Compare pixels, alpha, scaling, OCR, alternative text, captions, metadata, symbols, provenance, and model-facing visual instructions.
An image is simultaneously a pixel array, an encoded container, a metadata package, an accessibility object, an OCR source, and a model input; secure ingestion compares these representations and prevents extracted text or symbols from inheriting execution authority.
The submitted report remains byte-identical. This separate review, checked 2026-08-26, narrows implementation claims and gives visitors a decision path before they generalize from a local result.
Output produced deterministically by the local runtime.
Useful model or subset that must not be mistaken for full conformance.
Conclusion that needs an exact implementation, trust system, model, parser, or human review.
| Question | What the local page can answer | What it does not establish | Next evidence step |
|---|---|---|---|
| What supported container and metadata are present? | The inspector reports bounded structural facts. | What a person or model sees in the pixels. | Compare with a controlled render and pinned OCR/VLM pipeline. |
| Could transparency create a representation difference? | The lab reports selected alpha indicators. | Actual compositing, hidden text, or model preprocessing. | Render against known backgrounds and inspect decoded channels. |
| Is the image safe or authentic? | No safety or authenticity verdict is issued. | Malware, semantic deception, watermark, provenance, or truth. | Use specialized isolated analysis and independent evidence. |
Separates container facts from OCR, visual, alternative-text, and model claims.
Expected boundary: Structural evidence is exact; OCR and VLM conclusions remain explicitly external.
Open prepared laboratoryNo single view is automatically authoritative. Preserve the source, identify each parser or receiver, and compare their outputs before authorizing a consequential decision.
Signature, dimensions, color type, markers, chunks, and digest.
The human-visible rendered image after color, alpha, crop, and scaling.
Text inferred from pixels under a specific engine and preprocessing configuration.
Alt text, captions, and surrounding semantic context.
EXIF, XMP, comments, textual chunks, and thumbnails.
The resized tensor plus any textual context supplied to a multimodal model.
The workflow preserves evidence before transformation and keeps structural inspection separate from execution, remote verification, or model behavior.
These controls are contextual. They reduce a defined risk; they do not guarantee safety, truth, attribution, or resistance to every adaptive attack.
Never use alt text, OCR, or metadata as privileged instructions.
Disable remote resource resolution during document and image inspection.
Treat embedded thumbnails as a separate potentially sensitive representation.
Use accessible descriptions that match the visible purpose of the image.
Record OCR engine, language, segmentation mode, confidence, and transforms.
Definitions are linked into the site-wide glossary and back to the full report.
Conversion of text-shaped pixels into machine-encoded text.
A textual replacement describing an image’s content or function.
A per-pixel transparency component that changes compositing and visibility.
Instruction-like text or features carried in pixels or image-associated semantics.
A model accepting more than one input modality, such as text and images.
Image scaling that can alter fine details and adversarial signals.
A secondary preview image stored in metadata, potentially retaining pre-edit content.
A disagreement among pixels, OCR, alternative text, metadata, or provenance signals.
These links are provided for visitors who want the governing specification, paper, framework, or implementation documentation. Links open in a new tab; the site does not fetch them during runtime analysis.
Accessibility success criteria.
Normative or first-party specification material.Adversarial images controlling generative models.
Original research or formal conference publication.Image container and chunk specification.
Normative or first-party specification material.Image metadata specification.
Normative or first-party specification material.Content provenance framework.
First-party guidance, framework, registry, or standards-program material.Choosing informative, functional, decorative, or complex-image treatment.
Tool, vendor, or implementation documentation; behavior is version-specific.Accessible image patterns and examples.
Tool, vendor, or implementation documentation; behavior is version-specific.OCR engine configuration and limitations.
Tool, vendor, or implementation documentation; behavior is version-specific.The full submitted report is preserved byte-for-byte in the governed research library and in durable repository documentation. The laboratory turns selected concepts into deterministic local output without external calls or hidden persistence.
A detailed analysis of pixels, alpha, compression, OCR, alternative text, metadata, QR-like symbols, multimodal prompt injection, adversarial images, provenance, safe ingestion, and accessibility.
Read governed reportInspect a PNG or JPEG, compare container metadata and transparency with visitor-supplied visible and alternative descriptions, and show bounded prepared OCR context without external vision services.
Open bounded laboratory