Image Generation
Generate images from descriptive text prompts. Adjust resolution, steps, guidance, and output format.Front-end controls
Prompt
- Placeholder: “Describe the image you want to generate…”
Parameters (defaults)
- Resolution: 1080x1920 (portrait)
- Steps: 25 (1–50)
- CFG scale: 7.5
- Seed: Random (42 default, customizable)
- Format: JPEG (PNG optional)
Workflow
1
Write a prompt
2
Be specific about subject, style, lighting, and composition.
3
Tune parameters
4
Use defaults first. Increase steps for quality; adjust CFG for prompt adherence.
5
Generate
6
Start generation, then review. Re‑roll with a new seed if needed.
7
Export or iterate
8
Download the result or refine your prompt and generate variations.
Best practices
- Keep prompts under ~200 characters for predictable results
- Use high-level style tags plus concrete details (camera, lighting)
- Prefer PNG when preserving transparency; JPEG for smaller filesize
- Pick aspect ratio templates that match your publishing channel
For comprehensive guidance on writing effective prompts, see the Prompting Guides. These guides cover fundamental principles, text-to-image techniques, and image-to-image workflows that will help you get better results from the Image Generation service.
Prompting Resources
- Prompting Tips - Essential principles for all services
- Text-to-Image Guide - Comprehensive techniques for image generation
- Image-to-Image Guide - Advanced editing and style transfer methods
Cost and usage
- Billed per image per model invocation
- Track spend and balance at Billing & Usage and Pricing
Coming Soon
- Negative prompts and style strength sliders
- Batch grid and variation generator