Ashen is currently available at the address source.animeshon.com.
Ashen is an image and video reverse search engine built on top of perceptual hashing technologies. Its implementation is a rewritten and optimized version of the search engine in use by trace.moe. Ashen provides the core functionalities in use by Oracolo for image and video searches.
Ashen is currently providing core functionalities for image and video search only for
Further development in other domains such as
Fan-Arts is planned for future releases. For
Doujinshi reverse image search please visit Ashen Grey.
Ashen development state is shown in the Ashen Roadmap.
The Multimedia Community is currently relying on algorithms which, given an image or video file, generates a static signature (also known as hash or checksum) based on the content of the file itself. Corruption or changes of a single bit will result in a completely different signature and therefore the two files will be perceived as completely different. This might lead to a situation where lower quality images or videos are no longer recognized by indexing systems even if the content is perceptually still the same.
Ashen is built on top of a very different way of generating signatures for image and video files which relies on perceptual hashing and locality-sensitive hashing algorithms. This allows us to recognize the content of an image or video even if small artifacts (e.g. subtitles, adult content pixelization) are present, the pixel mapping is different (e.g. compression, scaling, different quality), the image is softly cropped or has extra borders. Ashen is therefore capable of returning metadata information for both original and modified content.
Ashen is the right product if you want to answer one of the following questions:
- Which Anime (or episode) was this screenshot taken from?
- Which Anime (or episode) was this video clip taken from?
- Which Anime (or episodes) are depicted in this AMV (Anime Music Video)?
- I have a collection of video files, can I group them by Anime?
- I have a large number of images / videos, can I extract their metadata in a quick way (>10 req/s)?
- I have a streaming service, can I prevent copyrighted material from being uploaded there?
Ashen might not be the best choice for the following queries:
- I have this heavily edited image, what Anime (or episode) was it taken from?
- Try trace.moe if image lookup fails on Ashen
- I have this meme (e.g. a cropped image with text), what Anime (or episode) was it taken from?
Ashen will support the following use-cases in future versions:
- Where was this fanart taken from?
- Use saucenao.com for the moment
- Which Anime / character(s) are depicted in this fanart? Who is its author?
- Where was this cosplay taken from?
- Which Anime / character(s) are represented in this cosplay? Who is the cosplayer / photographer?
- Which Manga / Doujinshi / Light Novel / Visual Novel was this picture taken from?
- For Manga, Light Novels, and Doujinshi please refer to Ashen Grey
Get started with Ashen functionalities and try out a new way of getting sources.
Stay up to date with Ashen development and new releases.
Integrate your service through the Ashen REST APIs.
Enjoy a rich set of tutorials and examples.
Integrate with Ashen using our official SDK.