The Basics: Welcome to the GreyHen Reverse Image Search, a page designed to let you find an image on GreyHen's image database using only the image itself. GreyHen keeps track of a lot of information that lets us identify duplicate images, in order to keep the database free of extraneous files. By attempting to find images that are duplicates of a given image, we can identify files which visually match that image, similar to IQDB or Google Images (although not quite as potent). This search is performed using three different algorithms, which work together to find you the best match.
SHA-1: SHA-1 is a hashing algorithm created by the United States National Security Agency... in 1995. This algorithm may be a few months older than I am, but it runs quicker than MD5, the most common modern hashing algorithm, and takes up less space. If you get a match in this section, then it's almost certainly an exact file match - you got the image from the same source we did, or at least, our sources got it from the same source. However, if there are tiny differences between two images, or if different levels of JPEG compression were used to make them, then their SHA-1 hashes will be completely different - In these cases, we need to rely on two other algorithms.
GreyHash-14: GreyHash-14 is GreyHen's own image hashing algorithm, which checks to see if images are similar based on their width:height ratio and dominant colors. It only samples a handful of pixels, though, so it's not very accurate on its own. However, because it runs so quickly, and errs on the side of false positives, it's a good way of narrowing down results before slower algorithms take over.
GreyHash-B: GreyHash-B is less of a searching algorithm and more of an evaluation algorithm, which returns a percentage chance that two images are identical. This is a relatively slow algorithm, but it's good at paring down the large list of possibilities that GreyHash-14 returns. GreyHash-B isn't perfect, though; it has trouble with line drawings, where the vast majority of the image is the same color. Future developments may fix this problem, but they are not currently a priority.