Prompt Engineering

Basics

Higgsfield models reward concrete, sensory prompts.

• Subject + setting + style: "a red fox curled in a snowy pine forest, golden hour, cinematic"
• Camera: lens (35mm, 85mm), angle (low, overhead), motion (dolly in, tracking shot)
• Lighting: rim light, neon glow, moody backlight
• Style/medium: oil painting, watercolor, photograph, anime, 3D render

Keep it under ~200 tokens. Models distort with very long prompts.

Image-to-image

When passing --image, the prompt should describe what changes, not redescribe the input.

Bad: "a man with brown hair in a leather jacket holding coffee, made into anime" Good: "transform into anime style, vibrant colors, soft cel shading"

Image-to-video

--start-image anchors the first frame. Prompt describes motion.

• Verbs: zooms in, dollies left, sweeping pan, slow push, fast whip
• Subject motion: "the dancer spins", "smoke rises slowly"
• Don't redescribe the static frame — model already has it.

Negative phrasing

Most models don't expose a negative_prompt. Phrase positively: - Instead of "no blur" → "tack sharp" - Instead of "no people" → "uninhabited landscape"

Aspect ratio guidance

• 16:9 — landscape, cinematic
• 9:16 — vertical, social
• 1:1 — square, profile / icon
• 4:3, 3:4, 21:9 — model-dependent, check higgsfield model get <jst>

Safety

Models reject prompts with nsfw or ip_detected terminal status. Avoid: - Real public figures - Sexual content - Trademarks / branded characters