Images
Text to image with two engines — Gemini and GPT Image. Reference images, aspect ratios, transparent backgrounds.
Generate images from a text prompt. Two engines, each with its own strengths.
What you can do
- Text → image with either engine.
- Reference image — pass an existing image to keep a character or style consistent across generations.
- Control the format — aspect ratio and resolution (Gemini), or transparent background and output format (GPT Image).
Choosing an engine
google_generate_image (Gemini) | openai_generate_image (GPT Image) | |
|---|---|---|
| Strengths | Photoreal, text rendering in images, up to 4K | Strong instruction-following, transparent background |
| Best for | Realistic scenes, posters, images with text | Logos, stickers, precise compositions |
| Reference image | ✓ (character consistency) | ✓ (character consistency) |
Interface
google_generate_image — Gemini (Nano Banana 2)
| Input | Values |
|---|---|
prompt | Text description (required) |
aspect_ratio | 1:1 (default), 16:9, 9:16, 4:3, 3:4, 21:9, … |
image_size | 512, 1K (default), 2K, 4K |
reference_image | An image to anchor character/style consistency |
openai_generate_image — GPT Image 2
| Input | Values |
|---|---|
prompt | Text description (required) |
size | 1024x1024, 1536x1024, 1024x1536, auto (default) |
quality | low, medium, high, auto (default) |
background | transparent (png/webp), opaque, auto (default) |
output_format | png (default), jpeg, webp |
reference_image | An image to anchor character/style consistency |
Both return an image synchronously, in one call.
Using it in a Skill
- Character consistency across scenes: generate a character sheet first, then pass it as
reference_imageto each later generation so the character stays the same. - Pick by need: photoreal / text-in-image / 4K → Gemini; transparent background / strict instructions → GPT Image.