Back from NAB

[David Mullen ASC]

Just got home tonight after nearly two days of NAB...

 

I almost hate to make a post about the RED camera because of the automatic posturing from people (on both sides) that always seems to result, but I'll do it anyway.

 

For me, the RED camera was the highlight of NAB. The Peter Jackson short film shot on the two RED prototypes was an amazing achievement. I would describe the look as something like 5245 50D 35mm scanned at 4K -- sharp & clean. Fine detail even in extreme long shots (of course, the demo also points out the beauty of 4K digital projection to show-off that detail.)

 

It's to the point where I don't care if "de-Bayered 4K Bayer-filtered is truly 4K", etc. because all that matters to me is that I didn't see any compromise in image resolution compared to 35mm, unlike with HD where it visibly craps-out in extreme long shot on the big screen.

 

Exposure range was excellent, though I would say that color negative still holds the edge in overexposure detail. This was closer to what you'd expect from a high-end digital SLR still image in terms of dynamic range, or a slightly contrasty color negative stock like 5245.

 

I didn't see any problems from the REDCODE compression, even though it's something like 12:1. It truly seems to work, although I'm sure someone could design a test that could stress it probably. But it seems a godsend for making RAW Bayered-4K field acquisition practical. An attached hard drive the size of a camera battery can hold something over three-hours of 4K material because the data rate is so reduced.

 

What was great about this demo was that it wasn't some studio controlled situation, but two days of shooting run-and-gun in changing outdoor light and weather many set-ups per day with barely a moment to grab a meter reading, fast camera moves from full sun to full shade and back, etc. the worst combination of elements for a digital camera, yet it handled it great and didn't slow down the production.

 

So congrats to the RED team -- they've leapfrogged a couple of years ahead of where I thought digital movie cameras would be today based on past experience.

 

Out of hundreds of shots, there were three or so maybe with a technical problem that the RED team recognizes, a faint patterning from a sensor that needs re-aligning, that only became visible when an underexposed shot was brightened in post (you saw something similar in some shots in "Deja Vu" shot on the Genesis). Graeme Nattress says that it will be fixed soon.

 

In a lot of ways, this camera is what the Kinetta promised to be three years ago, and I think its success is going to shake-up the long-held domination of corporations like Sony and Panasonic in the mid-range professional video camera world. Sony's talking point at NAB was to dismiss these cameras (RED, Dalsa, SI, Phantom, etc.) as "science experiments" rather than a practical reality, but I think they are a lot more real than Sony wants to admit.

 

Speaking of Sony, the new F23 is quite impressive in terms of what they improved over the F900 and F950. The dynamic range has been increased (by a couple of stops), the noise reduced, the Sony Log mode seems well-designed, etc. It can do 1-60 fps and speed ramps. As for the $200,000 price tag once you throw in the attached SRW1 recorder... well, it just depends on the client demand for the camera, because the rental houses will have to charge nearly $4000/day for the thing, it seems to me. I don't know enough about that market area to know if the camera will be a big success or not.

 

Other NAB items:

 

Tiffen has an affordable software version of their filters to similate their effects digitally. I think the cheaper version would be useful to give still shots the same filtered look planned for the project.

 

Rosco has this cheap plastic LED panel -- not that bright but perfect for applications like taping to a computer screen or car dashboard at night. The LED's run along the edge pointing inwards to a grid pattern that diffuses it. The whole thing is the thickness of a piece of plastic, like 1/8" I think, and the smaller one sells for $100.

 

ARRI tells me that the 2-perf movements are for the ARRI rental houses only and will be available by January.

 

Panavision told me that they had a few old 2-perf movements for their Panaflexes just sitting in a drawer for years and never told anyone (it's never been listed in their catalogs as an option.)

 

The Aaton Penelope looks great -- it's not much bigger than the Aaton XTR cameras or an Arri-235. And at 2-perf, you could shoot a feature with 400' mags and not get annoyed by the short loads.

[Alessandro Machi]

I didn't see any problems from the REDCODE compression, even though it's something like 12:1. It truly seems to work, although I'm sure someone could design a test that could stress it probably. But it seems a godsend for making RAW Bayered-4K field acquisition practical. An attached hard drive the size of a camera battery can hold something over three-hours of 4K material because the data rate is so reduced.

 

David, A, you are saying the images look terrific, and B, the images were shot with 12-1 compression.

 

Is what you saw at the show actually footage recorded from the on camera hard drive at 12-1 compression?

 

------------------------------------------------------------

 

What if I wanted to record at 6-1 compression, would that reduce the on camera harddrive record time in half?

