Wim Van Lancker has been active at the Multimedia Lab at Ghent University, dealing with various media-related issues including encoding and delivery techniques. His current interest goes out to scalable video coding technologies, media content adaptation and delivery, video quality, video streaming, media content presentation, Semantic Web technologies and multichannel publishing.

Karel Braeckman worked at VRT-medialab, playing a key role in the research and development of prototypes relating to innovative media workflows, managing media material and its associated metadata, and semantic metadata. His main interests include the management of media collections, with a focus on optimizing the search and navigation within large media collections.

Their competences will definitely be useful in NordicLight, a European project in which Limecraft will be investing a lot of energy for the next three years.

Limecraft now counts six media experts. The whole team wishes you a happy and succesful new year!

Despite the proliferation of user-friendly technology, the main challenge today for occasional and institutional producers remains the management of vast amounts of content, and in particular the retrieval of items of in large databases. In a world where digital content is ubiquitous, the ability to make sense of everyday information is key. 

During this project, we will set up an online application that is capable of supporting the different actors throughout the production processes of audiovisual media. The production platform will manage all media and their associated meta-data. The primary goal is to reduce the cycle time of the production by avoiding manual processing of information. Then again, based on a careful selection of open standards, we intend to deploy a system that is genuinely designed for collaborative production and geographically distributed operations. 

NordicLight will be built on Limecraft’s existing search technology. The core of this technology is an indexing system optimised for disclosure and re-use of audiovisual material. Due to its ability to understand the contents of the media assets, unrivalled retrieval rates in archives and production systems can be achieved. Based on this technology, Limecraft and Wappa will create a number of turn-key, cost-effective and time-saving applications that together support the production process end-to-end. In essence, by forming a conceptual and contextual understanding of any piece of audiovisual material, irrespective of data type and storage location, we will enable organisations to connect their information silos and make the most of their assets. Through meaningful computing, we will be able to automatically analyse, collect, organise and adapt assets according to the specific requirements of any particular production process, thus re-defining the meaning of Media Asset Management.

Limecraft is the coordinator of NordicLight and is responsible for the development of the production applications. As an independent solution provider, Limecraft delivers software infrastructure that automates complex operations on audiovisual material. Limecraft spans the full spectrum of mission-critical applications, including editorial, production, post-production and distribution.

Wappa is a Romania-based software development company specialised in complex back-end application logic. NordicLight will allow Wappa to strengthen its position in the media sector, by extending its portfolio with a content management framework, primarily addressing the media sector but eventually expanding to other sectors as well.

Professionals in the audiovisual field will help to define the functional requirements and will provide content. They will also test the robustness of the system and the user-friendliness and intuitiveness of the workflows and interfaces. Raindance reaches scriptwriters, producers, cameramen, sound engineers, production assistants and directors. Their members are active in every genre possible (documentary, feature film, shorts, animation and experimental films). Organised with local branches around Europe, their activities will allow us to fully test the collaborative aspect of our system and to become familiar with cultural or linguistic differences in production processes. Taglicht Media is a production company specialised in high profile feature documentaries. They represent the traditional production and distribution process, yet they have realised the importance of new media technology.
 
NordicLight is partially funded by the MEDIA programme. MEDIA is an instrument of the Executive Agency responsible for Education, Audiovisual and Cultural matter. It has been set up to support the European audiovisual sector. The specific objective of new technology pilots is to ensure that the latest technologies are developed into marketable solutions, and are thus made available to the European audiovisual sector, and in particular to strengthen the European cultural tissue.

Wappa - http://www.wappa.ro
Raindance - http://www.raindance.co.uk
Taglicht Media - http://www.taglichtmedia.de
Limecraft - http://www.limecraft.com

View the press release (Dutch)

As you might know, we’re building an end-to-end file-based production platform from scratch. This is good, because there’s no legacy code to drag along, no arcane product integrations to migrate or support, etc. On the other hand, building from scratch also means there’s loads of work to be done and we need to move forward as quickly as possible, requiring us to be as productive as possible.

The Material eXchange Format (or MXF for short) is the leading professional file format for the distribution of audio and video, as opposed to consumer-oriented formats such as the ISO MPEG-4 (mp4) or Matroska(mkv/webm) formats. MXF files are typically obtained from file-based camcorders at the beginning of the production process after which they are used throughout the production workflow: during editing, in media asset management systems, in play-out servers and for archival purposes. Additionally, MXF files contain a lot of valuable metadata about the audio and video material. This includes structural metadata (resolution, bitrates, frame rates, duration) but also descriptive metadata, as in: for which production this MXF file was created, who the director, producer, DOP, … were, what the recording camera properties were, etc.

Whenever we ingest an MXF file we also want to extract whatever metadata we can from it, as it might come in handy afterwards. Additionally, whenever we publish an MXF file for further processing outside of our platform, we want to extend the existing metadata set with information that we have gathered ourselves, e.g., object detection information, or a change log of changes that the file has gone through while being processed by our system.
Because MXF is an essence container, and is optimized for fast reading and writing operations, most software that deals with MXF files does this at a low level [1]. As such and unfortunately, obtaining high-level metadata from MXF files often means using toolkits written in C/C++ (e.g., the MXF::SDK, MXFTk, ...) that involve lots compilation and code verbosity in their use. Wouldn’t it be great if we could manipulate MXF files using a higher-level and more convenient language, like Python, Groovy/Scala or Ruby?

