Operating manual / 01

From image to useful action.

The complete guide to capturing a subject, reading its layers, asking better questions, and turning what you learn into something useful.

01

Quick path

One scan, five moves.

If you want to begin immediately, this is the entire operating loop.

  1. 01

    Add

    Capture, upload, drop, or link an image.

  2. 02

    Read

    Run the first read for a summary and detected aspects.

  3. 03

    Focus

    Tap an aspect, point to a detail, or shade an area.

  4. 04

    Ask

    Continue with dialogue, a deep read, or simplify.

  5. 05

    Act

    Restyle, translate, report, task, or organize it.

02

Input protocol

Give the scan good evidence.

Clear evidence produces a more useful first read. Choose whichever input method fits the situation.

Camera

Open the camera, place the subject in frame, hold steady, then capture. Best for objects in front of you.

Upload

Choose a JPG, PNG, WebP, or another supported image from your device. Images must be under 12 MB.

Drag & drop

Drop an image directly onto the scanner frame. The frame highlights when it is ready to receive it.

Public URL

Paste a public HTTPS image address. Private, local, or inaccessible links cannot be loaded.

Strong capture

Even light, sharp focus, the full subject visible, minimal glare, and enough surrounding context to understand scale and use.

If the result is vague

Retake from another angle, move closer to important markings, reduce reflections, or ask about a selected detail after the first read.

03

Reading protocol

Treat the first read as a map.

The first result is intentionally broad. It identifies the subject, summarizes what matters, and exposes aspects worth exploring.

Summary

A compact interpretation of the entire image. Use it to orient yourself before opening individual aspects.

Aspects

Detected layers such as material, condition, origin, function, care, safety, or value. Tap one to focus its short read.

Confidence

A directional model signal—not proof, a probability, or a professional assessment. Verify consequential findings independently.

Deep read is deliberate. Tapping an aspect keeps the interface quick and shows its existing summary. Choose Deep read only when you want a fuller AI analysis of that aspect.

04

Exploration protocol

Point, shade, then converse.

Use visual focus when the whole image contains too much competing information. Use dialogue when your question depends on context.

Point mode

Tap a detail. Thinking Scan reads a compact region around that point—fast for a component, mark, edge, or feature.

Shade mode

Drag a box around an exact area. This works better when nearby objects could confuse the read.

Dialogue

Ask naturally. Continue with the suggested questions or write your own; the conversation remains attached to the scan.

Useful prompts

What makes you uncertain?What alternatives could this be?Explain the selected detailWhat should I inspect next?Compare the visible materials
05

Execution protocol

Move from insight to output.

The action rail lets the same scan become a concept, a clearer explanation, a translated record, or a task.

Ask

Continue a dialogue about the image. Follow-up suggestions adapt to the latest answer.

Restyle

Describe a finish, colourway, visual language, or material direction to create a concept image.

Translate

Translate the title, summary, and detected aspects. The original remains available.

Task

Turn the focused aspect into a dated action that links back to its source scan.

Simplify

Ask for the entire read in clearer, less technical language.

Organize

Search saved scans and file them into collections in your local library.

Report copies the current scan summary, aspects, and exploration notes as text. Image downloads the original capture. A restyled concept remains visible in the scan record.

06

Storage protocol

A local library, by design.

No account is required. Records are kept in the browser you used, which makes the workflow immediate but device-specific.

Library

Search titles, summaries, and aspects. Create collection names, assign each scan, revisit the dialogue, or delete a record.

Privacy boundary

Saved records stay in this browser. Images and relevant text are sent for processing only when you request an AI action. Read the privacy notice for the complete model.

07

Recovery protocol

Resolve the common blockers.

Most scan issues come from image access, image quality, or local browser storage.

01

The camera does not open

Allow camera access in your browser settings, make sure another app is not using it, or switch to upload, drag and drop, or URL.

02

A URL will not load

Use a direct public HTTPS image link. Web pages, sign-in links, private-network addresses, redirects to blocked locations, and files over 5 MB will not load.

03

The read missed something

Use a clearer angle, include more context, or select the detail with point or shade mode. Ask what evidence is missing.

04

My library is empty on another device

The current release stores records only in the browser where they were created. There is no account sync yet.

05

Translation changed the nuance

Switch back to the original and ask a dialogue question about the exact phrase. Translation preserves intent where possible but remains AI-generated.

06

A result may affect safety or money

Stop and verify it with a qualified person or authoritative source. Thinking Scan is not professional advice or certification.