Arri Alexa and Mysterium-X...

[Roberto Lequeux]

Roberto

Why would you compress if there are available uncompressed recording devices like S.Two, Keisoku Giken, and Codex that record ARRIRAW?

With dual link and 3G HDSDI outputs Alexa will support "several" workflows. April 6th is just 4 days from now. We'll talk all about this... and more.

I guess what I am thinking about is size and money really. I look at these recorders and can't muster a reason to accept a system that requires a big external box that won't hold as much footage. One that will eat up a lot of budget.

But even if we took money out of the equation... say I don't have a small budget and I can spend more, and move slower... you still have to think about mobility and weight for steadicam, though if you really have cash you could hire a guy to run behind the operator with the thing on like a backpack I guess. The cables would be annoying but would only ruin one take out of many, hope not the right one.

But I guess on the 6th all of this could be addressed.

However with the camera costing more than an Epic, and having to get an additional module on top of that, I just don't see me agreeing to shoot Alexa only because a DP begs me not to shoot RED for whatever random reason they were force fed by those who don't really know what they are talking about (this frustration has nothing to do with you obviously). I need to have no-compromise look and malleability, while spending the smallest amount of my budget and making the process as easy as possible, so as to increase quality of execution, and speed for budgetary reasons. I want the money in the screen, not behind it.

I mean, you know I have a lot of admiration for ARRI -as anyone should- but I don't really see anything where Alexa will be better than RED.

...I guess not till the pictures come out and I can judge side by side at least, since so many say that Alexa's color will be more accurate, better, etc... but at this point I would need to hold the actual color chart in my hand, with a portable daylight lamp, and have access to a 4k projection of RED to find any discrepancies. Well... that doesn't sound ideal, but I get my point across I hope.

And still... even if I found that Arri colors were ever so slightly more accurate to the chart, not that my eyes are that amazingly trained -yet-, it wouldn't change how I feel when I see professional and filmic Red footage. It wouldn't take away from the fact that RED's color is already far beyond what anyone other than a computer could object to, unless you are in this field and the movie is about a love triangle between color charts.

An other area where we'd have to look closely though is this "usable DR". Not only does this range have to be clean of noise (and be recorded), but it also has to be stored properly. I wonder if this dual channel deal in Alexa helps malleability, making it's 2k RAW push farther than 5k down to 2k mathematically lossless our of Epic. Though keep in mind however that at 2k the measurable DR of Epic would be like 14.25 stops, right?

I am sorry, I guess what I am trying to say is that if Alexa and Epic do what they are expected to do, ARRI needs to catch up. To me, a hair ahead in color reproduction which I haven't yet seen myself, isn't a selling point that can overshadow the fact that M-X matches or beats all other aspects relevant to this particular filmmaking tool.

 

[Mark L. Pederson]

Why would you compress if there are available uncompressed recording devices like S.Two, Keisoku Giken, and Codex that record ARRIRAW?

Efficiency. Speed. Cost.

(I've actually used recorders from all three - and if you have to go down that road - Codex, IMO is the best.) But these recorders are NOT cheap and large raster uncompressed data files vs. Redcode with respect to speed and efficiency .... it's no contest.

Michael - I'm sure that at, or before NAB - ARRI will LOWER the price of Alexa 20% - in addition to announcing an on-board SSD recording option that will be "available some time later in the year".

I assure you, as soon as people can shoot on-board with ALEXA - about 5% or less of your users will work tethered to a recorder.

I guess after your on-board recording announcement - rental houses will realize if they are going to invest in these recorders for ALEXA customers - they better ROI them quickly.

 

[Michael Hastings]

Epic will be very nice for 3D for both beamplitter and side by side as it looks like it can get very close. I heard it's a 2.5" to 3" interocular.

Alexa in a side by side rig would get you to a 6" interocular.
Mr Bravin, is that about right?

