gpt-image-1-5_skill

Overview

This skill generates and edits images using OpenAI's GPT Image 1.5 model. It supports text-to-image generation, full-image edits, and mask-based inpainting for precise edits. Use it when you want automated filename generation, configurable quality/size/background, and reliable mask handling for edits.

How this skill works

For generation it calls the Responses API with the image_generation tool and saves a PNG to the current working directory. For edits it uses the Image API (images.edit) with optional mask input: a transparent mask area indicates where to inpaint. The script accepts prompt, filename, input-image, mask, quality, size, background, and API key parameters and prints the saved image path.

When to use it

• Create new images from a text prompt (text-to-image).
• Edit an existing image without specifying regions (full-image edit).
• Perform precise inpainting using a PNG mask to change only specific areas.
• Request different output sizes or background types (transparent/opaque/auto).
• Need automated, timestamped descriptive filenames for outputs.

Best practices

• Run the script from your current working directory so results save where you expect.
• Provide clear, specific prompts for generation or editing to preserve creative intent.
• Use a PNG mask with transparent areas (alpha=0) for exact inpainting regions.
• Choose quality based on trade-off: low for drafts/fast, medium for balanced, high for final assets.
• Supply an API key via --api-key or set OPENAI_API_KEY in your environment to avoid errors.

Example use cases

• Generate a high-res landscape: prompt a detailed scene and set --quality high --size 1536x1024.
• Remove or replace background: supply input image and use --background transparent (generation) or mask (editing).
• Retouch a photo: run a full-image edit with instructions like "make the sky more dramatic" and provide --input-image.
• Add an element precisely: create a mask for the target area and use prompt like "place a red bicycle in the masked area".
• Produce consistent filenames: the tool generates yyyy-mm-dd-hh-mm-ss-descriptive-name.png for traceable outputs.