See what changed.
Keep the context.
Visual monitors compare screenshots. Use them when appearance matters: a layout, chart, banner, or page section. For prices, wording, stock or new listings, a content monitor usually gives a more direct explanation.
A URL, a screenshot, and go.
- Choose Visual changes when adding a monitor. Paste a public webpage URL.
- Load the cloud preview. Keep the visible page, choose a full page within the capture limit, use a unique CSS element, or draw a region. Coordinates are also editable for keyboard users.
- Use the default 1% threshold, or expand Noise controls. Give the monitor a name and choose a schedule.
- Start monitoring. The confirmed image becomes a quiet baseline without an initial alert. Email and webhook controls remain on the details page.
A region drawn on a visible-page or full-page preview reuses that screenshot. Changing ignored selectors, the URL, the element, or page loading settings requires a new capture. Setup captures use the existing preview allowance.
The cloud does not inherit browser logins or cookies. An element passed from the Chrome extension remains selected, but PageFluxer must find it in its own public cloud capture.
Less noise. More useful changes.
- Change threshold: the percentage of unmasked pixels that must change before an event is recorded. Choose 0.1%–50%; the default is 1%. This describes changed area, not confidence or importance.
- Pixel tolerance: Precise notices smaller color differences, Balanced is the default, and Relaxed ignores more subtle differences. Anti-aliasing differences are filtered in all three modes.
- Ignored elements: up to eight CSS selectors, one per line. Matching elements are masked without removing their layout space. Choose a container whose box encloses the entire area you want to ignore. Use this for clocks, rotating ads or changing counters. Masks are excluded from the percentage calculation.
- Page loading: use the same extra wait, limited scrolling, and readiness selector as other Browser checks. Visual monitors always use a browser.
Captures use a fixed 1365px-wide desktop viewport, 900px viewport height, UTC, English locale, light appearance and device scale one. Animations, transitions and carets are disabled. These controls cannot freeze every canvas, video, random layout or personalized response; narrow the selection or mask that area.
Small changes accumulate against the last accepted baseline. A below-threshold check does not move the baseline forward. Once the threshold is met, PageFluxer records a change and uses the new screenshot for future comparisons.
Before, after, and the difference.
View the original screenshot, new screenshot, or highlighted pixel diff. Drag the comparison slider to reveal before and after. When dimensions differ, use the individual images or highlighted diff; images are not stretched to hide a layout change.
Only the baseline and meaningful changes are saved. Unchanged checks do not create image history. Screenshots stay in private object storage; Postgres stores references and small metadata records. Emails link to the authenticated comparison page. Screenshots are not sent to an AI provider.
Image history expires at the plan limit or is trimmed sooner when storage fills. Expired images become inaccessible in the app; background cleanup and bucket lifecycle rules remove stored objects. Physical deletion can lag access expiry. Event summaries can remain for the longer text-history period.
The current baseline is stored separately and renewed while checks continue. A baseline not renewed for 90 days expires, including a long-paused monitor; load a fresh preview to resume comparisons. Editing capture settings creates a quiet new baseline. A changed renderer profile also establishes a new baseline and shows a notice.
Downgrading shortens image access immediately and pauses excess active visual monitors. Upgrading cannot restore deleted images. Deleting a monitor or account queues screenshots for removal.
One allowance for Browser work.
| Plan | Active visual monitors | Image history | Image storage | Check allowance |
|---|---|---|---|---|
| Free | 1 during trial | Up to 7 days | 50 MiB | 168 attempts in the one-time 7-day Browser trial |
| Starter | 3 | Up to 14 days | 250 MiB | 750 Browser attempts/month |
| Pro | 10 | Up to 30 days | 1024 MiB | 5,000 Browser attempts/month |
Visual monitors use both total and Browser monitor slots. One attempted visual check uses one Browser check, including failed website captures. There is no additional pixel-diff credit. The Free trial is shared with other Browser monitors and does not restart for a new visual monitor.
Image storage is shared across the account and includes previews, baselines, before/after/diff images, and pending deletion. One MiB is 1,048,576 bytes. Near the limit, oldest image groups are trimmed. A full allowance may delay a new capture until cleanup runs; the previous baseline is kept. Schedules, service capacity and monthly allowances all apply.
Complete captures within clear limits.
- Public webpages only. PDF monitoring continues to compare document text, not PDF screenshots.
- Up to 1365 × 3000 pixels and 3 MiB per PNG. A full page beyond that height is rejected, never silently cropped. Use a region or element for long pages.
- 160 requests, 40 image requests, 12 MiB of image data, and 20 MiB total downloaded data. Individual image resources are limited to 2 MiB.
- Required images and fonts must finish loading. An incomplete or oversized capture keeps the previous baseline and reports an error.
- Frames, sign-in flows, browser cookies, arbitrary scripts and access-challenge bypasses are unsupported. Recorded interactions are not included in this release.
Too many changes? Use a smaller region, mask dynamic content, or raise the threshold. Missing element? Check that the selector identifies exactly one visible element in the cloud. Missing images? Try extra loading time or a smaller selection; some sites block cloud requests. A screenshot cannot prove a site is available to every visitor.
Your existing delivery workflow.
Email and Pro webhooks operate independently. Visual emails describe the changed percentage and link to the comparison. A visual webhook includes numeric change details and the authenticated change link, without image bytes or expiring storage URLs.
The Pro API can list visual monitors, retrieve change details, pause or resume them, update notification settings, and request checks under the normal allowance. Change-detail responses include short-lived image URLs for authorized callers; treat them as temporary credentials. Create visual monitors and edit capture settings in the website in this version. Existing content-monitor and extension requests keep their existing meaning.
The shared page-loading controls run before visual capture. Future recorded interactions can extend that sequence; changing an interaction recipe will need a fresh baseline, just like changing capture settings.