What Amped Authenticate is

The product page describes tools on two levels. At the file level it reads Exif and other metadata, the file structure, the JPEG quantisation tables and other bitstream details. It also checks whether an image carries the traces of a social media platform. At the pixel level it looks for the history of processing, such as a second JPEG compression, resizing or histogram and colour changes. It also looks for local inconsistencies that point to a manipulated region. A Face GAN filter labels faces as GAN or Not GAN with a confidence score and further filters look for the artefacts of diffusion models. Camera identification uses sensor noise: a Camera Reference Pattern is built from known images of a device and compared with the evidence, and the maker says the match survives mild compression, rotation by ninety degrees, downscaling and cropping. A video mode adds file information, frame hashing, group of pictures and macroblock analysis, motion vectors and continuity checks, with a warning that digital stabilisation can affect camera matching. Batch processing, a command line interface, side-by-side comparison and reports that cite the reference for each filter round it off. Supported inputs include JPEG, TIFF, BMP, PNG, HEIF, common video formats, camera raw files and images embedded in PDFs.

Authentication is an opinion built from converging results, never a single filter output. One tool showing a double compression means the file was saved twice, which every social media upload also does. A Not GAN label is the output of a classifier trained on certain generators and says nothing about the next generator. A highlighted region is a hypothesis to test with other filters and with the file's metadata and provenance. Validation is the examiner's job: each filter is run on known genuine and known altered sets before it is relied on, and the limits found are stated in the report. Missing metadata is not proof of tampering and present metadata is not proof of origin, since both can be edited. The output of this software is evidence for an expert to weigh, not a verdict.

Primary applications

Where it is used

  • Checking whether an image or video file has been altered since capture
  • Reconstructing the processing history of a file, such as recompression or resizing
  • Linking an image or video to a specific camera through sensor noise
  • Screening faces for signs of generative manipulation
  • Examining metadata and file structure against the claimed source
  • Batch triage of many files before detailed examination

Who uses it

  • Image and video forensic examiners
  • Media verification teams
  • Expert witnesses in authenticity disputes
  • Researchers in multimedia forensics

Capabilities and limits

Major capabilities

  • Metadata, file structure and quantisation table analysis
  • Processing history detection, including double JPEG compression and resizing
  • Tampering localisation filters at the pixel level
  • Face GAN and diffusion artefact detection filters
  • Camera identification with sensor noise reference patterns
  • Video mode with frame hashing, group of pictures, macroblock, motion vector and continuity analysis
  • Batch processing and a command line interface
  • Side-by-side comparison and open source intelligence helpers such as reverse image search
  • Reports that cite the reference for each filter used

Technical limitations

  • No single filter proves manipulation. Conclusions rest on several results that agree
  • Deepfake detectors are trained on known generators and may miss newer ones
  • Social media recompression and resizing hide many pixel-level traces
  • Camera identification needs enough known images from the device and is affected by stabilisation in video
  • The examiner must validate each filter on known material and state its limits
  • The product page checked does not state an operating system, so none is listed here

Learning with DFRTI

Independent professional education covering the principles and workflows of Amped Authenticate. DFRTI is not a vendor training centre and its courses are not vendor certifications.

Before you learn it

  • JPEG compression, colour and sensor fundamentals
  • Statistics at the level needed to read confidence scores and error rates
  • Validation practice and report writing

Disciplines

Related tools

Courses that cover this tool

Product names and trademarks belong to their respective owners. DFRTI is an independent educational and research institute. Inclusion in this directory does not imply manufacturer endorsement, partnership or authorisation. Facts are checked against the manufacturer's public documentation on the date shown and may have changed since.

Go deeper with DFRTI

Free pages explain the ground. Structured courses, lab work and a mentor take you from knowing about it to doing it on a case.