World's first side-by-side ARRI ALEXA and RED MysteriumX camera test

Posted In | Blog Categories: On set, Digital cameras, Compositing | Site Categories: 2D, Technology, Visual Effects

 

For our current film, Roland Emmerich's "Anonymous", a drama about Shakespeare (yes, you read this right. No destruction. No end of the world.), we just finished up three days of camera testing.

 

Not just any camera test, but the world's first side-by-side test of the new RED MysteriumX (which is the REDONE camera, with a brand new chip and a new software codec to go with it) and the ARRI ALEXA (which is a brand new camera, very different and way more compact than the current ARRI D-21). We were privileged to be provided by ARRI with a prototype of the camera, and by RED with one of the first delivery models of the Mysterium X.

 

Before I dive into the specifics, ladies and gentlemen, let me announce right here, on this blog, that     

 2010 IS THE YEAR THAT CELLULOID DIED.

Yes, all you physical-analog-film-lovers, I said it. Touch the smooth plasticky emulsion  one last time and relish the memory. Both these two digital cameras simply blow film out of the water. Period.

 

Daylight/Candlelight test

 

Mixed daylight and candlelight test

 

Now, to be honest, I wish the names of these two would be more celebratory for such an event (my guess is the "mysteriumX" was named by the creators of "Godzilla vs Mothra", and the "ALEXA" for somebody's girlfriend?). But that's beside the point.

 

The point is, both these cameras can do what you could not accomplish with any film stock, not in your wildest dreams. Here's how they work:

 

The RED and the ALEXA, while accomplishing approximately the same amazing feat of creating clean, VERY low noise, pictures in candlelight conditions, work in slightly different ways.

The RED still has a chip that's basically rated at 320 ASA (or more like 200 ASA practically, as many DPs told me), but you can underexpose it in extreme ways with only very little addition of noise.

For instance, we shot two stops under (equivalent of 800ASA) without any significant addition of noise.

3 stops under (1600 ASA!) is still absolutely feasible, with more noticeable noise, of course. Even 4-5 stops under (3200/6400 ASA) is not out of the question, if you use a good noise removal tool (we tested with ARRI's Relativity software and The Foundry's Furnace for Nuke) and are OK with loss of detail ion the blacks. But it's possible.

 

The ALEXA has a chip rated at 800ASA. But even with that starting point, it's possible to underexpose significantly with only little loss. We actually went so far to test 4 stops under, the equivalent of 12,800 ASA! And, while of course extreme, I would not be beyond saying it's usable for a certain type of shots (documentary style). Basically, I was absolutely blown away that we even saw ANYTHING on the footage.

 

We recorded the RED footage on CF cards, at a data rate of 42MB/sec (the highest data rate RED offers). The ARRI footage was recorded on CODEX portable machines. The CODEX records on little removable harddisc RAIDs with a capacity of up to 5 hours each. The RED records to their own RED RAW format, at 4480x1920 resolution, which is currently only supported via the redcineX (in beta) software, but is already also being implemented by other software packages, such as Digital Fusion and Baselight.While ARRI would not say what on-board recording solution they will show at NAB, in our test the ARRI output a 4:4:4 dual-HD-SDI stream at 1920x1080 resolution, which can be recorded by any HD recorder. The CODEX that we used, creates internally their own implementation of a JPG2000 file. During output via Ethernet or other means, it creates 10bit DPX files from the JPG2000s. In terms of file size, the Redraw 42 and JPG2000 are fairly similar. JPG2000 is a 1:4 compression, so about 2MB/frame. The Redraw comes to about 1.75MB/frame (at 24fps).

 

The DP for "Anonymous" is Anna Foerster. Anna started with Roland as DP for the miniature shoot of "Independence Day", and advanced to second unit director on many of his films, and is now also a director (she directed episodes of "Criminal Minds"). She prepared several lighting setups, one being a Vermeer-painting-like interior scene (pictured) and the other a greenscreen test setup, obviously of great importance to us. (attached are before-and-after pics of the greenscreen test).

 

At this point in time, I don't want to publish exact comparison details, because the ARRI ALEXA is a prototype camera. It will be shown at NAB and delivered in June this year. So not all the specs are finished yet. But I have to say, no matter what the final details will be - this is a damn fine camera, and the range of the chip is simply incredible. The RED Mysterium X is already shipping, and all I can say is - go out and test it for yourself, you'll be amazed.

So, it turns out that both cameras, in terms of performance of the chip, are very similar. And I'm not saying this for some bullshit political correctness reason, but because it's the truth. Since the ARRI is a prototype, and the RED software is still in beta, I don't want to go into the finer details that we found, because they can change at any point in time.

