Jump to content

Film Vs Digital - Why Is Arri Alexa So Expensive


Recommended Posts

I was amused to see the Alexa described as "soft" compared to a DSLR!

There was a lot of fuss a while back when people put the canon 5D Mark III up on resolution charts and discovered it was "barely able to resolve 720p" (although the moire problems had reduced over earlier models)

 

Freya

Link to comment
Share on other sites

Perhaps another a question the young might ask is not so much "why not DSLR" but "why not film". If the best the digital best can do is "almost film" why not use film? Is it that film would cost even more? Depends on what kind of work is to be done. Some works work well done on film. Others do not. Some works will work out cheaper done on film. Others will not.

I agree with this post, the fact is that film cameras, expecially Super 16mm are increadibly cheap and plentiful, working with film does teach you so many skills which are usefull when working with digital such as 'creative discipline'.

 

Pav

Edited by Pav Deep
  • Upvote 1
Link to comment
Share on other sites

I was amused to see the Alexa described as "soft" compared to a DSLR!

There was a lot of fuss a while back when people put the canon 5D Mark III up on resolution charts and discovered it was "barely able to resolve 720p" (although the moire problems had reduced over earlier models)

 

Freya

 

People should not out too much stock on uneducated freelance writers who write utter nonsense such as how turning up the iso on a camera "lowers low contrast detail areas" turning up ISO inherently does that unless the sensor is warped. LoL

  • Upvote 1
Link to comment
Share on other sites

  • Premium Member

One point of view that I don't think has been aired is that Alexa actually isn't that expensive. If Red have done anything, it's to enormously moderate the prices of high-end digital cinematography gear. When F23 and F35 were current, Sony were routinely charging a quarter of a million units of currency for high end cameras. The fact that Alexa's price is expressed in five rather than six figures is an expression, to some extent, of the fact that the technology is becoming easier to do, but I suspect it's also related strongly to the politics and what people expect to pay.

 

So, actually, compared to the reality only ten or fifteen years ago, it isn't expensive - it's actually pretty cheap. The flipside of this is that if they're charging you less for the camera, they're going to need to sell you a new one more frequently. The pace of change has become enormously faster recently, and anyone buying a high-end camera today needs to be confident that it'll pay for itself in well under two years. Beware.

 

P

Link to comment
Share on other sites

 

People should not out too much stock on uneducated freelance writers who write utter nonsense such as how turning up the iso on a camera "lowers low contrast detail areas" turning up ISO inherently does that unless the sensor is warped. LoL

 

I wasn't thinking of any freelance writer in particular or for that matter, even freelance writers in general!

I get the impression you have somebody in mind! ;)

 

Freya

Link to comment
Share on other sites

I don't have a lot of time right now or else I'd probably get into chatting about all this!

I am intrigued as to what the film vs video angle is however? ...Or is that just great marketing! ;)

 

Freya

Link to comment
Share on other sites

I think the bottom line is that you should shoot with whatever camera that gives you the look you are wanting to achieve! :)

 

If you prefer the look of a Canon 5D over an Arri Alexa for instance then you should definitely go with that, and you should feel very happy that you can save so much money over working with an Alexa! I don't think there is any one right camera anyway, there are lots of different looks available from different cameras and I think that diversity is a good thing. If anything the dominance of certain cameras has caused far too much stuff to have aspects that look a bit same-y.

 

Freya

Link to comment
Share on other sites

One point of view that I don't think has been aired is that Alexa actually isn't that expensive. If Red have done anything, it's to enormously moderate the prices of high-end digital cinematography gear. When F23 and F35 were current, Sony were routinely charging a quarter of a million units of currency for high end cameras. The fact that Alexa's price is expressed in five rather than six figures is an expression, to some extent, of the fact that the technology is becoming easier to do, but I suspect it's also related strongly to the politics and what people expect to pay.

 

So, actually, compared to the reality only ten or fifteen years ago, it isn't expensive - it's actually pretty cheap. The flipside of this is that if they're charging you less for the camera, they're going to need to sell you a new one more frequently. The pace of change has become enormously faster recently, and anyone buying a high-end camera today needs to be confident that it'll pay for itself in well under two years. Beware.

 

P

Phil I think you have this exactly right. I am calculating 18 months with enough residual value to sell the camera and turn a reasonable profit - or forget it. The Alexa XT pkg will cost at least 115,000 and that is far less expensive than an Arricam LT or ST ever was but you sure better be able to see a return on that in less than 2 years or it could be a real money loser. A rental day now and then won't cut it.

