What is Image Pack?
The image pack is the grid of images Google embeds mid-SERP for visual queries. Each image result includes a thumbnail URL, a full-resolution source URL, the hosting page URL, the page title, and the alt or title text when available. Scavio also resolves redirected image URLs to their canonical source, which saves a roundtrip per image when you want to download or embed the asset. For ecommerce, design, and dataset curation workflows, the image pack is a high-quality seed of visually relevant assets. AI agents use it to build mood boards, retrieve reference imagery for multimodal prompts, and power visual search features without running their own image index.
Example Response
{
"image_pack": [
{
"position": 1,
"thumbnail": "https://serpapi.scavio.dev/img/thumb/golden-gate-01.jpg",
"original": "https://example.com/photos/golden-gate-full.jpg",
"width": 3840,
"height": 2160,
"source_page": "https://example.com/blog/san-francisco-landmarks",
"source_title": "San Francisco Landmarks Guide 2026",
"alt_text": "Golden Gate Bridge at sunrise",
"domain": "example.com"
},
{
"position": 2,
"thumbnail": "https://serpapi.scavio.dev/img/thumb/golden-gate-02.jpg",
"original": "https://example.com/media/ggb-aerial.jpg",
"width": 4096,
"height": 2730,
"source_page": "https://example.com/travel/sf-aerial",
"source_title": "Aerial Views of San Francisco",
"alt_text": "Aerial view of Golden Gate Bridge",
"domain": "example.com"
}
]
}Use Cases
- Building reference image sets for multimodal AI agents
- Product image sourcing for ecommerce catalogues
- Visual search and reverse image workflows
- Design mood-board generation
- Dataset curation for computer vision training
Why Image Pack Matters
Getting full-resolution images at scale requires dodging Google's image proxy and resolving messy redirects. Scavio handles that resolution server-side and returns the original hosted URL with width, height, and alt text, so you can filter by quality and embed assets directly. Multimodal AI applications need this level of structured access to avoid the slow, error-prone scraping most teams otherwise write in-house.
LangChain Example
Drop image pack data into your LangChain agent in a few lines:
from langchain_scavio import ScavioSERPTool
tool = ScavioSERPTool(api_key="your_scavio_api_key", extract="image_pack")
result = tool.invoke({"query": "golden gate bridge sunrise"})
high_res = [img for img in result["image_pack"] if img["width"] >= 3000]
print(f"Found {len(high_res)} high-resolution candidates")
for img in high_res[:5]:
print(img["original"], img["alt_text"])