Codex XR.. finally

Codex XR.. finally

At last.. integrated Arriraw recording in the Alexa camera.

Codex Announcement >

Codex has worked closely with ARRI during the development and launch of the Alexa family of cameras. The Codex Onboard Recorder was the first recorder to be certified to record ARRIRAW from the Alexa back at the beginning of 2011 and has since been used on hundreds of feature films and commercials worldwide.

ARRIRAW has become the output of choice for feature films, including The Avengers and Skyfall and Codex recorders have become the recording standard. The ARRI Alexa features a 35MM CMOS bayer sensor. The sensor data is output over T-link to the Codex Onboard Recorder, where it is recorded on a Codex Datapack or Capture Drive. ARRIRAW is 12 bit logarithmic raw Bayer data. The resolution for 16:9 is 2880 x 1620 and for 4:3 it is 2880 x 2160. ARRIRAW can be output and recorded at up to 60FPS for 16:9 and up to 48FPS for 4:3 to an external recorder.

Fast forward to 2013. ARRI and Codex announce the new Alexa XR  (extended recording) Module. Building on the success of the Alexa/Codex combination,  Codex and ARRI have developed a module that incorporates Codex recording technology directly into the camera. This alleviates the need for cables between the recorder and camera, makes the camera package smaller, and further simplifies ARRIRAW recording.  It also enables higher speed ARRIRAW - up to 120FPS for 16:9 and up to 96FPS for 4:3.  These developments are bound to intrigue cinematographers and further cement the Alexa/Codex/ARRIRAW workflow as the standard for digital production.

The XR Module provides several recording options in a single package. ARRIRAW at up to 120FPS (16:9) can be recorded onto a high performance Codex Capture Drive. In addition, Apple ProRes or Avid DNxHD can be recorded to a Capture Drive, making longer recording times possible (up to 2.1 hours of ProRes 4444), or with an SxS adapter, to an SxS PRO card.

Anyone who's spent some time working with the Alexa and Codex on-board recorders can attest to how desperately needed internal Arriraw recording is. Just about everything about the Codex system is thoughtfully designed and executed - the stability of the recording, robust equipment, and control of metadata and deliverables with the Virtual File System. The big problem is the deck is quite large, is an expensive rental, and there are inevitably BNC cables running between it and the camera. Not to mention getting the Codex working happily with the Alexa means matching many menu items on the two devices, several of which are labeled differently and mismatch can result in irreversible recording errors. It's a good system but it's one that's by no means bullet proof.

In any video recording system, whenever a cable is introduced between the camera and the recorder the possibility for error is exponentially increased. A native, integrated recording system controlled solely by the menus within the camera greatly reduces all these problems and makes for a faster and more intuitive user experience. Needless to say I'm thrilled to see this functionality coming to the Alexa and think Codex was the obvious partner given their expertise with Arriraw. Why reinvent the wheel if you don't have to?

All new Alexa's (except the entry level 16x9 camera) will ship with the XR module and will be called Alexa XT (Extended Technology).

All existing Alexa cameras can be upgraded with the XR module. Sony SxS cards for ProRes recording only will still be able to be used with an adapter. In addition to native Arriraw recording, XT cameras will feature new processing hardware, 120 fps recording, 4:3 sensor and anamorphic desqueeze, and IFM In-Camera Filter System for behind the lens IR ND filtration. <via Film & Digital Times>

Behind-the-lens ND is actually another godsend and something I've been preparing a separate post on. Conventional, Front-of-the-lens filtration and use of heavy ND, sometimes 7 or more stops, is the most destructive factor in digital imaging and something that is easily remedied by putting the glass behind the lens. As Digital Imaging Technicians, painting out color temperature offests in neutral density filters is one of our most common tasks and another factor that it seems technology will alleviate us of. Much more on this later..

Conveniently the integration of Arriraw into the camera solves another big problem for us, and means another blog entry I don't have to finish writing, which is the problematic workflow for on-set color correction with Codex recording.

Because all the current external Codex decks - M, S, and Arriraw - require the Alexa's REC video outputs, getting a Log-encoded video signal to be used for on-set color correction presents a host of challenges. Ordinarily, the DIT would use one of the camera's REC Output's for a clean, 10 bit, 422, Log-encoded video signal to be used with color correction software such as LinkColor or LiveGrade. This allows the operator to create data directly on the set in the form of ASC-CDL or 3DLUT's to be used for color corrected production dailies. When the camera's REC Out's are being used to feed the deck and the deck in-turn doesn't output a video signal that's useful to us, the user has to get creative to get their workflow working.

This tends to be the workaround -

Alexa REC Out's to Codex Input's. Even with a Single-Link 3G recording, the camera's other Rec Out still ouputs a data stream that isn't useful for video monitoring.

So in order to get Log-encoded video for the DIT's use, the camera's MON Output needs to be put into Log C. Addtionally, the MON Out can only output a Legal levels video signal unlike the REC Out which can be set to Legal or Extended levels. There are workflows that require an Extended level video signal so if we can't get one out of the camera, this can present more problems to be solved.

