Multi-viewport ZIP packs

Request

{
  "url": "https://example.com",
  "output": "png",
  "full_page": false,
  "viewports": [
    { "name": "desktop", "width": 1280, "height": 720, "device_scale_factor": 1, "device": "desktop" },
    { "name": "tablet", "width": 768, "height": 1024, "device_scale_factor": 1, "device": "ipad" },
    { "name": "mobile", "width": 390, "height": 720, "device_scale_factor": 1, "device": "iphone_14" }
  ]
}

Rules

• viewports contains exactly 2 or 3 items. Names are unique 1–32 character identifiers using letters, digits, underscore, or hyphen.
• Only PNG, JPEG, or WebP is valid. The pack requires full_page=false and cannot use selector or clip.
• Every item uses the normal viewport and pixel limits. Its optional device applies device signals for that item.
• ViperCapture renders items sequentially with fresh isolated contexts inside one queue admission and one overall render deadline. It never consumes multiple browser slots concurrently.
• API cost equals image count: two items cost two credits and three cost three. Any failure aborts the pack and releases the entire reservation.
• The sum of stored image bytes plus manifest must fit the plan output-byte limit. ZIP uses stored entries because images are already compressed.

Archive layout

vipercapture-viewports.zip
├── desktop.png
├── tablet.png
├── mobile.png
└── manifest.json

manifest.json has schema_version: 1, output format, count, and one entry per image containing name, filename, media type, pixel width, pixel height, and navigation status. It does not include the source URL or request headers. The HTTP response uses application/zip and X-ViperCapture-Output-Count.

Website workspace

Enable “Responsive ZIP pack” in Capture settings. The workspace chooses desktop, tablet, and mobile viewports that fit the current account. ZIP files download directly and do not use temporary image delivery.