RED vs DALSA ORIGIN [Archive]

Felipe Jaco

10-26-2007, 07:37 AM

Graeme Nattress

10-26-2007, 07:53 AM

Bit depth != dynamic range. A higher bit depth can be necessary, in a linear system to get a higher dynamic range, but it does not in itself imply a higher dynamic range. Just think, you can put some VHS in a 16bit tiff, but it doesn't mean that because it's 16bit that the image has a high dynamic range....

4k makes a better looking 2k than a 2k camera, so that resolution is a great advantage. Also allows you to zoom and reframe without loosing detail.

I'd say the biggest difference between the two cameras occurs when you try to lift the Dalsa....

Graeme

Michael Schrengohst

10-26-2007, 08:13 AM

I'd say the biggest difference between the two cameras occurs when you try to lift the Dalsa....

Graeme

LOL!! - How about useability as a factor??
If it takes an army and a scientist to make images
then the 16 bit argument becomes moot.

I Bloom

10-26-2007, 08:25 AM

Is latitude and dynamic range more important than resolution? I mean if the final film print that we see in the theater is 2K then what is the point in having camera which produces images with resolution 4K and above.
DALSA ORIGIN has 16bit image and RED 12bit so it has better dynamic range. The resolution is about the same but is that 4bit defference significant in terms of closer to film quality?
Or the differnce in price beats everything?

Dalsa has a 16bit DAC, but... most actually crop that to 14 bits before storage because the extra two bits are lost in the noise floor. Here is what I can tell so far in terms of comparison:

Dalsa
16 bit, possibly more Dynamic range
Some say CCD is a better sensor, I say show me the money.
Optical Viewfinder
No rolling shutter artifacts
Some compression but lossless
Dalsa has anamorphics for their sensor.

RED
12 bit DAC
Lighter Weight, Modular design and formfactor, lots of fun accessories.
Affordable, available for purchase.
Redcode Compression allows for manageable onboard storage CF + Red Drive
Redcode also allows for proxy direct editing of source material.
Final Cut and Quicktime integration and RedCine
Oakley marketing genius, RED team personality, RedUser community.

Dalsa is great, and a really great company... certainly they've made their mark on the sensor world, on Earth and Mars! But they kind of missed the boat about three years ago with their moves on Digital Cinema. I'd like to see Dalsa instate a major change in the way they're doing bussiness with the film world and forget about the word "high end".

Right now I'm more interested to see what Silicon Imaging does in competition with RED.

IBloom

Kevin Halverson

10-26-2007, 08:55 AM

Is latitude and dynamic range more important than resolution? I mean if the final film print that we see in the theater is 2K then what is the point in having camera which produces images with resolution 4K and above.
DALSA ORIGIN has 16bit image and RED 12bit so it has better dynamic range. The resolution is about the same but is that 4bit defference significant in terms of closer to film quality?
Or the differnce in price beats everything?

How a resulting image is formated (bit depth) can be completely independent of the actual dynamic range. Just because the DALSA or any camera for that matter, chooses to use a 16 bit quantizer (and file format), does not necessarily mean that its sensor's analog performance justifies the data set used. In fact, a larger dynamic range could be represented by a smaller number of quantizer steps (bits) just as easily as it could be represented by a larger number.

In the end, trying to make a determination of the performance based solely upon the published numeric data accomplishes little more than playing into the hands of the marketing entity. Real testing, performed in a controlled environment, using proven testing regiments is the only thing that can reveal actual obtainable performance.

The old saying that "all it takes to get great specs is a typewriter" can just as accurately be transfered to modern products by the substitution of word processor for typewriter.

Felipe Jaco

10-26-2007, 09:02 AM

I can't wait when will somebody make shootout tests in controlled inviroment with RED and other current competitors as Dalsa.

Michael Stanmore

10-26-2007, 09:49 AM

Right now I'm more interested to see what Silicon Imaging does in competition with RED.

IBloom

Hmm, yeah I was wondering about that too...

Michael Lindsay

10-26-2007, 10:01 AM

Hi ibloom

Couple of questions:

Dalsa has a 16bit DAC, but... most actually crop that to 14 bits before storage because the extra two bits are lost in the noise floor. Here is what I can tell so far in terms of comparison:

Dalsa
16 bit, possibly more Dynamic range
Some say CCD is a better sensor, I say show me the money.
Optical Viewfinder
No rolling shutter artifacts
Some compression but lossless
Dalsa has anamorphics for their sensor.

How is the file truncated to 14bit? What file format?

