Image Forensics Inspector

0 comments

Inspect a JPEG or PNG with this free image forensics tool. Review embedded metadata, compare recompression differences through error level analysis (ELA), and examine high-pass detail alongside the original. Everything runs locally in your browser, with no login, installation or upload of your image for analysis.

The results are inspection clues, not an authenticity verdict. This tool does not decide whether an image is real, fake, edited or AI-generated.

Inspect JPEG/PNG files locally (10 MB, 16 megapixels, 12,000 px per side). ELA compares this decoded image with a new JPEG compression: quality, texture, edges, colour conversion and prior saves all affect it. Noise view shows high-pass detail, including edges. Bright areas do not prove editing or splicing. Metadata can be absent, removed or altered; no real/fake verdict is made. Transparent pixels are composited on white for analysis.

Loading browser toolโ€ฆ JavaScript is required.

How to inspect an image

  1. Choose a JPEG or PNG. Select a local image within the 10 MB, 16-megapixel and 12,000-pixel-per-side limits.
  2. Set the inspection controls. Start with the default ELA JPEG quality and amplification settings.
  3. Click Inspect image. Compare the original, ELA recompression-difference view and high-pass detail view. An embedded thumbnail appears if one can be extracted.
  4. Review the metadata. Look through the available fields and any parsing warnings. Some images contain little or no embedded metadata.
  5. Adjust and compare. Change the quality or amplification, then inspect again to see how the view changes.
  6. Save the inspection output. Download the ELA PNG or the JSON report if you want to retain the results and settings.

What does the ELA view show?

In this tool, error level analysis compares the decoded image with a newly JPEG-compressed version. The ELA image shows absolute pixel differences between those versions, multiplied by the selected amplification.

The JPEG quality setting ranges from 0.50 to 0.99. Difference amplification ranges from 1 to 50. Increasing amplification makes small differences easier to see, but can also make ordinary differences appear dramatic.

For example, if a textured area looks bright, inspect it in the original and try another quality setting. Texture, edges, colour conversion, earlier saves and the selected recompression settings can all affect the result. Brightness alone does not prove that an area was pasted in or edited.

The same calculation is available for PNG input, but it compares that decoded PNG with a newly created JPEG. It does not reveal a hidden JPEG editing history.

The displayed mean recompression difference is a pixel-difference summary on a 0โ€“255 scale. It is not an authenticity score, a probability of manipulation or a measure of how trustworthy the picture is.

What does high-pass detail show?

The high-pass view highlights local brightness differences, including edges and fine detail. Its amplification control ranges from 1 to 20.

This is not a sensor-noise identification test. A strong outline or a patch of visible texture can appear because of the image content itself. Compare it with the original and use it to decide where to look more closely, rather than treating a bright patch as a confirmed anomaly.

Read metadata and embedded thumbnails

The tool attempts to read embedded EXIF, XMP and IPTC metadata, including GPS fields when present. Depending on the file, those fields may describe a camera, software, capture settings or other recorded details. The tool shows what it can parse rather than filling in missing information.

Metadata can be absent, removed or changed. A missing camera model does not establish that a picture is fake, and a software field does not explain every step in its history. Review the fields in context with what you know about the file’s source.

If the image contains an extractable thumbnail, it is displayed as another inspection view. A missing or unreadable thumbnail is reported where applicable; the tool does not generate a substitute and present it as embedded evidence.

Downloads, transparency and limits

The ELA download is ela-inspection.png. The JSON report includes the image dimensions, selected quality and amplification settings, mean recompression difference, parsed metadata and an inspection-limitations note. The high-pass image is displayed for comparison, but has no separate download button.

Transparent pixels are composited onto white for analysis. Bear that in mind when inspecting PNGs with transparent backgrounds or partly transparent edges.

Use a valid JPEG or PNG no larger than 10 MB and 16 megapixels, with neither side above 12,000 pixels. Files must satisfy all three limits. Unsupported or damaged images may fail to decode or return incomplete metadata.

Use the results as part of a wider check

Keep the original file and record which version you inspected. A screenshot, resized copy or repeatedly saved image may give different inspection results from the source file.

This browser inspector helps you examine available file details and visual differences. It does not verify the image’s origin, prove a particular edit, detect AI generation or replace corroborating information about where the picture came from.

Related Articles

Wordsearch Generator

0 comments

Wordsearch puzzles are a fun way to build vocabulary, improve pattern recognition and challenge the brain. Whether you…

Leave a reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.