Instead of implementing an entirely new MXF manipulation toolkit, we’ve decided to wrap an existing one (with its various optimizations) and expose its features to a higher-level language using SWIG. SWIG is a tool that generates wrapper code to bridge from the original library to the target language, using an interface definition in which the functions and objects that must be exported are listed (shown below). The library we’re wrapping is the open source mxflib toolkit and we’re currently working with Ruby as the target language.

To give you an idea of what MXF manipulations look like using Ruby, next are a couple of simple examples. First of all, code that was simple in the C/C++ implementation remains simple in Ruby:

mxfFile->Open("1220444800140_123_C0004.MXF", true);

becomes:

mxf.Open("1220444800140_123_C0004.MXF", true)

Additionally, due to the dynamic nature of Ruby, we can declare and use variables without type definitions:

masterPart = mxf.ReadMasterPartition()
masterPart.ReadMetadata() unless masterPart.nil?

We don’t need to declare a type for masterPart, we can just call the methods that are also part of the original C++ object. Since the SWIG wrapper declares Ruby classes to represent the original classes, the Ruby runtime handles methods lookups as if object called was written in Ruby. Note too that the mxflib works with smart pointers for memory management. SWIG recognizes smart pointers and indirects them transparently (nil? also works with smart pointers). Then, whenever the Ruby reference to the wrapped object is garbage collected, the referenced smart pointer is destroyed and as a consequence, the referenced C++ object gets deleted.
SWIG also allows developers to define custom mappings between data types of the original library and the target language. Take for example the Universal Label, which is a 16-byte identifier used throughout the MXF specification to uniquely identify metadata elements. The UL for sample rate values is defined as follows in C++, using a constructor of the UL object that takes an array of UInt8 elements as its argument:

const UInt8 SampleRate_UL_Data[16] = ↩
{ 0x06, 0x0e, 0x2b, 0x34, 0x01, 0x01, 0x01, 0x01, ↩
    0x04, 0x06, 0x01, 0x01, 0x00, 0x00, 0x00, 0x00 };
const UL SampleRate_UL(SampleRate_UL_Data);

Using an appropriate SWIG typemap, we can do something similar for our wrapper and copy the Ruby array into a C array and then pass this array to the created UL object. The resulting Ruby declaration for the SampleRate UL then looks similarly convenient (and can be used as the argument of functions, as shown in the next example):

SampleRate_UL = Mxflib::UL.new( ↩
    [0x06, 0x0e, 0x2b, 0x34, 0x01, 0x01, 0x01, 0x01, 0x04, 0x06,  ↩
        0x01, 0x01, 0x00, 0x00, 0x00, 0x00])

Additionally, SWIG takes advantages of target language constructs and syntactic sugar to increase productivity. Using C++ STL, iterators can be used to traverse a list of elements:

MDObjectULList::iterator itPackages = Packages->begin();
while (itPackages!=Packages->end()) {
    MDObjectPtr Package = (*itPackages).second>GetLink();
    if (Package->IsA(SourcePackage_UL)) {
        // do things here…
    }
    itPackages++;
} 

This is fine in C++, but way too verbose and inelegant for Ruby. Luckily, C++ objects that implement a std::list or std::vector can be mapped to corresponding Ruby objects to take advantage of Ruby blocks:

Packages.collect { |p| p.second.GetLink() }
        .select  { |p| p.IsA(SourcePackage_UL) }
        .each {
            // do things here…
        }

This piece of code is much simpler and straightforward, and illustrates that the wrapped library is seamlessly incorporated into the Ruby code. The collect() and select() functions work just as they would in a native Ruby application which makes the code in question more declarative and easier to read.

The work on the wrapper interface is currently in its early stages and hasn’t been tested thoroughly. Give us a few more weeks and we’ll be publishing the wrapper and the accompanying modifications made to mxflib. We’ll also provide a couple of comprehensive MXF manipulation examples written in Ruby.

[1] In addition to the file format definition, SMPTE has also created the SMPTE 434 standard which defines a set of XML schema’s that can represent the metadata of an MXF file in a lossless way. I.e., the metadata can be extracted as an XML document, be modified and then be written back to the original file without loss of information, given a compliant decoder/encoder pair.

“The increased cost of online television has nothing to do with the cost of connectivity: distribution of radio and television over IP is often much cheaper compared to conventional distribution technology.  The real cost is caused by a lack of standards, unchallenged technology and inadequate production methods.”

Limecraft opts for an offensive strategy and is ready to take up the challenge and to develop the best and the most cost-effective system on the market.

Therefore we congratulate IBBT iCubes services, for their excellent care and we thank them for giving us the opportunity of a smooth start.

The new Limecraft office space on a peaceful Sunday afternoon.

Our conclusions so far are:  the market opportunities are gigantic, it will take us some time to develop the right tools and we will initially focus on a limited number of value-added assets.

An excellent setting and a productive meeting enabled us to outline the priorities in terms of our product portfolio and to address the main marketing requirements, on which we will craft our website. The good news is: there is a consensus! The bad news: we still have a lot of work. However this is also good news.

Nico did his very best to incorporate a mix of changes in the website. Despite several liters of coffee and Nico’s stamina of a trained long-distance runner, the effects of our marathon meeting are showing as you can see in the picture. Managing a startup is not for sissies. ]]> 2010-12-12T14:02:44+00:00