Skip to main content

Craft (Video & Image Generation)

Craft lets you generate AI videos and images on demand through the chat interface. Ask the agent to "generate a video of a sunset over the ocean" or "create an image of my floor plan" and it uses the configured provider to produce the media.

Open via ⚙️ Settings → Abilities → Craft.

What Can You Create?​

TypeExamplesProvider
VideosShort clips (5–16s), animated scenes, visual demonstrationsGoogle Vertex AI (Veo)
ImagesIllustrations, diagrams, concept art, styled photosGoogle Gemini, OpenAI (DALL-E)

Use Cases for Security​

While Craft is primarily a creative tool, it has practical applications for security:

  • Generate reference images: "Create an image of a suspicious person approaching a house at night" — use as a reference for training or presentations
  • Create visual aids: "Generate a diagram showing camera coverage angles for a typical home"
  • Test prompts: Generate sample images to test VLM analysis prompts before deploying them

Providers​

Google Vertex AI — Veo (Video Generation)​

Google's video generation model. Produces short video clips (5–16 seconds) from text descriptions.

Setup:

  1. Get a Google Cloud project with Vertex AI API enabled
  2. Generate an API key or service account credentials
  3. Paste the API key into the Vertex AI field in Craft settings
  4. Select a model (e.g., Veo 2) and set the default duration/aspect ratio

Settings:

  • API Key — your Google Vertex AI key
  • Model — select from available Veo models
  • Duration — default clip length (5s, 8s, 16s)
  • Aspect Ratio — 16:9, 9:16, or 1:1

Google Gemini (Image Generation)​

Google's multimodal AI model with image generation capability.

Setup:

  1. Get a Google AI Studio API key from ai.google.dev
  2. Paste the key and select a Gemini model that supports image output
  3. That's it — image generation is ready

OpenAI (Image Generation)​

Use DALL-E or GPT-4o for high-quality image generation.

Setup:

  1. Enter your OpenAI API key
  2. Select either DALL-E 3, DALL-E 2, or GPT-4o as the generation model
  3. Set the default image size

Settings:

  • API Key — your OpenAI key (same as the one used for LLM, if configured)
  • Model — DALL-E 3 (highest quality), DALL-E 2 (faster), or GPT-4o
  • Size — 1024×1024, 1792×1024, or 1024×1792

Using Craft​

Once configured, ask the agent in Aegis Home:

  • "Generate a video of a peaceful garden at sunset"
  • "Create an image of a modern security camera setup"
  • "Make a 16:9 video of rain falling on a window"

The agent routes the request to the configured provider, shows a progress indicator, and displays the result directly in the chat conversation with options to save the file.

Cost Awareness​

AI media generation uses cloud APIs, so each generation has a cost. Pricing varies by provider, model, resolution, and duration — check each provider's pricing page for current rates:

These costs go directly to the provider — Aegis doesn't add any markup.

Tips​

  • Be descriptive: longer, more detailed prompts produce better results — "a security camera mounted on a white wall overlooking a suburban driveway at dusk" rather than "a camera"
  • Specify aspect ratio: for videos, 16:9 is standard horizontal, 9:16 is phone/vertical
  • Start with images: images generate in seconds and are cheaper — use them to iterate on prompts before committing to video generation