Extracts GPS coordinates stored in EXIF metadata and visualizes the capture location on mapping services. This helps investigators quickly validate where a photo was taken and cross-check geographic claims. In forensic workflows, GPS evidence supports timeline reconstruction, travel path analysis, and correlation with other sources (CCTV, cell data, witness statements).
Similar Faces FAISS
Uses facial embeddings and fast nearest-neighbor search to locate visually similar faces across large evidence sets. This accelerates the identification of recurring individuals even when lighting, pose, or resolution changes. In forensic investigations, it supports suspect/victim recognition, cross-device linkage, and rapid triage of large seized collections.
Similar Images Embeddings
Computes global image embeddings to find visually related content across the dataset, including near-duplicates altered by resizing or recompression. This helps cluster evidence and reveal distribution patterns across folders, devices, or cases. Investigators use it to detect reused imagery, locate alternate versions, and quickly build coherent evidence groups.
Video Handling Frames
Extracts representative frames using intelligent sampling (≈10/50/90%) while preserving the original video file. This enables rapid screening without manually scrubbing hours of footage. In forensic practice, frame sampling speeds up triage, highlights key segments, and supports one-click export of originals for reporting or chain-of-custody workflows.
Restricted / NSFW Signals
Automatically flags potentially sensitive content using classification signals with adjustable thresholds and review flows. This helps prioritize items for examination while reducing unnecessary exposure during screening. In forensic environments, it improves triage speed, supports policy-based handling, and helps investigators focus on legally relevant material first.
Collections & Labels Categorization
Organize evidence into collections and apply labels to build a structured, searchable case view. Investigators can group items by subject, location, event, or investigative hypothesis and retrieve them instantly later. In real forensic work, consistent labeling improves collaboration, accelerates review, and supports clear courtroom-ready reporting.
Image View Analysis
Displays the original image inside the forensic workspace without altering pixel data, metadata, or color intent. This provides a reliable starting point for visual inspection before running analytical modules. In investigations, it is used to verify context, inspect fine details, and document what is visible in the original evidence file.
File Header Metadata
Shows the raw hexadecimal header and file segments with parsed markers and EXIF-related structures. This helps confirm that the file format is consistent with what it claims to be and exposes suspicious anomalies. Forensic analysts use header inspection to detect format spoofing, corruption, and traces of manual reconstruction or tampering.
Metadata Details Metadata
Presents EXIF/IPTC/XMP metadata in a structured table, including timestamps, device identifiers, software tags, and geolocation fields. This enables rapid validation of capture information and consistency checks across evidence sets. In forensic workflows, metadata supports timeline building, device attribution, and detection of post-processing or inconsistent provenance.
Thumbnail Check Metadata
Extracts embedded thumbnails and compares them to the full-resolution image using difference mapping. If the main image was edited after capture, the thumbnail may remain closer to the original, creating measurable inconsistencies. Investigators use this technique to flag post-capture manipulation and identify files that warrant deeper authenticity analysis.
Error Level Map Tampering
Visualizes compression error distribution to highlight regions that compress differently from the rest of the image. Areas with inconsistent error patterns may indicate localized editing, compositing, or selective recompression. In forensic practice, ELA is used as a fast triage indicator to guide further verification (clone detection, resampling checks, and metadata review).
Contrast Adjustment Tampering
Enhances local contrast and channel differences to reveal unnatural tonal transitions and potential selective edits. This can expose regions that were brightened, darkened, or blended to hide elements or alter interpretation. Forensic analysts use it to detect subtle manipulations intended to obscure objects, faces, text, or scene details.
Clone Detection Tampering
Detects copy-move manipulation by finding duplicated visual regions through keypoint matching and similarity thresholds. Repeated textures, patterns, or objects are highlighted to reveal likely cloning operations. In forensic investigations, clone detection helps identify removed items, duplicated backgrounds, and attempts to conceal evidence using patch duplication.
Filter Detection Tampering
Analyzes texture statistics and smoothing characteristics to identify filtered or stylized areas. Filters are often used to mask noise inconsistencies, blend edits, or hide compression traces. In forensic work, detecting filtering can reveal attempts to disguise manipulation or normalize edited regions to match the surrounding image.
Resampling Check Tampering
Uses frequency-domain evidence to detect artifacts introduced by resizing, rotation, skew correction, or other geometric transformations. Resampling creates periodic patterns that can indicate manipulation even when edits are visually subtle. Forensic analysts use this check to locate transformed regions and validate whether an image’s geometry matches a natural camera capture.
Double JPEG & Grid Tampering
Detects double-quantization artifacts and block grid misalignments that appear when a JPEG is saved more than once. These traces often suggest post-processing, editing, or recompression by software or platforms. In forensic analysis, double JPEG indicators help reconstruct processing history and flag files that may have undergone tampering.
JPEG Ghost Detection Tampering
Runs recompression sweeps to reveal “JPEG ghosts” — regions with a different compression history than the rest of the image. When elements are pasted or edited, their compression signature can diverge and become visible in the ghost map. Investigators use this to highlight suspicious areas for targeted follow-up with clone and resampling analysis.
Quantization & Huffman Check Tampering
Displays JPEG quantization and Huffman tables and scores anomalies against camera-like defaults. Non-standard tables can indicate re-encoding by editing software, messaging apps, or social platforms. In forensic workflows, this check helps assess authenticity claims and infer whether the file’s encoding matches the alleged capture device.
Social Media Detection Pipeline
Identifies processing fingerprints typical of social media upload pipelines, such as characteristic recompression and resizing patterns. This helps determine whether an image likely passed through a platform before acquisition. For investigators, it supports provenance analysis, distribution tracking, and validation of “original file” claims.
DeepFake Check AI/ML
Estimates the probability that facial content has been synthetically altered using machine learning signals and artifact detection. It focuses on inconsistencies typical of face swaps or AI-driven manipulation. In forensic investigations, this supports rapid screening of suspect media and helps prioritize advanced verification steps and expert review.
AI-Generated Check AI/ML
Predicts whether an image is likely AI-generated by analyzing coherence, texture statistics, and generative artifact patterns. This is useful when provenance is unknown or claims of authenticity are disputed. In forensic contexts, it supports triage and helps differentiate real captures from synthetic content during evidence screening.
PRNU Fingerprint Camera Attribution
Computes the Photo Response Non-Uniformity (PRNU) sensor pattern to support camera/device attribution. By comparing PRNU fingerprints, investigators can determine whether multiple images were captured by the same physical device. In casework, PRNU helps link evidence across devices, validate source claims, and strengthen attribution in forensic reporting; fingerprints can be exported as .fprnu for future comparisons.
LM Studio Vision AI AI Integration
Optional integration with a local LM Studio server enables AI-driven image analysis during indexing.
Using a configurable prompt, the model returns YES/NO responses that allow Media Insight to automatically
assign matching images to a chosen collection. The process is non-invasive and does not alter core
categorization logic — it only adds collection membership. Because analysis runs locally through an
OpenAI-compatible API, sensitive evidence never leaves the investigator’s system.
