A lightweight Swift package for interacting with OpenAI's gpt-image-1 model to generate or edit images using natural language prompts and image files.
- Supports multiple input images (
image[]) for editing - Uses
multipart/form-datarequests to match OpenAI API requirements - Returns generated image data in decoded PNG format
- Compatible with SwiftUI and UIKit
- Swift Concurrency (
async/await) support
- iOS 15.0+ / macOS 12.0+
- Swift 5.7+
- Valid OpenAI API key with account verification
- Network permission enabled
Add to your project via Swift Package Manager:
.package(url: "https://github.com/darya-varabei/OpenAIImageClient.git", from: "1.0.0")For image generation (generate image from prompt):
private let client = OpenAIImageClient(apiKey: "sk-***") // recommended to NOT store in code for production apps
let prompt = "Generate an image of a spaceship"
do {
let results = try await client.generateImage(prompt: prompt, size: .x1536x1024)
if let data = results.first?.data {
image = UIImage(data: data)
}
} catch {
self.error = error.localizedDescription
}For image edit (edit provided images according to provided prompt):
private let client = OpenAIImageClient(apiKey: "sk-***") // recommended to NOT store in code for production apps
let prompt = "Make the room look like a spaceship"
let images = [UIImage(named: "result")!.pngData()!]
do {
let results = try await client.editImages(images: images, prompt: prompt)
if let data = results.first {
image = UIImage(data: data)
}
} catch {
self.error = error.localizedDescription
}- Test coverage
- More OpenAI requst parameters
- Anything else proposed by community :)