marketing
image
Image-to-Image Portrait with Cartoon Doodles
A prompt for an image-to-image generation task using Nano Banana Pro. It instructs the model to preserve the identity of an uploaded person while placing them in a hyper-realistic indoor portrait setting (holding coffee,
Prompt
{
"input_type": "image_to_image",
"reference_image": "UPLOAD_IMAGE_HERE",
"prompt": "Take the uploaded image and perfectly preserve the person's face and features. Create a hyper-realistic indoor portrait of the same person sitting at a table, holding a cup of coffee, resting her face on her hand. Add playful hand-drawn cartoon doodles: an excited orange cat jumping next to her, whimsical cat ears on her head, and small floating stars and hearts around. Warm, cozy indoor lighting, soft-focus background, ultra-realistic skin textures, cinematic quality, high-definition.",
"negative_prompt": "blurry, distorted face, low resolution, extra limbs, text, watermark, logo, unnatural colors, unnatural pose",
"guidance_scale": 7.5,
"steps": 25,
"size": "1024x1024"
}
ตัวอย่าง Input
ต้องแนบรูปอ้างอิงก่อนสร้างภาพ
ผลลัพธ์ที่ควรได้
https://cms-assets.youmind.com/media/1766667276666_tov7le_G8vcGJBaQAAWtU-.jpg
ผลลัพธ์หลังลองใช้เป็นอย่างไร
ส่งสัญญาณสั้นๆ เพื่อช่วยให้ PromptThai จัดอันดับและปรับ Prompt ให้ตรงรูปตัวอย่างมากขึ้น
PromptThai