Move from ScreenshotOne.

Start with the boundary

Move the capture only if the smaller contract fits.

Latchshot is deliberately narrower than ScreenshotOne. The migration is straightforward when the job is “load this public URL at a bounded viewport and return an artifact.” It is the wrong move when the provider is also modifying the page, bypassing access controls, storing the result, or orchestrating a workflow.

Move the key, then rename the options.

The material security change is authentication. ScreenshotOne commonly receives an access key as a query parameter. Latchshot requires the key in an HTTPS Bearer header and does not accept it in the target URL or request query.

Before: synchronous ScreenshotOne request

curl -fSLo before.png \
  "https://api.screenshotone.com/take?access_key=$SCREENSHOTONE_ACCESS_KEY&url=https%3A%2F%2Fexample.com&viewport_width=1440&viewport_height=900&full_page=false&format=png"

After: synchronous Latchshot request

curl -fSL \
  'https://latchshot.fly.dev/v1/render' \
  -H "Authorization: Bearer $LATCHSHOT_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
    "url": "https://example.com",
    "kind": "screenshot",
    "format": "png",
    "width": 1440,
    "height": 900,
    "fullPage": false,
    "blockAds": true,
    "blockTrackers": true,
    "hideCookieBanners": true
  }' \
  -o after.png

Keep secrets in the runtime's credential store or environment. Do not copy a previous query-string key into logs, exported workflows, browser code, or a Latchshot target URL.

Translate only the controls you use.

Do not mechanically copy an entire ScreenshotOne options object. Start with the smallest request that produces the required artifact, then add one control at a time and compare representative pages.

On a narrow screen, scroll the option table horizontally.

Stop if these options carry the workload.

Latchshot has no equivalent for the following ScreenshotOne capabilities. Removing one can materially change the image or break the surrounding workflow.

• Page cleanup: custom or exhaustive ad/tracker/chat filter lists, popup handling beyond common known overlays, or any workflow that requires cleanup to be complete rather than best-effort.
• Element capture: selector screenshots, clip coordinates, click, hover, and element hiding.
• Page modification: injected CSS or JavaScript, HTML or Markdown input, and arbitrary request headers, cookies, or authorization.
• Network placement: country proxies, custom proxies, geolocation, private hosts, and non-standard target ports.
• Delivery orchestration: async jobs, webhooks, hosted artifact URLs, provider CDN caching, and direct S3 upload.
• Special outputs: WebP, AVIF, GIF, animation, metadata extraction, page HTML/Markdown, and provider-side vision analysis.

If one of these is required, keep ScreenshotOne for that job. A split-provider architecture is safer than silently degrading the result.

Cut over with evidence and a rollback.

1. Inventory the exact options present in production requests; reject the migration if any required feature appears on the stop list.
2. Create a representative set of public pages: ordinary, slow, animated, font-heavy, consent-banner, tall, and known-failure targets.
3. Render the same fixed viewport and format through both providers. Compare image dimensions, visible content, fonts, navigation completion, and latency.
4. Store Latchshot's X-Latchshot-Render-Ms, X-Latchshot-Navigation, X-Latchshot-Fonts, X-Latchshot-Scripts, and quota headers with job diagnostics.
5. Route a bounded percentage of eligible jobs to Latchshot. Keep the previous provider configuration until the sample passes and rollback is exercised.

Retry only 429, 502, 503, and 504, with a strict attempt cap. Treat 400, 401, 403, and target-policy failures as permanent until the request changes.

Compare successful output, not headline quota.

Latchshot counts successful direct renders. Failed direct renders do not consume monthly quota, and automatic overages are disabled during the beta. The current plans are Free at 100 successful renders each UTC month, Launch at $19 for 2,500, Build at $49 for 12,000, and Scale at $149 for 50,000.

Those prices are inputs, not proof that Latchshot is cheaper for your workload. Include the captures that cannot migrate, output storage, transfer, retry volume, integration work, and the operational value of the features on the stop list.

Create one Free-plan key, save it when it is returned, and spend the 100 monthly successful renders on a representative comparison set.