To receive a caption, let\u2019s add The image features a pair of purple and white sneakers. Next, we\u2019ll feed this caption to ChatGPT as well as any additional information we\u2019d like to include and receive the marketing copy back.\u00a0<\/p>\n\n\n ChatGPT without additional information returns the following marketing copy:<\/p>\n\n\n\n Step up your style game with our eye-catching purple and white sneakers! These shoes are designed for both comfort and fashion, making them the perfect addition to any outfit. Don’t miss out on the chance to make a bold statement with your footwear choice!!<\/p>\n<\/blockquote>\n\n\n\n This is a great placeholder, but we can add more info. Let\u2019s talk about info about pricing and features. Let\u2019s add: \u201cThey cost $200, and are perfect for playing basketball.\u201d This could also be stored in the asset\u2019s metadata and be fully automated:<\/p>\n\n\n\n Elevate your game with these $200 purple and white basketball sneakers. Designed for performance and style, this dynamic duo will have you dominating the court in both comfort and flair. Perfect for all your hoop dreams! #GameOn #HoopsHighFashion<\/p>\n<\/blockquote>\n\n\n\n Once we\u2019re happy with the product description, let\u2019s save this info to the asset\u2019s metadata.<\/p>\n\n\n Integrating Cloudinary with ChatGPT revolutionizes how we handle image uploads and content creation. This workflow automates the generation of engaging marketing copy, streamlining your production process with Cloudinary\u2019s AI-driven image analysis and ChatGPT\u2019s creative writing.<\/p>\n\n\n\n Stay ahead in content creation. Visit the Cloudinary AI page<\/a> or contact Cloudinary today<\/a> to transform your content creation workflow.<\/p>\n\n\n\ndetection:\"captioning\"<\/code> to the upload parameters.<\/p>\n\n\nasync<\/span> function<\/span> uploadImage<\/span>(<\/span>) <\/span>{\n\n\u00a0\u00a0try<\/span> {\n\n\u00a0\u00a0\u00a0\u00a0const<\/span> uploadResult = await<\/span> cloudinary.uploader.upload(\"assets\/images\/shoes.jpg\"<\/span>, { detection<\/span>: \"captioning\"<\/span> });\n\n\u00a0\u00a0\u00a0\u00a0caption = uploadResult.info.detection.captioning.data;\n\n\u00a0\u00a0\u00a0\u00a0imageURL = uploadResult.secure_url;\n\n\u00a0\u00a0\u00a0\u00a0useCaption(caption, imageURL);\n\n\u00a0\u00a0} catch<\/span> (error) {\n\n\u00a0\u00a0\u00a0\u00a0console<\/span>.error(error);\n\n\u00a0\u00a0}\n\n}<\/code><\/span>Code language:<\/span> JavaScript<\/span> (<\/span>javascript<\/span>)<\/span><\/small><\/pre>\n\n\n
We\u2019ll use this sample picture of a pair of shoes:<\/p>\n\n\n\n![](\"https:\/\/lh7-rt.googleusercontent.com\/docsz\/AD_4nXceDpdotw7-Ity5Go9bXn5yLrNDGVUtpvzhgBDJSGG1FX0Ijz5EutkB_QK5_kU_7Ti1zPY1UNLAAUynXgd0IWbMehT8Bow8I4FJl8eXGKcz7a04dxGbRTsGXqzHGIm1eHNlNPuOOwljaGrL9NkY0b1xj-ff?key=yLTgYk2NDUMnZxE2poYSBA\")
The Cloudinary AI Content Analysis add-on<\/a> returns the caption:<\/p>\n\n\n\n\n
<\/p>\n<\/blockquote>\n\n\n\nIntegrating With ChatGPT<\/h2>\n\n\n\n
async function<\/span> APIcall<\/span>(description)<\/span> <\/span>{\n\n\u00a0\u00a0const<\/span> openai = new<\/span> OpenAI({\n\n\u00a0\u00a0\u00a0\u00a0apiKey: process.env.OPENAI_API_KEY\n\n\u00a0\u00a0});\n\n\u00a0\u00a0const<\/span> chatHistory = [];\n\n\u00a0\u00a0do<\/span> {\n\n\u00a0\u00a0\u00a0\u00a0const<\/span> user_input = readlineSync.question(`Your picture appears to be ${description}, add more information? If<\/span> you're happy with the prompt, just submit: `);\n\n\u00a0\u00a0\u00a0\u00a0const messageList = chatHistory.map(([input_text, completion_text]) => ({\n\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0role: input_text === \"user\" ? \"user\" : (input_text === \"system\" ? \"system\" : \"assistant\"),\n\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0content: input_text,\n\n\u00a0\u00a0\u00a0\u00a0}));\n\n\u00a0\u00a0\u00a0\u00a0messageList.push({ role: \"system\", content: \"You are an amazing marketing creative, please use the following description of an image to write a short product summary. Make sure the post only uses factual information and is 50 words or less, is engaging, creative and ultimately makes the reader want to buy the product\n\n\" }); \/\/ Add system prompt\n\n\u00a0\u00a0\u00a0\u00a0messageList.push({ role: \"user\", content: `The default description is: \"${description} ${user_input.length > 1 ? `and I'<\/span>d like to add the following, ${user_input}` : \"\"<\/span>}` }); \/\/ Include description in user input<\/span>\n\n\u00a0\u00a0\u00a0\u00a0try<\/span> {\n\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0const<\/span> GPTOutput = await openai.chat.completions.create({\n\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0model: \"gpt-3.5-turbo\"<\/span>,\n\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0messages: messageList,\n\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0});\n\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0const<\/span> output_text = GPTOutput.choices[0<\/span>].message.content;\n\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0console.log(\"Your photo URL:\"<\/span>, imageURL, \"Your caption:\"<\/span>, output_text);\n\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0chatHistory.push([user_input, output_text]);\n\n\u00a0\u00a0\u00a0\u00a0} catch<\/span> (err) {\n\n\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0console.error(err);\n\n\u00a0\u00a0\u00a0\u00a0}\n\n\u00a0\u00a0} while<\/span> (readlineSync.question(\"\\nWould you like to add anything to the post? (Y\/N)\"<\/span>).toUpperCase() === \"Y\"<\/span>);\n\n}<\/code><\/span>Code language:<\/span> PHP<\/span> (<\/span>php<\/span>)<\/span><\/small><\/pre>\n\n\n\n
Refining Prompts<\/h2>\n\n\n\n
\n
cloudinary<\/span>.v2<\/span>.uploader<\/span>.update_metadata<\/span>(output_text<\/span>, public_id<\/span>, options<\/span>).then<\/span>(callback<\/span>);<\/code><\/span>Code language:<\/span> CSS<\/span> (<\/span>css<\/span>)<\/span><\/small><\/pre>\n\n\nConclusion<\/h2>\n\n\n\n