  • Upvote 1
Link to comment
Share on other sites

  • Premium Member

I went to a Procam-sponsored event introducing Amira earlier this year and they reported that the orginal Alexas had paid for themselves in under a year - they do have an absolutely enormous industry profile and a huge reputation.

 

But then, that was Procam.

 

P

Link to comment
Share on other sites

Recently DSLRs got that magic lantern with RAW capability (really crappy post-flow) but definitely I think ProRes encoders built into the Arri with ARRIRAW offstorage support is super fantastic. And as you mentioned, I do notice the ARRI ALEXA has this serious film-like...(whatever this means - i kind of dont have a clue) quality - its softer..just looks like...a movie? Not sure if you can understand what I mean...I guess that has to do with their proprietary lenses...probably it costs more to make 'filmic' lenses than 'dslr' lenses....thoughts sir?

 

There are in my mind 2 essential features of 'high end' vs 'consumer' cameras, and 1 philosophical issue. The philosophical issue is 'who is the market, and what are they willing to do to the 'image' before it is presented'. DSLR cameras are made for people who want to get the image to presentation as quickly as possible, and so the output of the camer is directed to that goal. Even in stills, most people shoot JPEG, some may alternate between the 'raw' and JPEG... and a limited few may shoot exclusively 'raw'.

 

The high data rate has some engineering costs. If a 2Kx1K (near enough to HD...) sensor is 'read out' at 30 bits per channel per cell (taking RGB as a 'unit') that would be 1024x2048x30 would yield a 60 Mbit 'chunk'. But... for a 1/48 shutter speed, one needs to read out those bits in 0.020 seconds, this is a 3 GBit/s stream.

 

There are some non-loss compression techniques to reduce this data rate, but even so, the amount of electronics to support such a data rate, requires 'more powerful circuits', along with some sort of cooling...

 

This is why ARRI is using some 32 'readout' circuits, such that each path is less than this high data rate. From their tech brief, they also have 'high gain' and 'low gain' amplifiers on each path, and then through the magic of DSP combine these two in to a resulting 'high dynamic range' value for each pixel read.

 

Even after all is said and done, the data rate is still fairly high and beyond the usual SD card or the like type recording device. The camera manufacturers who cater to the high end market have recording devices that are very expensive, but 'do the job' for these rates.

 

Most DSLR's shooters are not going to spend $100s of dollars for single item recording devices...

 

The other option is high speed cables to an external recording device via say SDI interfaces... again not a typical DSLR interface. The closest one gets is the HDMI output. The HDMI output is not really a 'professional' connection. To be sure people use it, and many DSLR shooters are satisfied with the results, in terms of 'industrial' design, it is pretty crappy... ranging from contacts to no effective strain relief when connecting to the camera or recording equipment.

 

The other issue is sensor yield. Depending on the specs for sensors, ARRI may not have a 'great' yield for their sensors because of the close tolerances they set on quality. Whereas a DSLR manufacturer may 'accept' sensors that have a larger variation of response across the sensor, the higher end camera manufacturers would reject such a device.

 

My 'advice' would be rent the ARRI, buy a reasonable DSLR for 'learning/experimentation'.

 

If one were to spend the 80K just to get on to the ARRI bus... I'd strongly suggest setting up a rental company to rent it out to recoup the cost...

  • Upvote 1
Link to comment
Share on other sites

Hi Adrien,

 

That sounds like such a good description...I never thought the sensors were proprietary...I did look into the ALEXA page and they do tout the fact that the sensor is custom built with Arii Scientist "magic" technology or whatever. Not exact words...but haha, still.

 

Recently DSLRs got that magic lantern with RAW capability (really crappy post-flow) but definitely I think ProRes encoders built into the Arri with ARRIRAW offstorage support is super fantastic. And as you mentioned, I do notice the ARRI ALEXA has this serious film-like...(whatever this means - i kind of dont have a clue) quality - its softer..just looks like...a movie? Not sure if you can understand what I mean...I guess that has to do with their proprietary lenses...probably it costs more to make 'filmic' lenses than 'dslr' lenses....thoughts sir?

 

There's a LOT more going on than just Magic Lantern hackery.

 

The sensor isn't just a chunk of silicon. It has a LOT of circuitry on it in addition to the photosites and microlenses and Bayer pattern filters, and on top of that, other than in cameras, just about everyone does everything they can to make their chips SMALLER. Larger chips reduce yields because you can fit fewer of them on a wafer, you lose more due to defects in the wafers, and you lose more chips the more complexity you put into them... like the circuitry for pulling data off of the chip and transferring it to the camera's main system bus.

