What is the difference between medium and high quality?

Medium is optimized for speed and testing. High is slower but generally delivers better fine detail.

Can GPT Image 1.5 perform image-to-image edits?

Yes. Provide input images plus an explicit edit instruction to preserve key visual attributes while changing targeted elements.

The current setup supports 1:1, 3:2, and 2:3.

Use concise, visual instructions with lighting, lens/style cues, and clear constraints on what must remain unchanged.

Create photorealistic outputs and controlled edits with quality-level tuning for faster drafts or premium detail.

Generation Mode

Reference Image0/8

Upload Reference Image

Drag and drop an image here, or click to select

Prompt0 / 6,000

Resolution

Aspect Ratio

Credits are deducted when generation starts. Failed generations are refunded.

Images are saved to My Creations. You can view or delete them there at any time.

Generation Mode

Reference Image0/8

Upload Reference Image

Drag and drop an image here, or click to select

Prompt0 / 6,000

Resolution

Aspect Ratio

Credits are deducted when generation starts. Failed generations are refunded.

Images are saved to My Creations. You can view or delete them there at any time.

Use medium mode for fast iteration and high mode for richer detail, while preserving composition intent across text and edit workflows.

Quality modes

Choose medium or high quality based on speed and fidelity needs.

Edit-preserving prompts

Good for changes that keep identity, pose, and scene structure.

Focused ratios

Simple 1:1, 3:2, and 2:3 ratio set for predictable composition.

Best for

Photorealistic content generation

Quality control

Medium / High

Quick answers about capabilities, input requirements, and output quality.