 

I've attached some pictures from the test shoot, a daylight/candlelight scene and a greenscreen shot (the raw, uncorrected footage and the keyed result). Do to the limitations of this blog format, these are just downrez'd JPGs, but even here you can see what I'm talking about. I will prepare links to high-rez files soon. The greenscreen footage of both cameras was very clean and - especially for 800 ASA, had very low grain/noise. While you still need post noise reduction to get a perfect key in hair details or the semi-transparent collar of the lady in the picture, we only had to apply very little NR, and it didn't create any visible softness in the picture.

 

 

"Raw" greenscreen image
"Raw" greenscreen image

 

Keyed image
Keyed image

 

 

 "Tracking Marc" will return next week: Shooting HDRIs for CG lighting and textures.







Comments


I think film is dead. It really only exists in large features (where the budgets run into the millions). In ad production is it is only used in about 10% of commercial productions. If anyone has learnt from history, the writing on the wall couldn't plainer. The advocates of film believe that if something is better, then it will not die. This is wrong. Why? Lets take a history refresher shall we?
1. The diamond disk record player (1916) has superior sound to the 78 record. The 78 record was cheaper, owned by more people, so artists sold more records on that format. Eventually they stopped making diamond disk records because the critical mass was no longer there.
2. The same can be said for quadraphonic tape. Better but stereo kicked its butt.
3. VHS and Beta. Beta was superior but lost the critical mass.
4. Superbit dvd audio. It was superior to the 44.1kHz cd yet the critical mass wasn't there.
5. Mp3 vs cd. Mp3 have worse sound yet the are seriously eroding sales of compact discs.
6. Typesetting vs Apple SE. When I was a paste-up compositor in 1988 the was the correct perception that an $80,000 typesetting machine would never lose its market share because it was better for setting type. All the professionals know this yet, they still made the change to setting type on a $5,000 desktop computer.
7. Digital still photography. Again, large format cameras were better, but now have lost the critical mass.
8. Audio tape vs Digidesign Protools. Tape has infinite audio detail. Yet, digital recording became commonplace because of the speed and the convenience.
In conclusion, I say that film is on its last legs now. Not because it's worse quality, indeed it is better, but because the critical mass won't be there soon. The support services will dry up and the decision will be made to the convenience digital. However in the far future, I believe you will always be able to order 35mm film from boutique enthusiasts, probably the same ones who still make 8mm film.

Geoff McLeod (not verified) | Wed, 03/10/2010 - 05:18 | Permalink

I have grown very frustrated waiting for RED's future camera announcements. I know that their saying is "Everything is subject to change. Count on it." but do you really have to announce dream platforms that are a decade away? 2 years ago they announced that they would have the 645 and 617 Epic monstro cameras by now, but they haven't even pulled off a full frame35 yet. Any company that makes such radical claims and doesn't deliver is dubious...I'd take the Arri by default.
Need another reason...just compare websites, the RED site looks like something that a highschool kid made.

Red with envy (not verified) | Wed, 03/10/2010 - 05:03 | Permalink

"2010 IS THE YEAR THAT CELLULOID DIED."

Yeah right... Try to restore a film backup in a hard drive or a tape-backup in 100 years from now...
Film is completely analog to our perception. No special equipment for viewing that there is an image encoded there.
I want to see some good digital format that is readable in the not so long future.
That is why major films are stored in 3 black and white internegatives ( one for each color channel ) separations and stored in special bunkers.

ariel wollinger (not verified) | Tue, 03/09/2010 - 14:18 | Permalink

Nice test and report Marc.
About the death of film - consider this: What percentage of people leaving the cinema after a seeing a movie shot on film are saying "I wish there had been less/more grain, noise, detail in highlights/shadows, color depth etc. etc."?
What percentage of cinematographers are saying the same?
Maybe film will die, but probably not this decade. I think. Also I have never heard of anyone predicting the future correctly... :-)

Mikael Kern (not verified) | Tue, 03/09/2010 - 07:56 | Permalink
mweigert's picture

not off topic at all. We tested with two different keyers, KEYLIGHT and PRIMATTE. The actual picture is done with keylight in Digital Fusion. And we used furnace on Nuke for (gentle) noise reduction.

mweigert | Mon, 03/08/2010 - 00:43 | Permalink

If it's not considered off topic, could someone mention what method/software was used for the green screen keying?

Tariq (not verified) | Sun, 03/07/2010 - 19:14 | Permalink
Jonathan Smiles's picture

Jamie,

The Mysterium-X (MX) sensor has an effective range from ISO 250-4000.

