Video Prompting Guide

Best practices and techniques for writing effective AI video generation prompts. Covers: Veo, Seedance, Wan, Grok, Kling, Runway, Pika, Sora prompting strate...

Security Scan

The skill is an instruction-only prompt-writing guide that is internally consistent with its stated purpose, but it references running a remote installer (curl | sh) and external CLI tooling without declaring credentials—verify the installer and exercise caution before executing remote scripts.

Purpose & Capability

Name/description (video prompt best practices) match the content: shot types, camera movement, style keywords and many example prompts for video generation services. Requested capabilities (none) and examples of CLI usage are coherent with a guide.

Instruction Scope

SKILL.md stays within the domain of prompting guidance and example commands for inference.sh; it does not instruct reading arbitrary files or harvesting unrelated data. However it explicitly recommends running a remote install script (curl -fsSL https://cli.inference.sh | sh) and calling infsh login, which introduces runtime behavior (downloading/executing remote code, authenticating) not strictly necessary to the static textual guidance.

Install Mechanism

There is no packaged install spec in the registry (instruction-only), but the document encourages piping a script from cli.inference.sh. That pattern (curl | sh) is convenient but higher-risk than documented package installs; the guide claims checksum verification is available on dist.inference.sh, but the skill does not verify or bundle those artifacts itself.

Credentials

The skill declares no required environment variables, credentials, or config paths. It does show an example 'infsh login' which implies credentials might be needed at runtime, but those are not requested by the skill and are appropriate for the external CLI usage shown.

Persistence & Privilege

No always:true, no install or code files, and no requests to modify other skills or system-wide configuration. The skill is user-invocable and can be used interactively; autonomous invocation is allowed by platform default but not excessive here.

Assessment

This is primarily a helpful, instruction-only guide for writing video-generation prompts and the examples are consistent with that purpose. Before following the install examples, do not run curl | sh blindly: inspect the installer script (or use the provided checksums) and prefer downloading verified releases or manual install steps. Be aware that running infsh login or the CLI will contact external services and may request credentials or tokens—only provide credentials you expect to give to that service. If you want to avoid executing remote code, you can still use the guide's prompt templates and examples without installing the CLI. If you need higher assurance, ask the publisher for the install script contents or use the manual checksum URL referenced in the doc to validate any downloaded binary.

Best practices for writing effective AI video generation prompts via inference.sh.

Quick Start

curl -fsSL https://cli.inference.sh | sh && infsh login

# Well-structured video prompt
infsh app run google/veo-3-1-fast --input '{
  "prompt": "Cinematic tracking shot of a red sports car driving through Tokyo at night, neon lights reflecting on wet streets, rain falling, 4K, shallow depth of field"
}'

Install note: The install script only detects your OS/architecture, downloads the matching binary from dist.inference.sh, and verifies its SHA-256 checksum. No elevated permissions or background processes. Manual install & verification available.

Prompt Structure Formula

[Shot Type] + [Subject] + [Action] + [Setting] + [Lighting] + [Style] + [Technical]

Example Breakdown

"Slow motion close-up of coffee being poured into a white ceramic cup,
steam rising, morning sunlight streaming through window, warm color grading,
cinematic, 4K, shallow depth of field"

• Shot Type: Slow motion close-up
• Subject: Coffee
• Action: Being poured
• Setting: White ceramic cup, window
• Lighting: Morning sunlight
• Style: Warm color grading, cinematic
• Technical: 4K, shallow depth of field

Shot Types

Shot Type	Description	Use For
Wide shot	Shows entire scene	Establishing location
Medium shot	Waist-up framing	Conversations, actions
Close-up	Face or detail	Emotion, product detail
Extreme close-up	Single feature	Drama, texture
Aerial shot	Bird's eye view	Landscapes, scale
Low angle	Camera looking up	Power, grandeur
High angle	Camera looking down	Vulnerability
Dutch angle	Tilted camera	Unease, tension
POV shot	First person view	Immersion

Camera Movements

Movement	Description	Effect
Tracking shot	Camera follows subject	Dynamic, engaging
Dolly in/out	Camera moves toward/away	Focus, reveal
Pan	Horizontal rotation	Survey scene
Tilt	Vertical rotation	Reveal height
Crane shot	Vertical + horizontal	Dramatic reveal
Handheld	Slight shake	Realism, urgency
Steadicam	Smooth following	Professional, cinematic
Zoom	Lens zoom in/out	Quick focus change
Static	No movement	Contemplation, stability

