Update pages/api/gemini.js

pages/api/gemini.js

@@ -31,7 +31,7 @@ export default async function handler(req, res) {
     const model = genAI.getGenerativeModel({ model: "gemini-2.0-flash" });
     
     // Use the custom prompt if provided, otherwise use the default
-    const promptText = prompt || "Look at this image and identify the main object or scene (ignoring any hands or fingers that might be visible). Generate a single, short, insightful thought (maximum 5 words) about this object or scene. Focus on something interesting, fun, or unexpected about it. DO NOT mention hands, fingers, pose estimation, motion tracking, or any computer vision technology in your response. Respond with just the thought, no additional text. Finish with 5 emojis in a coherent manner to convey something profound";
+    const promptText = prompt || "Look at this image and identify the main object or scene (ignoring any hands or fingers that might be visible). Generate a single, short, insightful or profound thought (maximum 5 to 10 words) (bonus if you can use an idiom) about this object or scene. Focus on something interesting, fun, or unexpected about it. DO NOT mention hands, fingers, pose estimation, motion tracking, or any computer vision technology in your response. Respond with just the thought, no additional text. Finish with 5 emojis in a coherent manner to convey something thought provoking and profound. Humor is appreciated and the goal should be the users continued well being";
     
     // Prepare the image part for the model
     const imagePart = {