Tests, actual files and the limits they show

Use these cases to inspect a result before trusting a long batch. They distinguish real recordings from synthetic fixtures and local software checks from production or independent-user verification.

Executed real-recording cases, 20 November 2026

Isolated desktop Chromium 151.0.7922.34 opened a creator-provided 1920 × 1008 screen recording and two Xiph camera sequences at 1280 × 720. The application’s normal controls generated the downloadable outputs.

  • PNG, JPG 40, JPG 95 and WebP 80: the same requested screen-recording moment produced four different file sizes. No OCR or perceptual-quality score was measured.
  • Repeated and nearby camera captures: two PNGs at 0.5 seconds matched; one at 0.6 seconds differed. Hashes do not decide semantic usefulness.
  • Native, Fill and Fit: inspect the 1280 × 720 master and two 1080 × 1920 canvas exports. Fill crops the sides; Fit retains the full image with alpha padding.
  • Real camera preview export: a first external-engine attempt timed out. A bundled-engine retry produced a 656,232-byte, three-second MP4 at width 640. Trimming and scaling make this an inappropriate like-duration quality benchmark.
  • Network observation during local capture: three page-image GETs and no HTTP data-submission attempt in the bounded window. External requests were blocked; this is not a production packet audit.

Download actual still outputs and the result record · Inspect source links, settings and hashes. The Xiph source listing explicitly labels FourPeople and Johnny public domain. The complete creator recording remains private; its historical marketing wording is not a test finding.

What a successful capture establishes

It establishes that this browser decoded this source and exported the checked result. It does not establish a verified source-frame index, HDR colour fidelity, universal codec support or AI restoration of missing detail. Keep the original, check one full-size output, then collect a manageable range.

The gallery retains up to 120 pictures and 96 MB of encoded captures. Contact sheets are sampled overviews. Frame-count calculators use the duration and constant FPS you enter; they do not measure a variable-rate source. Compare the visible controls and limits before a larger job.

Processing and first-use downloads

Frame capture and image resizing use browser media/canvas APIs without an AI model. Some browser encoders are unavailable on some devices. Compatibility processing first tries the pinned FFmpeg files bundled on this site; a missing file can use the pinned distributor fallback. Hashes are checked before execution. A download that stops making progress now gives a connection-stall error instead of waiting indefinitely. Active progress can still take time on a slow connection.

OCR, captions, portrait masks and optional enhancement need their own models and human review. A working frame export does not validate those tools. Cleared storage, another device or private browsing can require downloads again. No conversion is promised to be instant.

What remains unverified

The fresh real-recording cases are desktop Chromium application tests. Physical Android/iOS, Safari, every codec, independent first-time users and representative auxiliary AI results are not certified by those runs. Live Hostinger, current contact delivery, search indexing and AdSense account decisions are separate external checks.

Earlier controlled fixtures

The six-second authored slide example has a source, six actual PNGs and result record. It deliberately demonstrates how interval sampling can miss a short slide; it is not an actual lecture recording.

The older four-frame generated-demo archive and synthetic recovery record remain available with their original dated scope. Their dimensions and contents are not those of the tiny current sample or the new real-camera preview. Synthetic colour bars, tones and test illustrations are engineering fixtures, not customer case studies.

Corrections and reproducible reports

Send the page, browser/device, input container/codec if known, steps and error text through Contact. Do not send private media. Creator and editorial approach · Library/model notices · Privacy policy.

Current technical limits

Reviewed 10 October 2026. These limits come from the same configuration used by the processing modules. MiB and GiB are binary units; older interface labels abbreviate them as MB and GB. Limits are safety caps, not guarantees that every device can complete the largest job.

Limits by processing workflow
WorkflowEnforced limit
Frames / JPG / PNG / endpoints2 GiB source; 8,294,400 decoded pixels; 120 pictures or 96 MiB gallery
Contact sheetsSame source limit; 120 samples; 16,000,000 output pixels; 8192 pixels per side; 1–8 columns
Video conversion / audio / captions100 MiB source; 600 second selected range
Fast 2× video enlargementSame media limit; 8,294,400 output pixels (source pixels × 4)
Optional AI videoSame media file size; 320 × 180 source; 10 second range
Portrait removal / English OCR20 MiB source; 4,000,000 source pixels
Image enlargement50 MiB source; 16,000,000 source pixels; 8192 source pixels per side; 2× dimensions (up to 64,000,000 output pixels)
Image Resizer50 MiB source; 16,000,000 source/output pixels; 8192 pixels per side
Canvas animation preview2–60 images; 24 MiB total; 16,000,000 pixels per input; 640 × 360 preview
Frame calculatorsEntered FPS above 0 through 240; result up to 1,000,000,000 frames

Frame capture admits by total pixels, not one fixed landscape shape: 1080 × 1920 portrait fits, while 4096 × 2160 exceeds 8,294,400 pixels. There is no universal decoder or 8K capture. AI image enlargement can require several large buffers; a 64-MP output is particularly demanding on mobile.

Image Fast resize needs no model. Optional Auto switches to AI at 524,288 source pixels or below. AI video is independently sampled enhancement, not interpolation. Capturing every source frame and exact variable-frame-rate indexing are not implemented.