The equivalent of 0db gain is ISO 250 but it's better to use ISO 320 as the baseline as it gives us highlight protection.

We can increase ISO to 800 with no visible noise penalty so this combined with REDcolor and REDgamma  putting mid grey in the middle is nice and easy as a working baseline as you have correctly found.

The key thing is to use the cameras exposure tools to prevent clipping highlights and crushing the shadows to give the fattest negative for the DI.

Changing the ISO does not change what the sensor sees with MX, it's just metadata.

The other cameras you mention don't work the same way as the RED whose R3D files are a compressed RAW of what the sensor sees giving tremendous flexibility in post just like shooting RAW stills on a DSLR.

The wide Vermeer style shot above was shot in 4.5KWS mode on RED MX at ISO 800, I graded it in REDCINE-X on set.

I would be interested in seeing some of your work Jamie, as even with over 3 million feet shot on RED I am always learning.

Jonathan Smiles
RED Camera Supervisor on Anonymous, District 9, Green Zone, Red Riding 1983 etc.

Jonathan Smiles | Sun, 03/07/2010 - 16:41 | Permalink

I'm really excited about both of these formats and can't wait to shoot a project with them.

But writing: "2010 IS THE YEAR THAT CELLULOID DIED."

This is just a silly thing to write and didn't make me take the rest of the article seriously. I love both film and digi and I hope they'll both always have a place.

Stefan (not verified) | Sun, 03/07/2010 - 15:22 | Permalink

Marc,

Arrived at the ISO 800 based on detailed exposure testing.

If you wish to hire me let me know and I can go into more details.

I have ask how did you arrive at the ISO 320/200 rating yourself?

Others may arrive at a different ISO rating conclusion but I can only refer to my own exposure testing which revealed an ideal baseline ISO rating. The exposure testing I conduct has been accurate for the F35, F23, Red, Panasonic digital cinema cameras so I have an established set of parameters which I can rely on.

Regardless I do not rely on second hand information rather I verified the results myself via detailed testing with the actual camera system, I fully expect to conduct the same critical level of exposure testing with the Arri camera in the near future.

As you are aware the post production workflow is critical to maximizing the image quality possible from any digital cinema camera. Just as you point out the error's a DP can commit so can image processing errors be committed by an uninformed post production facility which does not understand how to best deal with the digital media they have in hand.

Please note no where did I refer to the Arri as I have not had the chance to test it yet. My comments were limited to the Red and F35. Of course my comments are subject to change as the product matures and future products such as Epic is released. All products are constantly changing and evolving which is a good thing for everyone involved.

On set monitoring is only as good as the calibration and viewing conditions. Exposure errors can still occur if the viewing equipment is not properly calibrated for an accurate baseline.

I can't comment on the DP issues you mentioned as I have not encountered such a problem.

I can agree that for cost effective consideration shooting in the digital realm does have major economic advantages.

Not sure why you were shocked to read from a DIT that "Film is not dead" but then again I was not surprised to read your comments from the perspective of a producer.

Bottom line is producers want to save money by whatever means possible.

Nothing wrong with that.

Let me end that there is no ideal or single camera system that is the end all solution for every application.

There will still be a place for motion picture film capture as there will also be a place for digital capture.

In some cases using both within the same project could make sense creatively and economically.

Combining any technology to achieve the creative vision of the DP, director, DIT, Visual EFX and even the producer is my ultimate aim. In the end the viewer receives the best all involved had to offer.

Jamie Jones

PS

Currently redcode 42 is possible with 4.5WS mode but not 4K 16x9.

jamie jones (not verified) | Sun, 03/07/2010 - 09:04 | Permalink
Jonathan Smiles's picture

On the Anonymous we shot 4.5KWS (4480x1920) mode REDCODE 42 at 24fps to 16GB REDFLASH CF cards.

4K 16:9 mode was not a requirement in this case, and is not a REDCODE 42 limitation just a limitation of the RED One bodies internal processing, that limitation is not present on Epic.

Jonathan Smiles
RED Camera Supervisor on Anonymous

Digital Production Supervisor & Consultant - UK, Europe & Worldwide
Camera Tech, On Set Colorist, DIT, Data, Workflow for Editorial VFX & DI
Credits: District 9, Green Zone, Red Riding 1983, Lesbian Vampire Killers, Unnaipol Oruvan
Tel +44 7966 014163 - Agent: Sara Putt Tel: +44 (0)1932 571044

Jonathan Smiles | Sun, 03/07/2010 - 08:31 | Permalink

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd>
  • Lines and paragraphs break automatically.
  • Use <!--pagebreak--> to create page breaks.

More information about formatting options

CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.