generate_image
Turn text prompts into images using a model catalog. Accepts optional aspect ratio, quality tier, and advanced style parameters. One integration works with all models.
Instructions
Generate an image from a text prompt using the MediaEngine model catalog. Use this for any image creation request: illustrations, photos, AI art, logos, icons, graphics, or visuals. Accepts a raw prompt and optional aspect ratio, quality tier, model id, and advanced style parameters. Quality tiers: basic/standard (Nano Banana — fast, inexpensive default), premium (GPT-Image — strongest prompt adherence and clean text rendering). One integration, every model: new models are added to the catalog server-side, so the same wiring keeps working. Style guidance: put the desired look in the prompt itself (e.g. "black and white archival photograph", "flat vector illustration") and bake negatives into the prompt ("no text, no words, no logos"). Use the intent-specific tools (generate_article_hero_image, etc.) when you know the content type.
Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| size | No | Legacy size hint; prefer aspect_ratio. | |
| model | No | Catalog model id (see https://mcpmediaengine.com/models). Overrides quality tier. Default: nano-banana. | |
| prompt | Yes | Text description of the image to generate. Include the desired style in the prompt (e.g. "photorealistic", "watercolor", "pencil sketch") and constraints like "no text, no words, no logos". | |
| quality | No | Quality tier: basic/standard (Nano Banana — fast default), premium (GPT-Image — text rendering, strict adherence) (default: standard). | |
| face_swap | No | Use face swap with reference images (default image model). | |
| hires_fix | No | High-resolution fix with super_resolution (default image model). | |
| scheduler | No | Advanced: sampler/scheduler algorithm (default image model). | |
| controlnet | No | Advanced: structural conditioning mode (default image model). | |
| film_grain | No | Add film grain effect (default image model). | |
| aspect_ratio | No | Output aspect ratio (default: 1:1). | |
| face_correct | No | Apply face correction (default image model). | |
| style_preset | No | Optional style preset (default image model). | |
| color_grading | No | Optional color grading preset (default image model). | |
| inpaint_faces | No | Inpaint faces for better quality (default image model, requires super_resolution). | |
| mask_image_url | No | URL of mask image for inpainting (default image model). | |
| controlnet_txt2img | No | Use structural conditioning for txt2img (default image model). | |
| controlnet_conditioning_scale | No | Conditioning scale 0.0-1.0 (default image model). |