Baraka is a 1992 American non-narrative documentary film directed by Ron Fricke and produced by Mark Magidson.

The word Baraka is defined within various Eastern religions as a powerful spiritual blessing that flows throughout many different aspects of life.

The Baraka project was filmed by a only 5 person crew, over a period of 14 months, in 24 different countries, and while spanning 6 of the 7 continents.

This is a short documentary released in 2008 about the 8K scanning and restoration process undertaken with the original film stock

Baraka was shot on 70mm film and then scanned at an oversampled 8K Ultra High Resolution (8,192 pixels across the frame) with DTS-HD Master Audio 5.1 @96k/24bits. It was then output onto Blu-ray in Full High Definition 1080p at a 16:9 Widescreen aspect ratio.

The scan done at 8K was oversampled, which means the film was scanned at a higher resolution than what was necessary for the final 1080p output that they intended and produced.

The purpose of scanning at 8K though was to create a super sharp, but down-sampled version of the film scan once it was output at Full HD 1080p. Thus, the resulting image from the 8K scan is a sharper quality 1080p output than if they had just scanned the film at 1080p or at a 4K scan resolution. The 1080p output from the 8K scan is also now perhaps an even better quality version than the original film print when viewed digitally on either a 1080p or 4K display.

The unanswered question though is if an 8K scan from the 65mm native film stock would produce an even more desirable looking image at 4K than at the 1080p resolution output which they had decided upon for this production.

But without there being any 4K sample outputs available (AFAIK) from the original film print to 8K scan (in order to rule out whether or not a 4K output might introduce any undesirable flaws, artifacts, or reveal any weaknesses from the physical film print at such a high resolution scan) it remains a question that still can’t be fully addressed.

In this video they also briefly mention DVD, which may be a bit confusing, but I can confirm that the 8K scan was never output at 480p resolution and was only ever released at 1080p resolution on Blu-ray.

One can still purchase the full Baraka 1080p film on Blu-ray from the 8K scan, which also includes this short documentary titled "Restoration" as a special feature on the disc. The Blu-ray disc also includes another 1+ hour documentary titled "A Closer Look” which is a BTS documentary on the shooting and production of Baraka itself.

Thus, the full 1080p version of the Baraka film, the "A Closer Look" documentary, and this short "Restoration" documentary are all included together on a single Blu-ray disc which is available from the link here: https://amzn.to/2Ii5na4

A final note in that I do not possess any copyrights to the content of this video nor is it my intention to infringe upon any existing copyrights of this content in any way. I also have no personal connection or otherwise to the producers of this video or to the creators of the Baraka film.

Certain patterns are innately hierarchical, like the underlying parse tree of a natural language sentence. A Recursive Neural Tensor Network (RNTN) is a powerful tool for deciphering and labelling these types of patterns.

Deep Learning TV on
Facebook: https://www.facebook.com/DeepLearningTV/
Twitter: https://twitter.com/deeplearningtv

The RNTN was conceived by Richard Socher in order to address a key problem of current sentiment analysis techniques – double negatives being treated as negatives. Structurally, an RNTN is a binary tree with three nodes: a root and two leaves. The root and leaf nodes are not neurons, but instead they are groups of neurons – the more complicated the input data the more neurons are required. As expected, the root group connects to each leaf group, but the leaf groups do not share a connection with each other. Despite the simple structure of the net, an RNTN is capable of extracting deep, complex patterns out of a set of data.

An RNTN detects patterns through a recursive process. In a sentence-parsing application where the objective is to identify the grammatical elements in a sentence (like a noun phrase or a verb phrase, for example), the first and second words are initially converted into an ordered set of numbers known as a vector. The conversion method is highly technical, but the numerical values in the vector indicate how closely related the words are to each other compared to other words in the vocabulary.

Once the vectors for the first and second word are formed, they are fed into the left and right leaf groups respectively. The root group outputs, among other things, a vector representation of the current parse. The net then feeds this vector back into one of the leaf groups and, recursively, feeds different combinations of the remaining words into the other leaf group. It is through this process that the net is able to analyze every possible syntactic parse. If during the recursion the net runs out of input, the current parse is scored and compared to the previously discovered parses. The one with the highest score is considered to be the optimal parse or grammatical structure, and it is delivered as the final output.

After determining the optimal parse, the net backtracks to figure out the appropriate labels to apply to each substructure; in this case, substructures could be noun phrases, verb phrases, prepositional phrases, and so on.

RNTNs are used in Natural Language Processing for both sentiment analysis and syntactic parsing. They can also be used in scene parsing to identify different parts of an image.

Have you ever worked with data where the underlying patterns were hierarchical? Please comment and let us know what you learned.

Credits
Nickey Pickorita (YouTube art) -
https://www.upwork.com/freelan....cers/~0147b8991909b2
Isabel Descutner (Voice) -
https://www.youtube.com/user/IsabelDescutner
Dan Partynski (Copy Editing) -
https://www.linkedin.com/in/danielpartynski
Jagannath Rajagopal (Creator, Producer and Director) -
https://ca.linkedin.com/in/jagannathrajagopal