Thank you for Subscribing to Media Entertainment Business Review Weekly Brief
This article is part of Media Entertainment Business Review Insights series featuring expert contributions nominated by our subscribers and reviewed by our editorial team.
COVID has impacted virtually every industry - media and entertainment is no exception. While the industry continues to see high demand for content and more aggressive production timelines, the pandemic has made it difficult to have large numbers of people on set (or even have real sets).
This has changed the way the industry is producing content and has accelerated the incorporation of animation and VFX into high quality productions. VFX shops are reaping the benefits and the demand for technology to accelerate workflows is increasing at a rapid rate.
For example, VFX companies are being called upon to: 1) execute traditional VFX work (ex. exploding a car or creating a monster) 2) create virtual sets containing people and other elements, and 3) complete nuanced work on things like set extensions (ex. adding ceilings or making rooms larger) that have become a staple of every show.
VFX is no longer limited to explosions and spaceships. It is now used to remove unwanted elements like cars, add signage, or even change the clothes people are wearing. This has created a strong need for VFX companies to reuse existing digital assets. Not just for reuse, but for the selection of looks. Searching and finding elements from past work allows new shows to reap the benefits of work done from previous shows.
This has highlighted the challenge of finding the right asset quickly at the right time, and within the end-user’s workflow. In many cases, VFX shops have stored hundreds of millions of assets over a long period of time using varying directory structures and naming schemes. Assets have been duplicated with different or misspelled names, and assets often have filenames that no longer make sense in the current pipeline.
What the artist needs is a way to search and find, directly within the workflow, the exact assets needed quickly. For example, they want to search for images or sequences or models, by resolution, by image orientation, by visual content, or by aspect ratio.
It has become almost impossible for artists to remember folder structures across many devices and locations. Finding files by their content is superior to finding by folder names and structure. You no longer need to remember where a file should be stored – you just search for files by their characteristics.
Additionally, a challenge arises with a geographically distributed workforce. It's easy to lose track of files as they move from home to office, between storage locations, or between offices and companies.
With large amounts of stored assets, remote production staff, reduced production timelines, and tight budgets, VFX companies have turned to the cloud for resources, and a combination of file and object storage for storage cost control. While the hybrid cloud is the ideal solution, its complexity has created a new long pole in the tent: finding content.
As the volume of content grows, stored in multiple locations and in both file and object-based storage, it has become a major challenge for studios to find what they need quickly. Today's complex scenes may contain tens of thousands of individual digital assets with complex relationships between these assets. Given this challenge, it is imperative that a search solution has the capability to search across the VFX shops’ entire storage network and include both files and objects. Equally important is to work with all storage types from all vendors, and on all operating systems. Studios are addressing this challenge through the automated creation of metadata catalogs utilizing AI/ML to both locate and catalog metadata, and also enhance the metadata in order to make it more easily searchable.
The basic building block in organizing unstructured data for easy, reliable, and quick search is the metadata catalog. A metadata catalog is simply a database of the data about the studio’s data. It structures the unstructured data’s attributes. Automating the creation of the metadata catalog becomes quite complex when considering the vast array of multimedia data. Simply extracting the basic metadata created by the data format provides only a limited amount of searchable information. Normalized and AI-augmented metadata is needed to enhance the quality of search results.
In order to build a metadata catalog that will accurately provide for the search results, various AI/ML algorithms must be deployed, and must be deployed using parallel processing techniques when considering petabytes of multimedia data.
Yadle continuously collects static, embedded, and behavioral metadata from files. The static metadata is details about the file itself – where it is stored, time and date information, etc. Embedded metadata is all metadata inside the actual file. Behavioral metadata is the data that explains where, when and by whom a file was created, modified, deleted, etc.
Yadle analyzes all collected metadata and augments it by generating normalized values to provide consistent fields across related but different file types. Yadle uses AI/ML analyze the actual content of files to generate tags for the specific visual and non-visual content in the file.
Yadle adds consistent, high-accuracy tags to files to simplify searching. Tags are added for visual content (#fire, #explosion), for file characteristics (#2k, #wide, #sequence, #stock) and for customer-specific elements (#lotr, #season1, #element)
Once the metadata catalog is created, a UI must be provided so the user can now search in an easy human-oriented way. The use of search filters helps in narrowing the search to specific media types.
Yadle File Channels allow the reorganization of related files without moving the files from their original location. This enables the aggregation of these files into virtual folders that include files from multiple storage and geographic locations into a single, searchable collection locations. The benefit is that all files for a specific show or episode can be searched as a single entity, regardless of where the files are actually located. File Channels can be created based on arbitrary needs, enabling the reorganization and “cleanup” of files without disturbing or changing the historic folder structure.
The Yadle app and plugins provide a simple “easy to get started” filter UI to allow users to find needed files quickly across one or more File Channels using tags, metadata, and normalized values. Today Yadle is supporting many of the most popular applications used by the industry including Adobe, Nuke, Maya, and USD.
In the highly competitive environment that the VFX industry is engaged, the time saved or lost by searching for content can be the difference between a making or missing a critical schedule.
