{"id":221893,"date":"2022-02-09T16:02:00","date_gmt":"2022-02-09T15:02:00","guid":{"rendered":"https:\/\/www.cined.com\/an-introduction-to-arriraw\/"},"modified":"2022-10-06T15:15:07","modified_gmt":"2022-10-06T13:15:07","slug":"an-introduction-to-arriraw","status":"publish","type":"post","link":"https:\/\/staging.cined.dev\/jp\/an-introduction-to-arriraw\/","title":{"rendered":"ARRIRAW\u5165\u9580"},"content":{"rendered":"\n

To understand how ARRIRAW works and what makes it different from other raw formats, it helps to go straight to the source.<\/strong><\/p>\n\n

In ARRI Academy\u2019s official Certified Online Training for Camera Systems course<\/a>, which is only available online at MZed<\/a>, ARRI\u2019s senior trainer Florian Rettich distills the complicated subject of raw recording in a way we can all understand.<\/p>\n\n

Even if you\u2019re not shooting with ARRI cameras, this lesson is an enlightening dive into camera sensors and photosites, bit depth, data compression, and color. There\u2019s a lot to learn and the knowledge will help you grasp the technology inside cameras today.<\/p>\n\n

ARRIRAW – Photosites, sensors, and analog to digital converters<\/h2>\n\n

To better explain ARRIRAW instructor Florian first gives a basic introduction to photosites and how they react to changes in light. You may already know this to some degree, but it\u2019s still a good refresher. The way Florian explains the digital imaging process is easy to understand for those of us who aren\u2019t already seasoned pixel peepers.<\/p>\n\n

So, all digital cameras have sensors that convert light into an electric charge, which then gets digitized and processed into data. But the way they go about that process differs.<\/p>\n\n

\"\"
Image source: ARRI<\/figcaption><\/figure>\n\n

First let\u2019s only think about light itself, before we add color to the mix. If you think of photosites on a sensor as buckets of light, then each bucket is white when it is completely full of light, and black when it receives no light. You could also use the analogy of a glass of water if that helps you picture what\u2019s happening.<\/p>\n\n

At some point the bucket is full, and the photosite is 100% saturated, or white, disregarding any additional light energy spilling over the bucket. And the same is true when there\u2019s no light and the photosite becomes 0% saturated, or black.<\/p>\n\n

\"arriraw-photosites-optical-electric-digital-conversion\"
Image source: ARRI<\/figcaption><\/figure>\n\n

Now if you have a 100% full bucket, and you stop the iris down by one stop, the bucket becomes only 50% full. If you stop down the iris an additional stop, the bucket is now 25% full.\u00a0<\/p>\n\n

The output of the photosite is a little analog charge that needs to be amplified before it gets digitized via an analog to digital (A\/D) converter. And at each stop down, the sensor is sending half the amount of analog energy to the digital converter.<\/p>\n\n

How ARRI gets to 14 stops of dynamic range<\/h2>\n\n

So if the camera is looking at a white cloud on a bright sunny day, stopping the lens down a stop from a clipping white cloud would result in a still very white cloud, but with the photosite bucket now at half full.\u00a0<\/p>\n\n

Florian goes on to show a linear graph where at each stop down of light, the bucket\u2019s signal gets dropped in half, and the output or code value is reduced linearly. With a 10-bit signal, there are only 1024 code values, so splitting it in half from 1024 to 512 to 256 and so on would result in only 9 stops of light until you reach 2 code values.\u00a0<\/p>\n\n

By adding two more bits and using a 12-bit A\/D converter, you can now quadruple the code values, from 1024 to 4096, but that still only provides 11 stops of light. (To do the math, just start with the number 2 and multiply it by 2 – you can only do that 11 times before reaching 4096.)<\/p>\n\n

\"arri-raw-12-bit-data-values\"
Image source: ARRI<\/figcaption><\/figure>\n\n

So how do ARRI cameras maintain 14 stops of dynamic range<\/a>? In a nutshell, they need to start with a 16-bit A\/D converter, but when the ALEV sensor was first designed that was difficult to acquire.<\/p>\n\n

As a solution, what the engineers did with the original ARRI ALEV sensor was use two A\/D converters, which were then mapped to a 16-bit linear signal. So now we have over 65,000 code values.\u00a0<\/p>\n\n

This is the foundation of the Dual Gain Architecture (DGA). One of the A\/D converters is looking at the highlights of the bucket, and the other at the shadows, with some overlap in the middle.\u00a0<\/p>\n\n

\"arri-arriraw-alev-sensor-14-bit-dual-gain\"
Image source: ARRI<\/figcaption><\/figure>\n\n

How ARRIRAW stays uncompressed<\/h2>\n\n

Now that there\u2019s a robust digital signal available, how do you take 65,000 code values per frame and fit all of that data onto a media card or drive? How do you reduce the amount of data without compromising the image quality?<\/p>\n\n

There are a few different ways to compress the data into a raw format, and most manufacturers have varying degrees of compression and style to minimize the effect on the image. For example, there\u2019s DCT (Discrete Cosine Transform) which works with square blocks of pixels, and there\u2019s Wavelet which is processor-heavy but takes a different approach to compress the raw image.\u00a0<\/p>\n\n

ARRIRAW, on the other hand, is able to record an uncompressed image, while reducing the amount of data, by using a logarithmic allocation of data rather than linear. Florian explains that ARRI looked at their history with film, which is not linear, to figure out how to translate that photometric approach into a digital output.<\/p>\n\n

\"arriraw-uncompressed-16-bit\"
Image source: ARRI<\/figcaption><\/figure>\n\n

So with 65,000 code values available in a 16-bit signal, a linear process would allocate one stop of light to half of all the code values – essentially over 32,500 code values for the difference between two shades of white on a white cloud. ARRIRAW, on the other hand, gives each stop of light an equal amount of code values.<\/p>\n\n

So the highlights aren\u2019t treated differently than the shadows. Each stop receives the same amount of code values, which makes the raw file more efficient at reducing data. And it\u2019s what enables ARRIRAW to save a raw image to media cards and drives, which all have limited data rates, without using compression.<\/p>\n\n

Adding color with the Bayer mosaic pattern<\/h2>\n\n

At this stage of the raw recording process with ARRIRAW, you\u2019re still only dealing with the monochromatic image, with photosites collecting light and converting each photosite into one pixel of data. In fact, ARRI does have a rental camera that only records Black & White: the ALEXA XT B&W. Here is an example<\/a> from a commercial that used this camera without an infrared filter for a cool effect.<\/p>\n\n

\n