Authenticate

Use filters in the Overview category for a preliminary evaluation of your evidence image: perform a deep visual inspection, with levels adjustment, pixel-fidelity magnification, and side-by-side comparison with possible reference images. Briefly analyze the primary file format elements with Amped Authenticate's automatic warning computation system, helping you to immediately spot suspicious/uncommon metadata and image properties.

Take a deep dive into all image metadata and encoding properties. This category includes filters for simple and assisted bitstream examination, Exif and other metadata extraction, JPEG Quantization Table analysis, and comparison with Amped Authenticate's internal database (featuring more than 14k tables from thousands of camera models). You can also check whether the image was downloaded from social media platforms such as Facebook, Flickr, Instagram, Whatsapp, Twitter, and others.

Reconstructing the processing history of the image is essential for a thorough forensic analysis. Global Analysis filters detect traces of multiple JPEG compression, reveal the previous compression quality, expose artifacts introduced by resizing/rotation/digital zoom, unveil traces of histogram and color processing, and uncover periodic noise typical of images recaptured from LCDs. 

Every camera leaves a distinctive noise in the produced images. Amped Authenticate lets you estimate this noise using a few images from that camera, thus creating a Camera Reference Pattern (CRP) file. You can then use the stored CRP to understand whether an evidence image was captured using that specific camera exemplar, with high accuracy. All of this still works even when the image has been mildly compressed, rotated by multiples of 90 degrees, downscaled, or cropped. 

Being able to localize manipulated regions is of the utmost importance. Amped Authenticate provides you with many filters for this task, based on complementary analysis domains. Traces introduced by JPEG compression, artifacts left by interpolation algorithms, local noise consistency analysis, and suspicious similarity of cloned regions, are all used to provide you with effective, precise, and fine-grained forgery localization maps. 

Searches for artifacts introduced by AI-based image generation tools, such as faces generated by a GAN or images entirely synthesized from a text prompt by a diffusion model (e.g., Midjourney, Dall-E, Stable Diffusion).

One of the hardest parts of forging an image is to keep lights, shadows, and perspective consistent. Geometrical Analysis filters help the analyst check the technical consistency of such physical elements. Since they don't depend on compression or acquisition traces, these filters are very robust to post-processing and are suitable for low-quality images such as those from social media platforms. Geometrical Analysis filters can often be applied to scanned images and single video frames. 

Use the Visual Inspection filter of the Overview category for a preliminary evaluation of the unprocessed pixels of your video evidence, with the ability to adjust the levels for better frame analysis and zoom with pixel-fidelity magnification. Transfer a frame of interest to Amped Authenticate's Image Mode for a more in-depth analysis.

Consult the Advanced File Info panel to collect a variety of useful metadata and file structure information for your video file analysis. Check if the format details correspond to the device or software generating the evidence video, to assess whether it's camera original. Employ the frame hash computation to spot the presence of duplicates quickly. Perform a frame or GOP analysis to get advanced insights from your evidence.

Compare the evidence video against a reference PRNU CRP file, to verify if it was captured from a particular exemplar of camera or device. The proposed methodology, based on scientifically validated techniques, yields highly accurate results, provided that the video is not affected by digital stabilization.

Analyze the video to detect traces of double compression and get an estimation of the previous encoding's GOP size. Double encoding is a significant indicator in assessing the video's integrity as questionable. Examine macroblocks to evaluate the frame reliability and use motion vectors to analyze motion across the frames of your evidence video.

Examine the video for local modifications over time. The comparison of frames aims to reveal any unusual patterns or statistics (e.g. to detect cloned sequences or frames).
Exploit the sensor pattern noise of your evidence video to determine if certain portions originate from different devices.