[David Sweetman]

Thanks for the report, from those of us with school and papers due...

 

What's the Rosco light called? -nevermind, found it: http://www.rosco.com/us/video/litepad.asp

 

those are slick.

[Emmanuel Decarpentrie]

Thank you very much David for your honest and detailed report. I know the RED camera has been very controversial in the past, but I hope we will now spend less time to argue about the existence of this camera (I know some people still can't believe it and still call it a "scam") and more time discussing its highlights and capabilities.

 

Is what you saw at the show actually footage recorded from the on camera hard drive at 12-1 compression?

 

The whole 2 days of shooting was less than 500GB in total! Several hours of 4K rushes for only 500GB, and without any sort of visible compression artefacts. Pretty amazing I believe!

[Andy_Alderslade]

The Peter Jackson short film shot on the two RED prototypes was an amazing achievement. I would describe the look as something like 5245 50D 35mm scanned at 4K -- sharp & clean.

 

Just out of curiosity what was the short film like, any good?

[Alexander Nikishin]

In a lot of ways, this camera is what the Kinetta promised to be three years ago, and I think its success is going to shake-up the long-held domination of corporations like Sony and Panasonic in the mid-range professional video camera world. Sony's talking point at NAB was to dismiss these cameras (RED, Dalsa, SI, Phantom, etc.) as "science experiments" rather than a practical reality, but I think they are a lot more real than Sony wants to admit.

 

David, I'm wondering what you mean by mid-range pro video camera?

 

If Red is mid-range, what would you consider to be high end pro video?

 

I honestly see Red as not only competition to the top grade digital cinema cameras but a giant killer at that.

 

--------------------------

Alexander Nikishin

Cinematographer

Vision Cinema Inc.

Long Beach, CA

Edited April 19, 2007 by Alexander Nikishin

[Kevin Zanit]

I think he was talking more price than anything else.

[Loi Banh]

It means a lot if David Mullen is favourably comparing RED to Kodak 5245 EXR filmstock. For the budget minded indie filmmaker, this is fantastic news.

 

REDCODE is VBR wavelet compression, so compressions of 12:1 is pretty much visually lossless since the format is so efficient and not prone to artifacts like DCT-based methods. Of course, everything I've read about it makes me think RED's implementation is extremely promising. For a strong showing of a first gen product, RED has got to have KODAK, SONY and others worried.

Edited April 19, 2007 by Loi Banh

[Hal Smith]

Just got home tonight after nearly two days of NAB..............

 

I would describe the look as something like 5245 50D 35mm scanned at 4K -- sharp & clean.............

 

It's to the point where I don't care if "de-Bayered 4K Bayer-filtered is truly 4K", etc. because all that matters to me is that I didn't see any compromise in image resolution compared to 35mm..............

 

What was great about this demo was that it wasn't some studio controlled situation, but two days of shooting run-and-gun in changing outdoor light and weather many set-ups per day .................

 

So congrats to the RED team..............

 

Thanks David, now I know something REAL about RED.

[Mark Williams]

It means a lot if David Mullen is favourably comparing RED to Kodak 5245 EXR filmstock. For the budget minded indie filmmaker, this is fantastic news.

 

REDCODE is VBR wavelet compression, so compressions of 12:1 is pretty much visually lossless since the format is so efficient and not prone to artifacts like DCT-based methods. Of course, everything I've read about it makes me think RED's implementation is extremely promising. For a strong showing of a first gen product, RED has got to have KODAK, SONY and others worried.

Hmm I probably have this wrong but at a 12:1 compression wouldnt this mean that for its got huge chunks missing? and has to reconstituted by copying bits? Im just wondering how you are applying the term lossless?Is that lossless as in for words?Or lossless as in plot?

 

Sorry if I seem cynical its just that I dont have one?

[Andrew Ray]

David, I'm wondering what you mean by mid-range pro video camera?

 

If Red is mid-range, what would you consider to be high end pro video?

 

I honestly see Red as not only competition to the top grade digital cinema cameras but a giant killer at that.

 

--------------------------

Alexander Nikishin

Cinematographer

Vision Cinema Inc.

Long Beach, CA

 

Alexander, at the current state of camera sensor technology the film is bit better in highlights (2-4 stops)

The only way to beat it is to use two digital sensors, one servicing shades the other highlights. Some will argue with this statement because high end camera sensors are better in the deep shade so you could push it down and get improved dynamic range.

There is a lot of money invested in the development of camera sensor technology right now, so I expect constant improvements there on almost monthly basis.