What kind of mechanical shutter does it employ?.. I'd always assumed a mechanical shutter had 'some' rolling shutter artifacts..

Also why would Dalsa's anamorphics not work on Red?

Thanks

Michael

Graeme Nattress

10-26-2007, 10:43 AM

Jeff Kilgroe

10-26-2007, 11:02 AM

IMO, Dalsa and RED are offering two different products to two different market segments using two very different business models. Technical issues aside, there are several business-related reasons for someone to choose one camera vs. the other. RED is positioned in a good price range for many independent operators where they can own the camera. Dalsa is rental only, so for people who try to shoot nearly every day, Dalsa doesn't make a whole lot of sense.

Other issues like the Origin's size and lack of modularity don't help it either. You're not going to be seeing one or two guys hiking off into the wilderness for a few days with a Dalsa Origin.. Oh, I suppose it could be done, but it's a very purpose-specific camera. It's best relegated to cine-style production with an adequate crew.

Dalsa's anamorphics only provide something in the neighborhood of a 1.25:1 horizontal compression, I don't think it's even a 1.33:1. Their purpose is to squeeze up to 2.40:1 out of a 2:1 sensor without sacrificing vertical resolution. Due to the very low horizontal compression, you're not going to get the elongated blue flaring or the stretched bokeh that you would get with normal 2:1 anamorphic glass. The flaring and bokeh are the reasons most would want to shoot anamorphic with something like RED. When we start getting up into 4K resolution, the benefits of horizontal compression just for the sake of resolution have become somewhat moot.

Soderbergh is shooting anamorphic with RED. Here's a thread talking about it: http://www.reduser.net/forum/showthread.php?t=4932&highlight=anamorphic

Steve Gibby

10-26-2007, 11:05 AM

I'd say the biggest difference between the two cameras occurs when you try to lift the Dalsa....

...and when you try to buy the Dalsa...

I Bloom

10-26-2007, 12:11 PM

Hi ibloom

Couple of questions:

How is the file truncated to 14bit? What file format?

What kind of mechanical shutter does it employ?.. I'd always assumed a mechanical shutter had 'some' rolling shutter artifacts..

Also why would Dalsa's anamorphics not work on Red?

Thanks

Michael

No idea, about the file format. This was told to me by a RAID vendor at Cinegear.

Mechanical shutters have soft rolling shutter artifacts, digital are hard.
http://www.redhax.net/wiki/Understanding_Rolling_Shutter_Artifacts

The anamorphics would work.

IBloom

Peter McCully

10-26-2007, 12:25 PM

Dalsa's anamorphics are semi-scope, so they only stretch their 2:1 to 2.35:1 or so. You can use normal anamorphics on RED if you want as I'm sure someone is doing at this very minute....

I'd recommend people take a look at the images on the Dalsa website, open them up and pixel peep.

Graeme

I have done that and I think the sharpness and color of the Dalsa Image is incredible. And so it should be. I don't think we have anything to gain by trying to score points against it. They are chalk and cheese, these two systems. Red is incredible too and we all know the reasons why.

tj williams

10-26-2007, 02:02 PM

Yea when you lift it. I offered to put it on my shoulder at last years NAB, the guy kinda looked funny at me. I wondered if He didn't think I could lift it??

Felipe Jaco

10-26-2007, 04:11 PM

Alexander Nikishin

10-26-2007, 04:15 PM

RED will always be around as long as there are a team of talented individuals and a mastermind willing to create the next big thing.

Mathieu Ghekiere

10-27-2007, 01:02 AM

Does anybody know how long will RED camera be avaliable on the market, I guess untill there is demand that is payable for Jim.I mean will he run factory if there is demand of 20 cameras per year. Are there any teories of how much time will be needed for market to get saturated? But no. On a second taught , this just a beginning, when the world sees the quality of RED I think Jim will have to do a 'little' expandation.

I think if people use RED, RED will stay on the market.
They will make their camera better, add more features, make more lenses, make displays, 4k projector's, other camera's... They aren't RED Digital Camera Company anymore (if they ever were), but RED Digital CINEMA Company.
So don't be worried...

Michael Lindsay

10-27-2007, 02:29 AM

No idea, about the file format. This was told to me by a RAID vendor at Cinegear.

Mechanical shutters have soft rolling shutter artifacts, digital are hard.
http://www.redhax.net/wiki/Understanding_Rolling_Shutter_Artifacts

The anamorphics would work.

