1. Can image search identify an actress?
Can an AV actress be identified through image search?
If the image quality is sufficient to make out the face, AI can compare facial features against an actress database and present candidates even when no identical image exists on the web. Unlike general-purpose similar-image search, this answers who the person in the image is. Results present candidates and do not definitively establish identity.
Site search covers Japanese-language content; enter queries in Japanese.
2. Why AV titles can be hard to find with general-purpose image search
General-purpose similar-image searches such as Google Images find identical or similar images published on the web. If the image you want to search for appears on the web in exactly the same form, you may find the page that published it. However, the following situations can prevent you from reaching the title or actress you are looking for.
- The image has a different composition or crop from images circulating on the web, as with screenshots or cropped images.
- The search settings or service design exclude adult content from search results.
- Even if you find the cover image, the page does not list the featured actress's name.
General-purpose image search looks for where the same image appears; it is not designed to answer who the person in the image is. This distinction is the main reason image searches for AV titles can reach a dead end.
3. Image search using facial recognition as an alternative
Rather than matching the image itself, this site's image search compares the facial features in the image against an actress database. This means it can present candidates even for screenshots not circulating on the web or photographs you took of a screen, provided the image quality is sufficient to make out the face. The ability to search without an identical image existing elsewhere is the biggest difference from general-purpose similar-image search.
Supported formats are JPG / PNG / WebP / HEIC / HEIF, with a file size limit of 5MiB at the time of submission. Even if an image exceeds 5MiB, supported browsers automatically optimize it before submission as long as the original file is no larger than 25MiB (files over 25MiB are not eligible. Also, many environments cannot load HEIC / HEIF images and therefore cannot optimize them, so please convert these files to JPEG or PNG if they exceed 5MiB).
4. What makes an image suitable for image search?
Because the system uses faces as clues, results vary greatly depending on how the face of the person you are looking for appears in the image.
- The face is large, clear, and viewed at an angle close to front-facing.
- Most of the face is not obscured by mosaic censoring or text overlays.
- If several people appear in the image, the face of the person you are looking for is clearly distinguishable.
If you cannot provide an image that meets these conditions, try a different shot. If a product code is printed on the cover, looking it up by product code is more reliable.
5. Search coverage
This site includes titles across FANZA and MGS. From the search results, you can follow links to candidate actresses' pages and lists of the titles they appear in. If you want to search by actress name, you can also use the search box and the Japanese syllabary listings on the home page.
The only inputs accepted for recognition are an image file and the URL of a publicly available image. A title page URL (a product page URL) cannot be submitted because it is not an image itself. A product code (title code) is not an input for recognition either; use How to look up titles and performers by product code to search the site instead.
6. Frequently asked questions
Why can Google Images fail to identify an AV actress?
General-purpose similar-image search is designed to find where the same image appears on the web, not to answer who the person in the image is. It reaches a dead end when the composition differs from images on the web, as with screenshots or cropped images, or when the source page does not list the actress's name.
How is image search using facial recognition different?
Instead of matching the image itself, it uses facial features in the image as clues to compare against an actress database. This means it can present candidates even from your own screenshots or photographs of a screen that do not have an identical image on the web, as long as the image quality is sufficient to make out the face.
What makes an image suitable for image search?
There are three requirements: the face should be large, clear, and close to front-facing; most of the face should not be obscured by mosaic censoring or text overlays; and if several people appear in the image, the face of the person you are looking for should be clearly distinguishable.
Which titles are covered by the search?
The site includes titles across FANZA (Video and Amateur) and MGS動画. From search results, you can follow links to candidate actresses' pages and lists of the titles they appear in.
7. Related pages
For how recognition works and how to read the similarity scores in the results, see How AV actress face recognition works. For when to use the different recognition methods available on this site (Web, LINE, and the browser extension), see Ways to identify an AV actress from an image. For detailed operating instructions, see the Actress Recognition AI User Guide.