 

Again, we are comparing film masters with the camera sensors.

Since most movies right now are going through DI process anyway that uses similar digital sensors as the 4K cameras do, the final distribution copy will have the same limitations in highlight. ARRI scanners though use two passes process though, one for neutral and one for highlights.

 

Andrew

[Alex Montoya]

Hmmm... or maybe it is just that video compression ain't one of your strengths.

[Andrew Ray]

David, A, you are saying the images look terrific, and B, the images were shot with 12-1 compression.

 

Is what you saw at the show actually footage recorded from the on camera hard drive at 12-1 compression?

 

------------------------------------------------------------

 

What if I wanted to record at 6-1 compression, would that reduce the on camera harddrive record time in half?

 

Alessandro, yes if RED CODE compression codec would have the compression ratio option say 6-1 then the final data output will be two times bigger as compared to 12-1, so it will reduce recording time by about 50%

 

However the on camera circuitry, may not be fast enough to record/process the data.

I think you can push the recording on RED to 6-1 with good safety margin but then when you use memory card recording media instead of hard drive it will cost you exponentially more.

 

Andrew

 

Hmmm... or maybe it is just that video compression ain't one of your strengths.

 

You will see the same dynamic range on the RAW output from the camera that clicks at 9GB/sec so I don't think so.

Edited April 19, 2007 by Andrew Ray

[Jeff Tanner]

David,

 

I'm glad that you had the opportunity to see the RED ONE. I had a feeling that you would be impressed. As I mentioned in another post, I went to NAB for one purpose...to see the RED camera and get some real information for myself instead of reading opinions about it. I fully expected to walk out of the booth knowing that the RED was a good start to a good idea but wasn't quite ready for high-end production work. I must admit that I was wrong. The footage in the demo was a great "real world" test and showed that the technology works. I would like a serious testing session with their lenses however. I'm not suggesting that they are inferior but I do want to see an in depth test of their performance versus other cine lenses.

 

As I've said before, this camera will change the way that my company works and I'm excited about the future of digital aquisition.

 

Respectfully,

 

Jeff

[Mark Williams]

Hmmm... or maybe it is just that video compression ain't one of your strengths.

So your saying there isnt huge chunks of data missing from 12x compression? what exactly are you saying happens to that missing data? and how is it reconstituted without copying from surviving information? how would that stand on fast moving action or even slow moving? Please explain?

[Michael Newton]

So your saying there isnt huge chunks of data missing from 12x compression? what exactly are you saying happens to that missing data? and how is it reconstituted without copying from surviving information? how would that stand on fast moving action or even slow moving? Please explain?

Unless you're looking for something that doesn't exist, I would summit that David Mullen's report pretty much answered your questions and more. I will take his word for it, if I were you and quit this line of thinking that will fetch you nor any of us for that matter, any gains.

[M Joel W]

So your saying there isnt huge chunks of data missing from 12x compression? what exactly are you saying happens to that missing data? and how is it reconstituted without copying from surviving information? how would that stand on fast moving action or even slow moving? Please explain?

 

Calm down for a minute. Yes, there are "huge chunks of data" that compression removes, but they aren't so much missing as they are reinterpreted into a more efficient (but less accurate) form. JPEG, one of the most basic there is, works like this: http://en.wikipedia.org/wiki/Jpeg or by using a more color space that is less accurate but visually as accurate to the human eye and cosine transformation from 1X1 pixels into combinations of 8X8 pixel grids. To be honest, I only understand some of it...as you can probably tell! At high image quality, you can make a very large image (20mb as a TIFF) into a 2mb JPEG that looks nearly as good, except under close scrutiny.

 

Redcode similarly uses different color space (it compresses BEFORE transforming bayer into RGB) that looses a bit of color quality in favor of efficiency, and uses wavelet based compression which is much more efficient than JPEG (I think the grids can be of any size, for instance.) Unlike H264, HDV, MPEG2 and similar codecs, Redcore is INTRAFRAME meaning that fast motion will have no effect whatsoever. No matter how fast you move the camera, the image will hold up just as well.

 

But...wavelet codecs have problems with areas of high detail. If you gain the camera up very high (1600ISO or so) you will get more noise. If you shoot deep focus, your scene will have more high detail areas. From what I've seen from some tests, Redcode softened footage a tiny bit...but not enough to really matter. I would worry a bit about deep focus photography with lots of gain...but if you have enough light or shoot wide open you will probably be all set....

