OYYO AI NotzUp: Capture All · guide
Visual understanding
Use local OCR, scene descriptions and sampled silent-video analysis without identity or verified-brand claims.
Before you begin
- An authorized image, screenshot, scan, document image or supported video.
- A compatible local visual/OCR model and sufficient resources; support varies by device and file.
Step by step
Open visual media
Import the image or video and open its file workspace. Video does not need speech for visual analysis.
Analyze locally
Run Visual analysis for detailed scene and object descriptions, labels and face counts. Face counts do not identify people. Video analysis describes sampled frames, not necessarily every frame or event.
Review visible text
Open Visible text to check on-device OCR and visible-text extraction. Compare spelling, numbers and layout with the original; small, blurred or obscured text may be missed.
Treat name and brand hints cautiously
Possible name or brand hints can come from visible text or labels. They are not verified brand matches, identity findings, ownership evidence or authenticity checks.
Search only when you choose
Explicitly start a web search for a detected object or name if wanted. Inspect the query first: the search service receives the submitted terms, and related results are not proof of a match.
Correct and export
Edit Visual analysis and Visible text, then export the reviewed sections or the full printable report.
Expected result
Reviewed descriptions, labels and extracted text, including for sampled frames of silent video.
Privacy and permissions
Recordings, files, searches, scans, OCR, barcode counts and supported local-AI results do not leave the device by default. Only a deliberate online action—Share, Search online, connector authorization or encrypted backup—transmits the selected data or value. Language-pack downloads do not upload the source text. Files, recordings, scans, local searches and local-AI results stay on your device by default. All inference runs on your own awake, authorized device, with no cloud AI, hidden cloud processing or remote fallback. You choose what to share, with whom, in which format and when. Online searches, connectors, sharing and optional backup transmit the data needed for the action you explicitly choose.
Troubleshooting
Incorrect label or hint
Correct it in the workspace. Improve lighting/resolution and check visible text; never assume the model verified a brand or identified a face.
A video event is missing
Sampled frames are not exhaustive. Inspect playback locally and supply a clearer image of the relevant moment if supported.