This has changed the way the industry is producing content and has accelerated the incorporation of animation and VFX into high quality productions. VFX shops are reaping the benefits and the demand for technology to accelerate workflows is increasing at a rapid rate.
For example, VFX companies are being called upon to: 1) execute traditional VFX work (ex. exploding a car or creating a monster) 2) create virtual sets containing people and other elements, and 3) complete nuanced work on things like set extensions (ex. adding ceilings or making rooms larger) that have become a staple of every show.
VFX is no longer limited to explosions and spaceships. It is now used to remove unwanted elements like cars, add signage, or even change the clothes people are wearing. This has created a strong need for VFX companies to reuse existing digital assets. Not just for reuse, but for the selection of looks. Searching and finding elements from past work allows new shows to reap the benefits of work done from previous shows.
This has highlighted the challenge of finding the right asset quickly at the right time, and within the end-user’s workflow. In many cases, VFX shops have stored hundreds of millions of assets over a long period of time using varying directory structures and naming schemes. Assets have been duplicated with different or misspelled names, and assets often have filenames that no longer make sense in the current pipeline.
What the artist needs is a way to search and find, directly within the workflow, the exact assets needed quickly. For example, they want to search for images or sequences or models, by resolution, by image orientation, by visual content, or by aspect ratio.
It has become almost impossible for artists to remember folder structures across many devices and locations. Finding files by their content is superior to finding by folder names and structure. You no longer need to remember where a file should be stored – you just search for files by their characteristics.
Additionally, a challenge arises with a geographically distributed workforce. It's easy to lose track of files as they move from home to office, between storage locations, or between offices and companies.
With large amounts of stored assets, remote production staff, reduced production timelines, and tight budgets, VFX companies have turned to the cloud for resources, and a combination of file and object storage for storage cost control. While the hybrid cloud is the ideal solution, its complexity has created a new long pole in the tent: finding content.
As the volume of content grows, stored in multiple locations and in both file and object-based storage, it has become a major challenge for studios to find what they need quickly. Today's complex scenes may contain tens of thousands of individual digital assets with complex relationships between these assets. Given this challenge, it is imperative that a search solution has the capability to search across the VFX shops’ entire storage network and include both files and objects. Equally important is to work with all storage types from all vendors, and on all operating systems. Studios are addressing this challenge through the automated creation of metadata catalogs utilizing AI/ML to both locate and catalog metadata, and also enhance the metadata in order to make it more easily searchable.
The basic building block in organizing unstructured data for easy, reliable, and quick search is the metadata catalog. A metadata catalog is simply a database of the data about the studio’s data. It structures the unstructured data’s attributes. Automating the creation of the metadata catalog becomes quite complex when considering the vast array of multimedia data. Simply extracting the basic metadata created by the data format provides only a limited amount of searchable information. Normalized and AI-augmented metadata is needed to enhance the quality of search results.
In order to build a metadata catalog that will accurately provide for the search results, various AI/ML algorithms must be deployed, and must be deployed using parallel processing techniques when considering petabytes of multimedia data.
Yadle continuously collects static, embedded, and behavioral metadata from files. The static metadata is details about the file itself – where it is stored, time and date information, etc. Embedded metadata is all metadata inside the actual file. Behavioral metadata is the data that explains where, when and by whom a file was created, modified, deleted, etc.
Yadle analyzes all collected metadata and augments it by generating normalized values to provide consistent fields across related but different file types. Yadle uses AI/ML analyze the actual content of files to generate tags for the specific visual and non-visual content in the file.
Yadle adds consistent, high-accuracy tags to files to simplify searching. Tags are added for visual content (#fire, #explosion), for file characteristics (#2k, #wide, #sequence, #stock) and for customer-specific elements (#lotr, #season1, #element)
Once the metadata catalog is created, a UI must be provided so the user can now search in an easy human-oriented way. The use of search filters helps in narrowing the search to specific media types.
Yadle File Channels allow the reorganization of related files without moving the files from their original location. This enables the aggregation of these files into virtual folders that include files from multiple storage and geographic locations into a single, searchable collection locations. The benefit is that all files for a specific show or episode can be searched as a single entity, regardless of where the files are actually located. File Channels can be created based on arbitrary needs, enabling the reorganization and “cleanup” of files without disturbing or changing the historic folder structure.
The Yadle app and plugins provide a simple “easy to get started” filter UI to allow users to find needed files quickly across one or more File Channels using tags, metadata, and normalized values. Today Yadle is supporting many of the most popular applications used by the industry including Adobe, Nuke, Maya, and USD.
In the highly competitive environment that the VFX industry is engaged, the time saved or lost by searching for content can be the difference between a making or missing a critical schedule.
MORE FROM Innovation Insights
David Lundberg, Founder & CTO
Inqwizitiv
How Virtual VFX Teams Are Using Metadata Catalogs to Find and Manage Assets in a Hybrid Cloud Environment
EXPLORE OUR KNOWLEDGE NETWORK
The articles from these contributors are based on their personal expertise and viewpoints, and do not necessarily reflect the opinions of their employers or affiliated organizations.
