Camera
Open the camera, place the subject in frame, hold steady, then capture. Best for objects in front of you.
Operating manual / 01
The complete guide to capturing a subject, reading its layers, asking better questions, and turning what you learn into something useful.
Guide index
Quick path
If you want to begin immediately, this is the entire operating loop.
Capture, upload, drop, or link an image.
Run the first read for a summary and detected aspects.
Tap an aspect, point to a detail, or shade an area.
Continue with dialogue, a deep read, or simplify.
Restyle, translate, report, task, or organize it.
Input protocol
Clear evidence produces a more useful first read. Choose whichever input method fits the situation.
Open the camera, place the subject in frame, hold steady, then capture. Best for objects in front of you.
Choose a JPG, PNG, WebP, or another supported image from your device. Images must be under 12 MB.
Drop an image directly onto the scanner frame. The frame highlights when it is ready to receive it.
Paste a public HTTPS image address. Private, local, or inaccessible links cannot be loaded.
Even light, sharp focus, the full subject visible, minimal glare, and enough surrounding context to understand scale and use.
Retake from another angle, move closer to important markings, reduce reflections, or ask about a selected detail after the first read.
Reading protocol
The first result is intentionally broad. It identifies the subject, summarizes what matters, and exposes aspects worth exploring.
A compact interpretation of the entire image. Use it to orient yourself before opening individual aspects.
Detected layers such as material, condition, origin, function, care, safety, or value. Tap one to focus its short read.
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.
Exploration protocol
Use visual focus when the whole image contains too much competing information. Use dialogue when your question depends on context.
Tap a detail. Thinking Scan reads a compact region around that point—fast for a component, mark, edge, or feature.
Drag a box around an exact area. This works better when nearby objects could confuse the read.
Ask naturally. Continue with the suggested questions or write your own; the conversation remains attached to the scan.
Useful prompts
Execution protocol
The action rail lets the same scan become a concept, a clearer explanation, a translated record, or a task.
Continue a dialogue about the image. Follow-up suggestions adapt to the latest answer.
Describe a finish, colourway, visual language, or material direction to create a concept image.
Translate the title, summary, and detected aspects. The original remains available.
Turn the focused aspect into a dated action that links back to its source scan.
Ask for the entire read in clearer, less technical language.
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.
Storage protocol
No account is required. Records are kept in the browser you used, which makes the workflow immediate but device-specific.
Search titles, summaries, and aspects. Create collection names, assign each scan, revisit the dialogue, or delete a record.
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.
Recovery protocol
Most scan issues come from image access, image quality, or local browser storage.
Allow camera access in your browser settings, make sure another app is not using it, or switch to upload, drag and drop, or URL.
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.
Use a clearer angle, include more context, or select the detail with point or shade mode. Ask what evidence is missing.
The current release stores records only in the browser where they were created. There is no account sync yet.
Switch back to the original and ask a dialogue question about the exact phrase. Translation preserves intent where possible but remains AI-generated.
Stop and verify it with a qualified person or authoritative source. Thinking Scan is not professional advice or certification.