163 lines
5.3 KiB
JavaScript
163 lines
5.3 KiB
JavaScript
#!/usr/bin/env node
|
|
import { GoogleGenerativeAI } from '@google/generative-ai';
|
|
import { readFileSync, writeFileSync, existsSync } from 'fs';
|
|
import { resolve } from 'path';
|
|
|
|
interface EditOptions {
|
|
model?: string;
|
|
}
|
|
|
|
async function editImage(
|
|
sourcePath: string,
|
|
prompt: string,
|
|
outputPath: string,
|
|
options: EditOptions = {}
|
|
): Promise<void> {
|
|
const apiKey = process.env.GEMINI_API_KEY;
|
|
|
|
if (!apiKey) {
|
|
console.error('Error: GEMINI_API_KEY environment variable is required');
|
|
console.error('Get your API key from: https://makersuite.google.com/app/apikey');
|
|
process.exit(1);
|
|
}
|
|
|
|
const resolvedSourcePath = resolve(sourcePath);
|
|
|
|
if (!existsSync(resolvedSourcePath)) {
|
|
console.error(`Error: Source image not found: ${resolvedSourcePath}`);
|
|
process.exit(1);
|
|
}
|
|
|
|
const { model = 'gemini-2.0-flash-exp' } = options;
|
|
|
|
console.log('Editing image...');
|
|
console.log(`Source: ${resolvedSourcePath}`);
|
|
console.log(`Prompt: "${prompt}"`);
|
|
console.log(`Model: ${model}`);
|
|
|
|
try {
|
|
const genAI = new GoogleGenerativeAI(apiKey);
|
|
const generativeModel = genAI.getGenerativeModel({ model });
|
|
|
|
// Read and encode the source image
|
|
const imageData = readFileSync(resolvedSourcePath);
|
|
const base64Image = imageData.toString('base64');
|
|
|
|
// Determine MIME type from file extension
|
|
const mimeType = getMimeType(resolvedSourcePath);
|
|
|
|
console.log(`Image size: ${(imageData.length / 1024).toFixed(2)} KB`);
|
|
console.log(`MIME type: ${mimeType}`);
|
|
|
|
// Use Gemini's vision capabilities to analyze and describe the edit
|
|
const enhancedPrompt = `You are an image editing assistant. Analyze this image and describe how to apply the following edit: "${prompt}". Provide detailed instructions for the transformation.`;
|
|
|
|
const result = await generativeModel.generateContent([
|
|
{
|
|
inlineData: {
|
|
data: base64Image,
|
|
mimeType: mimeType
|
|
}
|
|
},
|
|
enhancedPrompt
|
|
]);
|
|
|
|
const response = result.response;
|
|
const editInstructions = response.text();
|
|
|
|
console.log('\nEdit Analysis:');
|
|
console.log(editInstructions);
|
|
|
|
// For actual image editing with Gemini, you would typically:
|
|
// 1. Use the Imagen model's image editing capabilities
|
|
// 2. Send the source image with the edit prompt
|
|
// 3. Receive the edited image as base64
|
|
// 4. Save to output path
|
|
|
|
console.warn('\nNote: This is a demonstration implementation.');
|
|
console.warn('For actual image editing, you would use Gemini\'s image editing API.');
|
|
console.warn('The model has analyzed the image and provided edit instructions.');
|
|
|
|
// In a real implementation with Imagen editing:
|
|
// const editedImageData = Buffer.from(response.candidates[0].content.parts[0].inlineData.data, 'base64');
|
|
// writeFileSync(resolve(outputPath), editedImageData);
|
|
|
|
console.log(`\nTo implement actual image editing:`);
|
|
console.log(`1. Use Gemini's image editing endpoint`);
|
|
console.log(`2. Send source image with edit prompt`);
|
|
console.log(`3. Parse the edited image data from response`);
|
|
console.log(`4. Save to: ${resolve(outputPath)}`);
|
|
console.log(`\nRefer to: https://ai.google.dev/docs/imagen`);
|
|
|
|
} catch (error) {
|
|
if (error instanceof Error) {
|
|
console.error('Error editing image:', error.message);
|
|
if (error.message.includes('API key')) {
|
|
console.error('\nPlease verify your GEMINI_API_KEY is valid');
|
|
}
|
|
} else {
|
|
console.error('Error editing image:', error);
|
|
}
|
|
process.exit(1);
|
|
}
|
|
}
|
|
|
|
function getMimeType(filePath: string): string {
|
|
const extension = filePath.toLowerCase().split('.').pop();
|
|
const mimeTypes: Record<string, string> = {
|
|
'jpg': 'image/jpeg',
|
|
'jpeg': 'image/jpeg',
|
|
'png': 'image/png',
|
|
'gif': 'image/gif',
|
|
'webp': 'image/webp',
|
|
'bmp': 'image/bmp'
|
|
};
|
|
return mimeTypes[extension || ''] || 'image/jpeg';
|
|
}
|
|
|
|
// Parse command line arguments
|
|
function parseArgs(): { sourcePath: string; prompt: string; outputPath: string; options: EditOptions } {
|
|
const args = process.argv.slice(2);
|
|
|
|
if (args.length < 3) {
|
|
console.error('Usage: edit-image.ts <source-image> <prompt> <output-path> [options]');
|
|
console.error('\nArguments:');
|
|
console.error(' source-image Path to the image to edit');
|
|
console.error(' prompt Text description of the desired changes');
|
|
console.error(' output-path Where to save the edited image');
|
|
console.error('\nOptions:');
|
|
console.error(' --model <string> Gemini model to use (default: gemini-2.0-flash-exp)');
|
|
console.error('\nExample:');
|
|
console.error(' GEMINI_API_KEY=xxx npx tsx scripts/edit-image.ts photo.jpg "add a blue sky" edited.jpg');
|
|
process.exit(1);
|
|
}
|
|
|
|
const sourcePath = args[0];
|
|
const prompt = args[1];
|
|
const outputPath = args[2];
|
|
const options: EditOptions = {};
|
|
|
|
// Parse options
|
|
for (let i = 3; i < args.length; i += 2) {
|
|
const flag = args[i];
|
|
const value = args[i + 1];
|
|
|
|
switch (flag) {
|
|
case '--model':
|
|
options.model = value;
|
|
break;
|
|
default:
|
|
console.warn(`Unknown option: ${flag}`);
|
|
}
|
|
}
|
|
|
|
return { sourcePath, prompt, outputPath, options };
|
|
}
|
|
|
|
// Main execution
|
|
const { sourcePath, prompt, outputPath, options } = parseArgs();
|
|
editImage(sourcePath, prompt, outputPath, options).catch((error) => {
|
|
console.error('Fatal error:', error);
|
|
process.exit(1);
|
|
});
|