Lighting Keywords

Keyword	Effect
Golden hour	Warm, soft, romantic
Blue hour	Cool, moody, twilight
High key	Bright, minimal shadows
Low key	Dark, dramatic shadows
Rim lighting	Subject outlined with light
Backlit	Light from behind subject
Soft lighting	Gentle, flattering
Hard lighting	Sharp shadows, contrast
Neon	Colorful, urban, cyberpunk
Natural lighting	Realistic, documentary

Style Keywords

Cinematic Styles

cinematic, film grain, anamorphic lens, letterbox,
shallow depth of field, bokeh, 35mm film,
color grading, theatrical

Visual Aesthetics

minimalist, maximalist, vintage, retro, futuristic,
cyberpunk, steampunk, noir, pastel, vibrant,
muted colors, high contrast, desaturated

Quality Keywords

4K, 8K, high resolution, photorealistic,
hyperrealistic, ultra detailed, professional,
broadcast quality, HDR

Prompt Examples by Use Case

Product Demo

infsh app run google/veo-3-1-fast --input '{
  "prompt": "Smooth tracking shot around a sleek smartphone on a white pedestal, soft studio lighting, product photography style, reflections on surface, 4K, shallow depth of field"
}'

Nature Documentary

infsh app run google/veo-3-1 --input '{
  "prompt": "Slow motion extreme close-up of a hummingbird hovering at a red flower, wings in motion blur, shallow depth of field, golden hour lighting, National Geographic style"
}'

Urban Lifestyle

infsh app run google/veo-3 --input '{
  "prompt": "Tracking shot following a cyclist through busy city streets, morning rush hour, natural lighting, handheld camera feel, documentary style, authentic and candid"
}'

Food Content

infsh app run bytedance/seedance-1-5-pro --input '{
  "prompt": "Close-up of chocolate sauce being drizzled over ice cream, slow motion, steam rising, soft lighting, food photography style, appetizing, commercial quality"
}'

Tech/Futuristic

infsh app run xai/grok-imagine-video --input '{
  "prompt": "Futuristic control room with holographic displays, camera slowly pans across the space, blue and cyan lighting, sci-fi atmosphere, Blade Runner aesthetic, 4K",
  "duration": 5
}'

Common Mistakes to Avoid

Mistake	Problem	Better Approach
Too vague	"A nice video"	Specify shot, subject, style
Too complex	Multiple scenes	One scene per prompt
No motion	Static description	Include camera movement or action
Conflicting styles	"Minimalist maximalist"	Choose one aesthetic
No lighting	Undefined mood	Specify lighting conditions

Model-Specific Tips

Google Veo

• Excels at realistic, cinematic content
• Supports audio generation (Veo 3+)
• Best with detailed, professional prompts
• Frame interpolation available in 3.1

Seedance

• Strong at dance and human motion
• First-frame control available
• Good for consistent character motion
• Works well with reference images

Wan 2.5

• Best for image-to-video
• Animates still images naturally
• Good motion prediction
• Works with any image style

Grok

• Good general-purpose video
• Configurable duration (5-10s)
• Creative interpretations
• Works well with abstract concepts

Workflow: Iterative Prompting

# 1. Start with basic prompt
infsh app run google/veo-3-1-fast --input '{
  "prompt": "A woman walking through a forest"
}'

# 2. Add specificity
infsh app run google/veo-3-1-fast --input '{
  "prompt": "Medium tracking shot of a woman in a red dress walking through an autumn forest"
}'

# 3. Add style and technical details
infsh app run google/veo-3-1-fast --input '{
  "prompt": "Cinematic medium tracking shot of a woman in a flowing red dress walking through an autumn forest, golden hour sunlight filtering through leaves, shallow depth of field, film grain, 4K"
}'

Related Skills

# Generate videos
npx skills add inference-sh/skills@ai-video-generation

# Google Veo specific
npx skills add inference-sh/skills@google-veo

# Generate images for image-to-video
npx skills add inference-sh/skills@ai-image-generation

# General prompt engineering
npx skills add inference-sh/skills@prompt-engineering

# Full platform skill
npx skills add inference-sh/skills@inference-sh

Browse all video apps: infsh app list --category video

Documentation

• Running Apps - How to run apps via CLI
• Video Generation Guide - Comprehensive video guide