompting
QUICK FACTS
Generated-checked block (scripts/build_index.py verifies anchors). Read the linked sections for full context — these lines are routing aids, not the rules themselves.
- Scope: still-image composition only;
[img 1]is the Soul ID / uploaded reference token — replace with a character description or @ Element when no reference is used → - Base pattern: [Shot keyword] of [img 1 or character description] + [pose/action] + [environment detail] + [lighting/atmosphere] →
- 10 framing sizes, EWS → Macro: EWS tiny-in-landscape, Wide full body, Full head-to-toe, MLS knees up, Cowboy mid-thigh up, Medium waist up, MCU chest up, CU face, ECU single detail, Macro texture →
- Framing keywords: "EWS, Vast, Landscape" / "Wide Shot, Full Body" / "Cowboy Shot" / "MCU" / "Close-Up" / "ECU" / "Macro" →
- 10 angles: Eye-Level, Low Angle, High Angle, Overhead, Worm's Eye, Dutch/Canted, OTS, POV, Selfie, Ground Level →
- Angle = emotional read: Low = power, High = vulnerability, Dutch = unease/chaos, Worm's Eye = extreme power/surreal →
- Multi-word angle keywords: "POV, First Person Perspective" / "Selfie Angle, Arm's Length" / "Ground Level, Surface View" →
- OTS = looking past one person's shoulder at another — essential for dialogue →
- Movement keywords create the FEELING of motion in a single frame — motion blur, implied movement, frozen mid-movement →
- Static/Basic movements: Static, Pan, Tilt, Zoom In / Zoom Out, Pedestal Up/Down →
- Advanced Physical movements: Dolly In, Dolly Out, Truck Left/Right, Orbit, Crane →
- Cinematic & AI movements: Dolly Zoom (Vertigo), Crash Zoom, FPV, Bullet Time, Handheld Follow, Camera Roll, Rack Focus, Pull Back Reveal, Fly-Through →
- Strongest prompts layer all three: [Distance/Size] + [Angle] + [Movement hint] + Subject + Environment + Lighting →
- Full formula closes with [Lighting — time of day, source, quality] then [Style — cinematic, film stock, color grade] →