 

Of course, that's just speculation. But compressing isn't like scaling down an image and "repeating" data, it's like giving a very accurate summary of data, from which it can be reconstructed so accurately, you (ideally) won't notice any effects of compresison.

 

So don't sweat it, or just shoot miniDV!

[Andy_Alderslade]

Hopefully this whole Red thing will shake up the medicority of the big video technology manufacturers, when I went to Video Forum at Earls Court, London a few months ago - the Panasonic and Sony forum/talks were the most boring not to mention depressing things i've ever attended - like who cares if you record to tape or solid state, its quality, versatility and reliability are what matter.

 

Thank goodness I quickly went into the Arri tent, their open minded 'can-do' spirit brought me back out of my commar.

 

Its seems like Red is a more serious threat to Panasonic and Sony than to film companies like Kodak.

[Hal Smith]

If RED RAW data is coming of the camera at 9Gbs it could be directly recorded with 3 or 4 current technology SATA drives slaved. I just bought a replacement 160GB Seagate Barracuda SATA drive for a workstation. The Seagate has a max serial data rate of 3Gbs so a small array of them could keep up with a RED RAW data stream (how long you can record is an issue of how many parallel sets, about 16 drives would equal a 35mm 1000' mag).

 

Given the RED crew's computer "chops" I suspect they've been doing a lot of talking to R & D people at high tech firms and designing their gear around the NEXT generation of technology. That way when their camera was ready, matching technology was readily available in the consumer market. \

 

Microsoft has been doing that for years, writing their next operating system for computer technology that was still in development. This is why you always have had to buy the latest, newest, baddest computer on the market to take full advantage of the latest Windows OS. When 3.1 came out, you'd better have a 486, when 95 came out, you had to get a Pentium I, 98 required a PII, etc., etc.

[Andrew Ray]

Calm down for a minute. Yes, there are "huge chunks of data" that compression removes, but they aren't so much missing as they are reinterpreted into a more efficient (but less accurate) form. JPEG, one of the most basic there is, works like this: http://en.wikipedia.org/wiki/Jpeg or by using a more color space that is less accurate but visually as accurate to the human eye and cosine transformation from 1X1 pixels into combinations of 8X8 pixel grids. To be honest, I only understand some of it...as you can probably tell! At high image quality, you can make a very large image (20mb as a TIFF) into a 2mb JPEG that looks nearly as good, except under close scrutiny.

 

Redcode similarly uses different color space (it compresses BEFORE transforming bayer into RGB) that looses a bit of color quality in favor of efficiency, and uses wavelet based compression which is much more efficient than JPEG (I think the grids can be of any size, for instance.) Unlike H264, HDV, MPEG2 and similar codecs, Redcore is INTRAFRAME meaning that fast motion will have no effect whatsoever. No matter how fast you move the camera, the image will hold up just as well.

 

But...wavelet codecs have problems with areas of high detail. If you gain the camera up very high (1600ISO or so) you will get more noise. If you shoot deep focus, your scene will have more high detail areas. From what I've seen from some tests, Redcode softened footage a tiny bit...but not enough to really matter. I would worry a bit about deep focus photography with lots of gain...but if you have enough light or shoot wide open you will probably be all set....

 

Of course, that's just speculation. But compressing isn't like scaling down an image and "repeating" data, it's like giving a very accurate summary of data, from which it can be reconstructed so accurately, you (ideally) won't notice any effects of compresison.

 

So don't sweat it, or just shoot miniDV!

Matthew, this compression doesn?t really matter since in most project you will use RAW port anyway. And what is coming from the raw port can print the same film copies that now DI process is printing.

So compression is just academic discussion here.

[Mark Williams]

Calm down for a minute. Yes, there are "huge chunks of data" that compression removes, but they aren't so much missing as they are reinterpreted into a more efficient (but less accurate) form. JPEG, one of the most basic there is, works like this: http://en.wikipedia.org/wiki/Jpeg or by using a more color space that is less accurate but visually as accurate to the human eye and cosine transformation from 1X1 pixels into combinations of 8X8 pixel grids. To be honest, I only understand some of it...as you can probably tell! At high image quality, you can make a very large image (20mb as a TIFF) into a 2mb JPEG that looks nearly as good, except under close scrutiny.

 

Redcode similarly uses different color space (it compresses BEFORE transforming bayer into RGB) that looses a bit of color quality in favor of efficiency, and uses wavelet based compression which is much more efficient than JPEG (I think the grids can be of any size, for instance.) Unlike H264, HDV, MPEG2 and similar codecs, Redcore is INTRAFRAME meaning that fast motion will have no effect whatsoever. No matter how fast you move the camera, the image will hold up just as well.