I think you better check again. From published pictures with a known reference I estimated SCARLET at just under 4" minimum intraocular (best case a little over 3" if you could remove the bulge on the left side where the start stop and VF are and I don't think that's possible) and from handling Epic at RED day it seemed that epic was significantly larger.

Using the 30x rule of thumb for IO to near subject distance that would give a near subject distance of 10 feet which is fairly limiting.

 

[Sanjin Jukic]

Roberto,

don't you get finally that Michael wants to sell you Alexa and also to other REDUsers

just because he is an ARRI seller that means in t(his) case everything ARRI is better than RED.

You could make straight away a short movie out of this story (a comedy with Monty Python style).

Also you've got all the dialogs, so just "twiddle" and "spin" them a bit around then you are ready to roll.

And of course you should shoot that comedy on R-1-M-X.

 

[Michael Hastings]

On the other hand, I would imagine that the smaller a file is, the easier and cheaper it is to store it in a reliable way till it gets backed up.

So I totally understand the idea of retaining all information, but I don't see a justification for keeping an uncompressed file from becoming smaller through mathematically lossless compression. Right?

Roberto:

I like REDCODE and think it is a tremendous application for field shooting and many other types where size and speed are essential - AND IT WILL BE MY CHOICE. But I think you are missing two points:

1) ANY form of compression REDUCES reliability. For the sake of simplicity let's do a simple analysis (I know the real math can be a bit more complicated) : A two drive RAID 0 array like the REDDRIVE is TWICE as likely to fail as a single drive. By the same token any bad bits even on a mathematically lossless 2.5 compression would cause 2.5 times the errors on reconstruction and a 10 to 1 compression scheme could be 10 times as destructive. (This is why a simple dropout on an HDV tape can cause loss of frames while the same dropout on more robust (less compressed) formats is handled by the error correction and there is no loss.)

That might not be so bad when Mike H. is swimming around filming sharks - or on a microbudget feature. But when you are spending half a million a day on a production, even losing a minute or two of footage that might mean a reshoot could cost hundreds of thousands of dollars and all of a sudden that $50K you saved on the camera doesn't look so smart.


Second: Drives and storage are getting really cheap. I just bought top quality 2 terabyte drives for $150 each. That's something like 18 hours of 4K RC36 footage. If that is 10:1 then even uncompressed would be 1.8 hours so even at 25 to 1 shooting ratio that's only about 30 drives or $4500 - trivial for any normal budget movie (less than half that for an hour TV show).

ARRIRAW may be substantially more data but even if it is two or three times that plus double drive backups plus tape backup it is not a particularly large expense.

 

[Roberto Lequeux]

There isn't a person on a set that could care more about the safety of the cans than a writer director. We are keenly aware of the impact each moment lost -footage and schedule- has/would have on the likelihood of the purest manifestation of our story, which we usually bleed for for at least a few years before anyone even hears about the "job". I assumed everybody would realize such an evident and logical thing.

My point was that a smaller file would require a less complex media, smaller means media can also be slower... taking us in the other direction, safer. If you read my post closely you'd see that my point was not on storage, but magazines. Of course, all moot when considering things such as SENSITIVITY, DYNAMIC RANGE, RESOLUTION...

I think my polite tone might be getting confused with naiveness. I can't help it people... I will do anything to get to the truth. And it is only fitting IMO, since I am not a DP, rental company, single cinema camera owner, camera manufacturer... etc. Perhaps I got carried away, trying to overcompensate for SJ's direct and passionate observations. I have seen what happens to forums, DSLR forums are sand-boxes all too often, C.com had many a flame fights... I think we are a different breed there, and SHOULD ACT LIKE IT. If you are at the front of the line, wear it out.

Respecting Michael will only get me more information, despite how many times over I may be sold for Epic. I would hope he would respond to my last post, since I wrote so much... if he doesn't dispute anything it means I was right. He sort of gets paid for disputing erroneous statements regarding Alexa, doesn't he? And do any of us need to be clarified/sold on any of these EPIC points behind Epic? Really?