IBloom

Thanks Ibloom

Sorry I was a little disingenuous with the last too questions... You are of course right about hard/soft but it is worth pointing out.. As for Dalsa's anamorphics? I think if we could get 4520x2260 on Red with Dalsa's lenses it would be an interesting scope option.

As for the file format GN may have some clues? It was why you would cause bandwidth challenges to only truncate or dither down in post to 14bit I was confused about...

thanks and sorry

Michael Lindsay

Graeme Nattress

10-27-2007, 06:30 AM

http://www.dalsa.com/shared/content/PDFs/Origin_Launch_NAB_2003_April_7.pdf is one of the only pieces of info about Dalsa I have - stating 14bit a-to-d and other stuff. It's way out of date, but Dalsa don't seem to have updated specs anywhere. Take it for what it is, an interesting historical document.

Graeme

Chris Nuzzaco

10-27-2007, 07:00 AM

Hey Graeme,

I was talking to a engineer friend of mine about sensors and ADC's and he mentioned something to me that I never really thought about. He said that even though you have two equal performing analog sensors, with ADC's running at the same bit depth, they can have radically different performance simply due to the quality of the ADC. Very interesting stuff, and it makes a lot of sense too. That ADC's noise floor needs to be as good as or even better than the sensor it's sampling from. Really makes me wonder how good the sensors in "lower" performing cameras really are.

What costs more? A good sensor, or a good ADC?

Thanks!

Graeme Nattress

10-27-2007, 07:04 AM

I don't know costs, but with CMOS, the ADC is built into the sensor, whereas in CCD, it's seperate. And yes, you can ruin the image by a poor ADC on a good sensor, or you can waste your effort putting a really good ADC on a poor sensor.

Graeme

Chris Nuzzaco

10-27-2007, 07:08 AM

you can ruin the image by a poor ADC on a good sensor, or you can waste your effort putting a really good ADC on a poor sensor.

Graeme

Well said!

Chris Nuzzaco

10-27-2007, 07:09 AM

Would it be more advantageous to not build the ADC into the sensor? I'm curious about that design decision actually, I never knew that.

Kevin Halverson

10-27-2007, 08:29 AM

What costs more? A good sensor, or a good ADC?

Hi Chris,

I know that you addressed this question to Graeme, but I thought I would give you my perspective on the subject as I have been involved in manufacturing of hardware for nearly 30 years. In general, there are two things that drive the costs of any Si which utilizes the same process. The first is die area (size) and the second is yield. Generally, the die area of a sensor is much greater than an ADC so this would push the cost of the sensor. Yield is a more complex concept to fully understand as it is somewhat dependent upon the task that the Si is expected to perform. Being that photosite sensor array can have a defect and still be considered usable (depending upon the manufacture res expectations of performance), this gives back a bit to the costs of the sensor as you don't have to have zero defects if you can map out bad photosites.

There are numerous techniques for digital to analog conversion and each effects performance and costs. Considering the large number of conversions that need to be performed and how quickly the converter needs to settle (given the high photosite count) this would suggest that one would trade off conversion accuracy for speed in this type of application. Higher speed can be achieved by several approaches, but an obvious one would be to have a high number of converters operating in parallel, this approach would quickly increase the converter costs. There are also trade offs to consider such as how much post fabrication (packaging) expenses for things like high pin count packages versus the desire to increase the number of channels between the sensor and ADC.

In the end, its speculation, but generally, I would estimate that the sensor is more costly, but given the task, the ADC wouldn't be too far behind. Add to this the additional PCB area and labor and it can be seen that the separate sensor / converter approach is going to be considerably more expensive to manufacturer. Development costs likely favor this approach, but how these get factored in is difficult to predict as it will be effected by the business model and market segment.

Graeme Nattress

10-27-2007, 08:35 AM

Would it be more advantageous to not build the ADC into the sensor? I'm curious about that design decision actually, I never knew that.

CMOS sensors have the ADC on-chip. That's one of the key differences between CCD and CMOS.

Graeme

Omnius

10-27-2007, 09:20 PM

Does anybody know how long will RED camera be avaliable on the market, I guess untill there is demand that is payable for Jim.I mean will he run factory if there is demand of 20 cameras per year. Are there any teories of how much time will be needed for market to get saturated? But no. On a second taught , this just a beginning, when the world sees the quality of RED I think Jim will have to do a 'little' expandation.

Who is this guy? :unsure: :huh: :umm:

Do I detect a case subject with an anti-red borderline personality?:detective2:

Take the RED pill, it'll make you feel better in the morning