 

Then you have to account for the fact that an Arri Alexa can do 240fps with no compromise other than a shorter exposure time (i.e. no windowing like Red does), which requires a HUGE amount of bandwidth (one reason that Red uses it). Then on top of that, account for the fact that an Alexa allows you to build a color LUT on camera, which is a pretty processor intensive task... and it also has to be quiet enough to let you record sound on set while filming... it adds up!

 

And don't forget support.

 

If you're getting an Alexa, I'll envy you... I'm stuck with Black Magic. :)

 

They're great cameras for the money and they're enabling me to do phenomenal work, but their support is... lacking. It's a great starter camera though, a much better choice by far than a Rebel for casual filming, especially their Pocket camera.

 

Don't forget lenses though... none of those cameras are much good without good glass in front of them. :)

Link to comment
Share on other sites

 

Then you have to account for the fact that an Arri Alexa can do 240fps with no compromise other than a shorter exposure time (i.e. no windowing like Red does), which requires a HUGE amount of bandwidth (one reason that Red uses it). Then on top of that, account for the fact that an Alexa allows you to build a color LUT on camera, which is a pretty processor intensive task... and it also has to be quiet enough to let you record sound on set while filming... it adds up!

 

I'd agree with some of what you have written, but the use of a LUT does not add 'all that much'. There are several 'LUTs' being used in DSLRs. For one, is in 'white balance'. Another is most likely in 'shaping' of the sensor 'raw' data into the various 'profiles' that the camera may have.

 

The problem with the DSLRs often there is no way that a user can create a custom LUT for many cameras. There may be 'tweaking' of existing LUTs, etc.

 

Here the pseudo code for how raw data is transformed to output data via a LUT...

 

output = LUT[ raw ];

 

Here's the pseudo code for scaling and offesting via multiplies and adds...

 

output = raw*scale + offset;

 

(Examples simplified from R, G, B... but that would only mean something like:

 

output_red = LUT[ raw_red ];

output_green = LUT[ raw_green ];

output_blue = LUT[ raw_blue ];

 

Multiplies and adds are usually more processor intensive... unless the processor is in fact doing a lookup to produce the multiply or add results... not usually feasable...

 

Further, the LUT transformation is requires no more clock cycles than addressing the LUT element. Multiplies usually require several clock cycles.

Link to comment
Share on other sites

I forgot to also note, that a LUT operation can implemented in hardware with no processor involvement at all...

 

So the pixel data can pass through a chain of LUTs each one transforming the value according to some 'rule', and not ever be processed by the camera's micro processor, or DSP processor, again depending on the hardware implementation.

Link to comment
Share on other sites

  • Premium Member

I think you're really arguing semantics there.

 

Alexa is implemented almost entirely in FPGAs, according to Arri. They're expensive and power hungry, and are the sorts of devices that are used in most products implementing HD LUTs.

 

Some devices may do it in a specific bit of DSP, but really the FPGA is being used to implement DSP there. 3D LUTs can be very hard work, especially if you include a weighted interpolation between the stored cube points.

 

P

Link to comment
Share on other sites

Hi All,

 

An Update - I went ahead and bought the Arri Alexa...it was a lot more than I bargained for. A lot of people are mentioning recouping the costs...I'll be honest in saying I just happened to get this much money to buy just the kit from AbelCine - but not the lighting etc (worked out a deal on this). Also the fact that I'm not taking film professionally...it is just a hobby. I wouldn't even know how to start renting out my camera...or that I could even afford that type of risk

 

I am however stuck...I don't know which lens to use.

 

I want a more FILMIC look - so I have been looking at the 'anamorphic' lenses. Apparently they capture a wider FOV (field of view) than spherical lenses. Is this true? I am horrible when it comes to aspect ratio..I saw videos on youtube...but no one quite explains what lenses are.

Link to comment
Share on other sites

  • Premium Member

I want a more FILMIC look - so I have been looking at the 'anamorphic' lenses. Apparently they capture a wider FOV (field of view) than spherical lenses. Is this true? I am horrible when it comes to aspect ratio..I saw videos on youtube...but no one quite explains what lenses are.

I am amazed at how someone would spend so much to try to get the filmic look when you could have easily got the real thing by purchasing a true 35mm kit and LOADS of stock.

 

What a waste.

  • Upvote 1
Link to comment
Share on other sites

Congratulations, but, to be honest I don't get why you bought an Alexa if you're not a professional. There are a number of other professional camera options which would leave you money for lights, lenses and other accessories, examples would be the RED Epic or Scarlet, Sony F55, a number of hobbyists have bought these.