We've all seen 5k at REDCODE 28. We've all played with RC36 and turned it inside out without a single issue! If it worries you, I'll tell you right now that I'll probably shoot 5k at about RC42-50, because I am aware of how affordable 2TB enterprise drives are, and how many I can run at 800MB/s off one single RAID card. RC42 would make me feel as safe as a baby in a mother's embrace, I am convinced it is more than necessary for 4.5k. I am only interested in lossless, but I am yet to think of a single sequence where I would use it over something like RC80-100. We are talking 5k to 2k with the distinct possibility of 4k.

 

[Graeme Nattress]

On the other hand, Michael H, reliability is inversely proportional to amount of data you need to retrieve / store. The more data you store, the more likely some of it will get an error.

HDV is bad for dropout because of it's temporal nature where the current frame is based on up to 15 previous frames and any visible error will propagate over the length of the GOP.

Good compression is a wonderful tool. But like all tools, abuse it, and it is bad. Use it properly and you get all the tremendous advantages we've had posters mention above, not least cost, efficiencies, weight, and speed. And if even just one of those benefits allows you to get a shot you'd otherwise be un-able to, you can see how valuable good compression is.

Graeme

 

[Roberto Lequeux]

On the other hand, Michael H, reliability is inversely proportional to amount of data you need to retrieve / store. The more data you store, the more likely some of it will get an error.

What he said... ALL of it! Even the posts I haven't read.

 

[Jeff Coatney]

Phil

Alexa has a 3.5K sensor and outputs 2K ARRIRAW data which is uncompressed off the CMOS sensor (no luck needed).

I don't understand how you can take a 3.5K sensor, down rez it to 2K and say that it is UN-compressed. If you aren't capturing the RAW sensor data pixel-for-pixel, you are by definition compressing the image.

To truly claim that you're capturing uncompressed data, your stored media bandwidth must match your native sensor bandwidth. Doesn't the very architecture of the CMOS sensor limit your ability to capture pixel-for-pixel, thus requiring a codec to resolve "native" resolution?

Also the moniker "ARRIRAW" paints a picture of a specialized version of RAW that seems like a cover term. Is this term just a way to make people think you're getting RAW when in fact you're getting something less than? What is the "ARRI" part of "ARRIRAW" mean? What does it do and why is it there?

 

[Tim Morten]

Roberto:

1) ANY form of compression REDUCES reliability. For the sake of simplicity let's do a simple analysis (I know the real math can be a bit more complicated) : A two drive RAID 0 array like the REDDRIVE is TWICE as likely to fail as a single drive.

[SNIP]

Second: Drives and storage are getting really cheap. I just bought top quality 2 terabyte drives for $150 each. That's something like 18 hours of 4K RC36 footage. If that is 10:1 then even uncompressed would be 1.8 hours so even at 25 to 1 shooting ratio that's only about 30 drives or $4500 - trivial for any normal budget movie (less than half that for an hour TV show).

Michael -

As has already been pointed out, the less storage space consumed, the less likely that you will encounter faulty storage. This is a clear benefit of compression.

Also, you are wildy mistaken about your assumption that any cheap storage will suffice for recording uncompressed. The fact is that the data rate required by uncompressed predicates an extremely expensive SSD RAID (assuming that we're talking about high resolutions - SSD may not be required for simple HD, but we're definitely still talking RAID).

As a fellow Red user, you have my support, but please be cautious in the technical assertions that you are making - these two are simply not correct.

- Tim

 

[Graeme Nattress]

I don't have any problems with good compression. A typical 4k R3D file may be around 10:1 compressed, and as such, the data rate is similar to that of uncompressed PAL SD. Now, ask me which I'd take - the ~10:1 compressed R3D file, or the uncompressed PAL SD file. Answers on a postcard. Winners chosen randomly. No purchase necessary.

Graeme

 

[Tim Morten]

Until we see some images, this is my impression based on the specs:

Alexa is comparable to an R1-MX, but take away 75% of the resolution and triple the price.

