# Built-in image-generation platforms.
#
# To add a new platform: append an entry under `platforms:` with a unique `id`,
# a human label, a one-line description, and the platform's syntax hints.
# Restart the server (or rebuild) to pick up the change.

category:
  id: image
  label: Image
  description: Image generation
  defaultPlatform: midjourney
  defaultMode: detailed

platforms:
  - id: midjourney
    label: Midjourney
    description: Artistic, stylized imagery
    syntaxHints:
      - "--ar"
      - "--v 6.1"
      - "--style raw"
      - "--chaos"
      - "--weird"
      - "--q"
      - "--s"

  - id: dall-e
    label: DALL-E 3
    description: Natural language, versatile
    syntaxHints:
      - natural language
      - "size: 1024x1024, 1792x1024, 1024x1792"

  - id: stable-diffusion
    label: Stable Diffusion
    description: Open source, highly customizable
    syntaxHints:
      - negative prompts
      - CFG scale
      - steps
      - samplers
      - LoRA
      - embeddings

  - id: flux
    label: Flux
    description: High detail, photorealistic
    syntaxHints:
      - natural language
      - high detail focus
      - guidance scale

  - id: ideogram
    label: Ideogram
    description: Best for text in images
    syntaxHints:
      - magic prompt
      - text rendering
      - typography

  - id: leonardo
    label: Leonardo AI
    description: Preset styles, game art
    syntaxHints:
      - preset styles
      - guidance scale
      - contrast
      - alchemy

  - id: firefly
    label: Adobe Firefly
    description: Commercial safe, natural
    syntaxHints:
      - natural language
      - style references
      - effects

  - id: grok-aurora
    label: Grok Aurora
    description: xAI, fast and creative
    syntaxHints:
      - natural language
      - creative interpretation
      - fast generation

  - id: imagen
    label: Google Imagen 3
    description: Photorealistic, via Gemini
    syntaxHints:
      - natural language
      - photorealistic
      - "size: 1024x1024"
      - aspect ratios

  - id: nano-banana
    label: Nano Banana (Gemini 2.5 Flash Image)
    description: Natural-language scene direction; character consistency, multi-image edits, in-image text
    syntaxHints:
      - describe the whole scene in natural language (direct the scene, don't list keywords)
      - photographic terminology to control the look (camera/lens, angle, depth of field)
      - explicit lighting (e.g. "three-point softbox setup", "golden-hour backlight")
      - editing — state what CHANGES and what STAYS the same (preserves subject identity)
      - combine multiple reference images for character consistency / product placement
      - reliable in-image text rendering
      - specify aspect ratio

  - id: recraft
    label: Recraft
    description: Vector design, brand assets
    syntaxHints:
      - style selection
      - vector output
      - brand colors
      - SVG export

  - id: higgsfield
    label: Higgsfield (multi-model platform)
    description: Multi-model creative platform via MCP / CLI / Skills. Routes one natural-language prompt to Soul 2.0, Soul Cinema, Soul Cast, Flux 2, Seedream 5, Nano Banana Pro, or GPT Image 2. Soul ID for face-faithful character reuse. No API key — auth via Higgsfield account.
    syntaxHints:
      - "models: soul-2.0 | soul-cinema | soul-cast | flux-2 | seedream-5 | nano-banana-pro | gpt-image-2"
      - long-form natural-language prose (composition + lighting + textures + mood)
      - Soul ID for face-faithful identity reuse
      - multi-reference compositing for edits
      - output up to 4K
      - "modes: Marketing Studio, Fashion Factory, Photodump Studio"