 

Anamorphic lenses have a wider field of view for the focal length being used by squeezing the image horizontally, you reverse this during projection. .Downside, they're larger, heavier, more expensive, often not as wide an aperture http://en.wikipedia.org/wiki/Anamorphic_format

 

You can get wider field of view spherical lenses,

Link to comment
Share on other sites

You might have been better off with an Amira if you are just going to shoot to SxS cards as it has the same sensor.

I would shoot fast and sell it on quickly.

If I had infinite money and wanted a "filmic" look I would probably shoot on Cooke S4's. Maybe Cooke S4 mini's if I didn't need the extra stop or so. They are beautiful lenses that have been used on a lot of feature films and they are cheaper and easier to work with than Anamorphics.

 

Unless you have a particular desire to shoot in cinemascope aspect ratio?

 

Freya

Link to comment
Share on other sites

  • Premium Member

Plenty of movies have been shot on film with spherical lenses...

 

The most common anamorphic lenses have a 2X horizontal squeeze that expands the FOV horizontally by 2X. However the image has to be unsqueezed at some point, doubling the width of the image, so if your recording is 16x9 HD, for example, which is a 1.78 : 1 aspect ratio (just divide 16 by 9) then doubling the width to get rid of the squeeze gets you a 3.56 : 1 aspect ratio. Generally the widest aspect ratio currently used for movies is 2.40 : 1. Working backwards, with a 2X anamorphic lens on the camera, this means you end up only using a 1.20 : 1 area of the sensor or recording for a final 2.40 unsqueezed image.

 

Some Alexa's allow you to record the squarer / taller area of their 4x3 sensor above and below the 16x9 area normally used, but that may only be in ArriRaw. Otherwise if you record 16x9, then in post you'd have to crop the sides to get the 3.56 : 1 unsqueezed image back to 2.40 : 1.

  • Upvote 2
Link to comment
Share on other sites

Plenty of movies have been shot on film with spherical lenses...

 

The most common anamorphic lenses have a 2X horizontal squeeze that expands the FOV horizontally by 2X. However the image has to be unsqueezed at some point, doubling the width of the image, so if your recording is 16x9 HD, for example, which is a 1.78 : 1 aspect ratio (just divide 16 by 9) then doubling the width to get rid of the squeeze gets you a 3.56 : 1 aspect ratio. Generally the widest aspect ratio currently used for movies is 2.40 : 1. Working backwards, with a 2X anamorphic lens on the camera, this means you end up only using a 1.20 : 1 area of the sensor or recording for a final 2.40 unsqueezed image.

 

Some Alexa's allow you to record the squarer / taller area of their 4x3 sensor above and below the 16x9 area normally used, but that may only be in ArriRaw. Otherwise if you record 16x9, then in post you'd have to crop the sides to get the 3.56 : 1 unsqueezed image back to 2.40 : 1.

 

Hi Mr. Mullen

 

This is what confuses me - do the lenses have a wider FOV? Because I don't understand quite how squeezing the image would actually lead to a more wide aspect ratio which does not look badly stretched out without the lens actually capturing more horizontal detail over spherical lenses (wider FOV) As you mentioned that a lot of films were shot on spherical....but I am essentially trying to get close to film (despite people having a heartattack I didn't just get actual film stock, etc.) I want an updated digital workflow for my small-time production. And I don't want to add black bars in post to fake the filmic aspect ratio look (losing vertical resolution detail).

 

"Working backwards, with a 2X anamorphic lens on the camera, this means you end up only using a 1.20 : 1 area of the sensor or recording for a final 2.40 unsqueezed image."

 

Based on this, the way my brain is interpreting this is that if you have a 16:9 image...and use an anamorphic lens - you get a much bigger capture area than most films? I think I probably have this insanely wrong. Awaiting your response and I thank you for your time. I have read several of your posts and am glad you decided to reply to mine.

 

Thank you,

 

Michael

Edited by Michael Landon
Link to comment
Share on other sites

Some Alexa's allow you to record the squarer / taller area of their 4x3 sensor above and below the 16x9 area normally used, but that may only be in ArriRaw. Otherwise if you record 16x9, then in post you'd have to crop the sides to get the 3.56 : 1 unsqueezed image back to 2.40 : 1.

Then, you'd have a pixel count just slightly higher than cropped 1080. :D Better colors, though!

 

Fun topic, but has anyone considered that the original poster might be trolling?

 

-Carl.

 

It could be but just being on this forum is awfully niche...

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...