Using a 3-wire BNC video harness, one cable is used to monitor the Codex' output which unfortnately isn't useful for anything other than verifying the recording. The second cable takes the Log-encoded MON Out from the camera to the DIT where it's color corrected and then using the third cable in the harness, is fed back to the camera to be used by the assistant's on-board monitors.

The main problem with this is it's often impractical to send a color corrected return back to the camera to be used by the AC and operator. This can be for any number of reasons - long cables runs, wireless video for a vehicle shot, process trailers, etc. In practice, it really isn't fair to them to have to focus using Log C video as this flat, washed out image makes their job even more difficult. The operators tend to hate it as well. Codex XR solves these problems. A separate video channel for users at the camera, an in-camera recording, and a single video link to the DIT is always ideal. Or better yet - a single wireless link. Less is more. Keep the recording and monitoring as simple and discrete as possible and everyone in the camera department - DP, operator, assistants, and engineering - can do their best work.

Cutting the Cord - Follow Up

Cutting the Cord - Follow Up

February 4, 2013

As is always the case with content published on this site, the first version is certainly not the last. Through valued reader feedback, these articles are amended and updated over time. Chris MacKarell at Arri CSC Digital, whose feedback has routinely been beneficial to this site, raised some valid points on my post on Wireless HD-SDI that I'd like to include. The main concern is that for Wireless HD Video, because this digital signal has to be converted to analog for transmission and then converted back to digital upon reception, this resulting signal can never really be "identical" to a cable-bound one no matter how high quality it is.

How true! 

From Chris -

.. as far as wireless monitoring is concerned, here some questions you didn't touch on which certainly affect the quality of the image on the monitor:

At the Transmit end, precisely how is the deserializing and digital-to-analogue conversion of the original HDSDI signal performed?

How is the reverse, the analogue-to-digital conversion and serializing of the signal at the Receive end performed?

How high a fidelity are these DAC and ADC functions, what quantizing steps are taking place to perform them?

How are overly high peak-to-average power ratios in the modulated signal dealt with? - Note that this is usually done by clipping the transmitted sine wave.

How is intermodulation distortion i.e. non linearity in the signal chain handled?

If COFDM, how is this managed?

Of course, there are industry standards ( e.g. IEEE 802.20 ), but were they designed for the kind of critical application the DIT requires?

The key question is, can you trust what you see on your monitor at all times?

I suspect that for many, any questions at all around the image processing techniques applied and therefore the integrity of the subsequently transmitted image, are enough by themselves to mitigate against that signal's utility for serious critical imaging work.

And since all this processing will always be a necessary precursor of image transmission and reception, the wireless utility you seek for on set monitoring applications perhaps maynever be approached.

In short, whatever future technology developments occur, the underlying principles governing wireless transmission will not change. Wireless may therefore never yield the kind of critically accurate signal you require in your day-to-day work.

All of the above points will certainly affect image quality and is something to be aware of.  As Chris pointed out, "the underlying principles governing wireless transmission will not change," so no matter how good these systems are, just how cautious should we be in using them? Should we even be using them at all?

Unfortunately, we often don't have much choice in the matter and if the shot calls for wireless, then it has to be wireless. In my opinion, if you know the nuances of your wireless system and have spent some time testing and evaluating the image it produces, you can come to trust it. My favorite thing to say on the set when things become questionable is, "the scopes don't lie." Waveform and Vectorscope are probably the only thing on the entire set that are truly objective. They reveal problems now that will certainly be problems later and basically tell you everything you need to know about the video images you're working with. In the case of wireless video, the scopes can tell you just how well your system is putting that signal back together upon reception.

Here's a great way to test this - using scopes, compare the exact same image - one coming to you over the air and the other through a cable. Use one of the camera's HD-SDI outputs to feed the wireless Tx and then run a hardline to you from the camera's other output. Now switch between the two and study the waveform and vector. Are you seeing any shift in Chroma in the wireless image compared to the cable? What about contrast? Does the highlight and shadow information in the wireless waveform sit at the same place in the cabled image? What about midtones? Are they more compressed in the wireless image? 

Really if you're going to use this video signal for anything other than basic monitoring, any differences between the wireless and cabled image should be minimal. In the case of the Boxx Meridian, in my experience very little if any, shift in chroma or contrast is evident in the wireless image. As the signal degrades, the image becomes noisier and blockier but largely maintains its correct luma and chroma information. This is why we spend so much time testing gear. You're going to have to use it and often in very compromised situations. The more you know about its strengths and weaknesses, the more confident you can be in its operation. 

I encourage anyone who follows this site to contribute to the knowledge base and feel free to eviscerate anything I've written here with a fine-toothed comb. I'm a technician and not an academic so many of the topics I address here are much more focused on practical application or the "end-user experience" and not the hard science driving it.