[10:42 Fri,4.June 2021 by Thomas Richter]
Since our last News there have been some interesting changes to Flim, the AI search for movie stills, which we would like to summarize for you here. The inventory of searchable stills from movies, documentaries, TV series and animations has been expanded to include images from the areas of advertising as well as music videos - a total of over 300,000 movie stills are now available.
More comprehensive search.
Particularly exciting, however, is the upgrade of the search to an even better deep-learning model for searching: while Flim previously worked with rather simple AI-generated tags for movie stills, the new version now also allows more abstract visual searches, for example, for the artistic style and general mood of an image as well as its atmosphere and composition, and not just of image objects.
So now film stills can be searched for criteria as diverse as a concept (e.g., a minimalist landscape), an action (e.g., lighting a cigarette), typography, by beta.flim.ai/? ft=nudity (nude scenes), by multiple words (such as red lips), geography (a street in Paris), by context (such as man on a park bench), or descriptions of a mood (such as the feeling when you are alone).
In addition, if no image matching the search exactly is found, a set of results matching the mood and atmosphere of the search will be suggested from now on.
Freemium instead of Free.
In order to finance itself (and further development), Flim is now switching to a freemium model, meaning that basic features can continue to be used for free, but more advanced features will require payment. In the case of Flim, this means that while stills can still be searched and browsed, non-paying users can only view a limited number of images in full (HD) resolution. A premium account with unlimited use costs around per month.
Daily limit reached
What is Flim?
Flim is a search engine equipped with DeepLearning features for movie stills from many well-known (and unknown) movies. So far you could already search for classical categories like the name of a movie, the director, the DOP, the actors, the genre, ratio or year of release, and by AI (version 1) already for objects appearing in the pictures (like angels, clocks, cabs..) or the environment of a scene (for example desert, city or forest). With the new version, the possibilities of intelligent searches have been extended significantly (see above).
In addition, an AI color analysis can be called up for each image, which shows which colors were used in a scene - likewise, images can be searched for in which a selected color occurs dominantly. The search for objects can also be narrowed down to a specific film (or films by a director), such as cars in films by Quentin Tarantino.
Lighting a cigarette
The project is still in beta and is based on the fair use principle, i.e. you are allowed to use the movie stills (which are in HD resolution) for educational or personal purposes only (such as inspiration for mood boards), but never commercially. Rights holders can contact us to remove images of their films from the search.
more infos at bei beta.flim.ai
deutsche Version dieser Seite: Flim: KI-basierte Filmstillsuche bekommt Intelligenz-Upgrade