 

But...wavelet codecs have problems with areas of high detail. If you gain the camera up very high (1600ISO or so) you will get more noise. If you shoot deep focus, your scene will have more high detail areas. From what I've seen from some tests, Redcode softened footage a tiny bit...but not enough to really matter. I would worry a bit about deep focus photography with lots of gain...but if you have enough light or shoot wide open you will probably be all set....

 

Of course, that's just speculation. But compressing isn't like scaling down an image and "repeating" data, it's like giving a very accurate summary of data, from which it can be reconstructed so accurately, you (ideally) won't notice any effects of compresison.

 

So don't sweat it, or just shoot miniDV!

 

So your saying there are large chunks of information missing and this was what I said and it reconstitutes it why reiterate what I already said? The thing is this becomes more noticeable at larger sizes and the more information contained in the image and movement.

 

NOW fine if this can reconstitute to an accuracy where its undectable at projection sizes however this does seem unlikey and in the absence of any evidence at least nothing that can be downloaded and in the absence of being able to buy this equipment only preorder it by giving a 10% deposit that wont be taken until the thing is made if it is made as stated by them themselves then this is unproveable non existant far fetched and the science is to say the least a little skew wiff.. However what do I know quantum computers are proveably demonstrably workable.

[David Mullen ASC]

Yes, Kevin was right -- I'm referrring to RED competing with other cameras in the same price range, since I figure people with a budget of $20,000 to $50,000 to spend on a package are going to look in that arena.

 

People who were looking at a $4000 HVR-V1U are not going to need or afford a RED camera, and people for whom spending over $100,000 on equipment aren't necessarily looking for a bargain but probably have some specific format that they need to shoot for their specific market. If they need to string a bunch of Ikegami's for a live rock concert in 1080/60i with live switching, then they will probably go with what they are comfortable with.

 

I'm not going to make ridiculous predictions about RED versus film cameras; for people that love film and can afford to shoot film, there's no reason to not keep using film.

 

For people like me who rent cameras for feature productions (and now TV), RED will become another option to consider, so I'm curious to see what places like Clairmont do to the camera.

 

You have to understand that this particular moment in time, the presentation at NAB, is really a time to celebrate for the RED team after pushing so hard to get there with a working camera and something to show people. So I'm going to cut some of the more ardent fans some slack here, just as I'm not going to take anything too seriously that a fan says after a winning SuperBowl game. They are excited and now is the time for the marketing hype of the introduction to begin to peak and then die down as the camera gets released to the public over the course of the year. So I don't think we'll truly be able to gauge the long-term effects for awhile now. Anyone who makes sweeping predictions risks looking foolish in the future.

 

Obviously I'm going to eventually want to see some side-by-side comparisons to 35mm of the same subjects, both projected in 4K and both projected in film prints, etc.

 

And I'm sure someone is going to find ways to push the limit of the REDCODE compression to see where it gets stressed.

 

One thing to remember about all of this is that the RED camera will push all the other manufacturers to improve their products (and hopefully reduce their prices) so ultimately we all win whether or not we buy a RED camera.

[Patrick Neary]

>>Anyone who makes sweeping predictions risks looking foolish in the future.

 

Actually, a lot of them don't need to wait that long...

[Alessandro Machi]

Hopefully this whole Red thing will shake up the medicority of the big video technology manufacturers, when I went to Video Forum at Earls Court, London a few months ago - the Panasonic and Sony forum/talks were the most boring not to mention depressing things i've ever attended - like who cares if you record to tape or solid state, its quality, versatility and reliability are what matter.

 

Wow, I find the issue of tape versus solid state vs harddrive recording to be the biggest issue in the entire media spectrum, and will be for several years to come, unless you're a film guy, then it's probably irrelevant.

 

Thank goodness I quickly went into the Arri tent, their open minded 'can-do' spirit brought me back out of my commar. Its seems like Red is a more serious threat to Panasonic and Sony than to film companies like Kodak.

 

That's a very interesting way to look at it.

[Adam Frisch FSF]

For me personally, the two biggest benefits I see are:

 

1. If RED is what they say it is, it's going to ruin Sony and Panasonic's markets overnight. I personally hope it is all they say it is just so I can see them go down in flames on the pro camera side. They've screwed us over with news-cameras-sold-and-priced-as-film-cameras for long enough now.

 

2. Film cameras are going to drop in price and flood the second hand market. Which means I can buy a lot of them and continue to use film.