Nodaro DocsSite Capture
Capture a web page at phone width and get its sections as stills in page order, a full-page image, the section map and the exact figures the page writes.
The Site Capture node opens a web page the way a phone shows it and shoots each of its main sections as a still: the hero, the features, the pricing, the reviews. The node scrolls through the page first, so sections that appear as you scroll are captured too. The stills come out in page order, shaped like phone cards, ready to put on screen in a video. The node also returns the whole page as one image, a map of its sections and the exact figures the page writes, such as prices, counts and ratings.
When to use it
- Show a real product page in a video, for example as cards over a talking-head clip.
- Describe or animate each section of a page, one run per still.
- Turn a page into a slideshow or a collage.
- Read a page's exact prices and numbers into a script, instead of copying them from a screenshot.
Quick start
Add the node
Press Tab on the canvas and choose Automate › Get Content › Site Capture.
Enter the page
Type the page's address in Page, as you type it in a browser. https:// is optional. Or leave Page empty and connect a node whose text holds the address to the Page URL input.
Run it
Click Run. A capture usually takes under a minute, and at most a few minutes. The node shows the elapsed time while it runs, and you can stop the run. The stills appear on the node.
Use the stills
Connect the Stills output to the next node. Set the connection to Each to run that node once per still, or to Bundle to hand it every still.
Settings
| Setting | Default | What it does |
|---|---|---|
| Page | Empty | The page to capture. When it is empty, the node takes the first web address in the text connected to the Page URL input. |
| Stills | 8 | How many section stills to aim for, from 3 to 8. A page with fewer sections gives fewer. |
Inputs
| Input | Required | What to connect |
|---|---|---|
| Page URL | No | A text that holds the page's address, such as a Text node. From a List of addresses, the node runs once for each address. |
Outputs
| Output | What it carries |
|---|---|
| Stills | The section stills, in page order. A plain connection passes the first still. An Each connection runs the next node once per still, and a Bundle connection passes all of them. The full-page image is never one of the stills. |
| JSON | The whole capture: the full-page image, the stills, the map of the page's sections with their text, the page's exact figures (facts) and any warnings. Use Extract Field to take one part of it, such as facts. |
The JSON output looks like this, shortened:
{
"finalUrl": "https://www.example.com/",
"title": "Example: the product",
"fullPage": { "url": "https://…/full.jpg", "width": 412, "height": 6420 },
"stills": [{ "index": 0, "label": "Ship faster", "category": "hero", "url": "https://…/0.png" }],
"sections": [{ "order": 0, "label": "Ship faster", "category": "hero", "text": "…", "stillIndex": 0 }],
"facts": ["$12 per month", "4.8/5", "10,000 teams"],
"warnings": []
}A section's category is hero, feature, pricing, proof, faq, cta or other.
Credits
Each capture costs 11 credits, whatever the number of stills.
- A failed capture is refunded: a page that blocks the visit, a page with nothing on it, a page that cannot be reached, or a capture that runs out of time.
- Some pages are refused before any charge: a page whose site asks automated visitors to stay away (robots.txt), or an address that does not exist.
On a self-hosted install, Site Capture needs the web-scraping key described in Provider keys, or a connection to Nodaro Cloud. An install can also turn capture off; the node is then not offered.
Tips
- Copy the exact numbers. Take the page's prices and figures from
factsin the JSON output, so a script never misreads them from an image. - Pick the stills you need. Each still has a
category, so you can keep the hero and the pricing and leave the rest. - The stills stay in your library. They are hidden from the media picker until you save one.
Troubleshooting
| Problem | Cause and fix |
|---|---|
| The capture fails with a "blocked" message. | The site shows automated visitors a challenge page or refuses them. Upload your own screenshots of that site instead. |
| Fewer stills than Stills. | The page has fewer sections, a section came out blank, or a section's heading never appeared on screen. Such sections are skipped, and the warnings in the JSON output name them. |
| A still shows a cookie or app banner. | The node hides cookie notices it recognizes. A site's own banner can remain. Pick another still, or crop it in a later node. |
From the API
POST /v1/site-capture runs the same capture from code, with { "url": "…", "maxStills": 8 }. It always answers at once with a jobId; read the capture from the completed job with GET /v1/jobs/:id. See Run a single node and Jobs.
Frequently asked questions
What does the Site Capture node return?
Up to 8 stills of the page's sections, such as the hero, the features and the pricing, in page order, each shaped like a phone card. It also returns the whole page as one image, a map of the page's sections and the exact figures the page writes, such as prices and ratings.
How many credits does Site Capture cost?
Each capture costs 11 credits, whatever the number of stills. A capture that fails, for example because the page blocks automated visits, is refunded.
Why did I get fewer stills than I asked for?
Stills is a ceiling. A page with fewer sections gives fewer stills, and a still that comes out blank is skipped. The warnings in the JSON output say which sections were skipped.
Can I capture several pages in one run?
Yes. Put the addresses in a List node and connect it to the Page URL input. The node runs once for each address.
Why did the capture fail on some sites?
Some sites block automated visits with a challenge page, or answer with an error. The node fails with that reason and refunds the credits. Upload your own screenshots of those sites instead.
Related
Web Scrape
Extract Field
List
Lists and batching
Last updated on
Web Scrape
Fetch Google Search results, the text of a web page or site, RSS or Atom feed items, or TikTok posts, and pass them on as JSON to AI and list nodes.
Meta Ads
Search Meta's public Ad Library by keyword, advertiser or Facebook Page, and get each ad's copy, call to action, images and videos as JSON for AI nodes.