No offense to Michael, but why would I buy that?

 

[Martin Stevens]

We are looking for LOSSLESS image recordation, not UNCOMPRESSED
image recordation.

In the past the only virtue to uncompressed was that it was lossless.
However with the advancement of science we now have an option to
record images that are compressed but are lossless.

So Red wins again!

My mind explodes with how smart
the Red Team is!
:emote_head_explode: :emote_head_explode: :emote_head_explode:

 

[Wendy Woods]

I am very interested in the 4K loosen, when is that announcement Bravin and when do you think you will be shipping? Is the Alexa the "Scarlet" in the Arri family, and the Loosen the "Epic" in the Arri family?

 

[Jarred Land]

I think you better check again. From published pictures with a known reference I estimated SCARLET at just under 4" minimum intraocular (best case a little over 3" if you could remove the bulge on the left side where the start stop and VF are and I don't think that's possible)

I think you better check again Everything in your statement above is false.

 

[Casey Green]

On the other hand, Michael H, reliability is inversely proportional to amount of data you need to retrieve / store. The more data you store, the more likely some of it will get an error.

HDV is bad for dropout because of it's temporal nature where the current frame is based on up to 15 previous frames and any visible error will propagate over the length of the GOP.

Good compression is a wonderful tool. But like all tools, abuse it, and it is bad. Use it properly and you get all the tremendous advantages we've had posters mention above, not least cost, efficiencies, weight, and speed. And if even just one of those benefits allows you to get a shot you'd otherwise be un-able to, you can see how valuable good compression is.

Graeme

Right on. (Sometimes I think you help the competition a bit too much! Hehehehe! But that's what leaders do...)

 

[M Most]

I don't understand how you can take a 3.5K sensor, down rez it to 2K and say that it is UN-compressed. If you aren't capturing the RAW sensor data pixel-for-pixel, you are by definition compressing the image.

To truly claim that you're capturing uncompressed data, your stored media bandwidth must match your native sensor bandwidth. Doesn't the very architecture of the CMOS sensor limit your ability to capture pixel-for-pixel, thus requiring a codec to resolve "native" resolution?

Also the moniker "ARRIRAW" paints a picture of a specialized version of RAW that seems like a cover term. Is this term just a way to make people think you're getting RAW when in fact you're getting something less than? What is the "ARRI" part of "ARRIRAW" mean? What does it do and why is it there?

Michael was referring to the reconstructed image, not the RAW file record. Arri is saying that their 3.5K sensor yields a proper 2K image when reconstructed. The RAW record is, by definition, a value for each of the pixels in the sensor.

I do wish that some people here would have the courtesy and common sense to refrain from jumping to conclusions about products that have not yet been built, and jumping on people for using terminology that doesn't conform to Red's redefinitions of those terms. Production units of the Alexa do not yet exist. Neither do production models of Epic or Scarlet. Until they do, there is little to compare and no conclusions to be properly drawn.

 

[Martin Stevens]

I am very interested in the 4K loosen, when is that announcement Bravin and when do you think you will be shipping? Is the Alexa the "Scarlet" in the Arri family, and the Loosen the "Epic" in the Arri family?

Am I mistaken?.... I thought this "Arri Loosen" was some forum humor
suggesting that Arri was going to be losing. If it is not a play on words
then could someone please clarify.

I mean no offense to Arri.

:violin::violin::violin:

 

[Michael Bravin]

I am very interested in the 4K loosen, when is that announcement Bravin and when do you think you will be shipping? Is the Alexa the "Scarlet" in the Arri family, and the Loosen the "Epic" in the Arri family?

Wendy

This was made up by several folks here on Reduser. We do not have any other cameras besides the 3 soon to be released, ALEXA models and the current D21. This has been clearly stated before, but requires repeating... I guess.

 

[Matt W.]

Arri rocks! But the MX is pretty dope, too.