DALSA Evolution

[Brian Drysdale]

Correct! Now we don't have any problems :-) I thought you were intentionally ignoring me when you kept listing this symptom. Well, it's actually more complicated, because all lenses are brighter in the center and get dimmer as you move out to the edge, with digital, film, or Daguerrotype plate.

 

This is why you don't want to shoot lenses wide open when you have the stop down one or two stops from wide open. I forget the Physics behind it at the moment, but it works. I try to print all of my pictures with the enlarge lens stopped down two or three, at the very least one stop. Some digital sensors exhibit further problems though, with weird color shifts and artifacts at the edges, which go beyond just vignetting toward the edges.

 

No, I was just taking a different line of thought regarding extra thickness of the film emulsion layers compared the single sensor, in that focus pullers were noticing less of a focus "fudge factor" with the single sensors.

[Gary McClurg]

Okay... how was the show from Dalsa the other night...

 

On the new Evolution camera... did they say how much rentals would be... I could be wrong but I was under the impression that this was a smaller camera... or dare I say a cheaper camera...

 

Also any clue to the cost of work flow... in the post realm...

 

Also I know this sounds crazy... but I'm not so worried about 4k as working in a good or better 1080p mode... I work on smaller productions... the ones that end up on Sci Fi Channel... etc...

Edited April 11, 2007 by Gary McClurg

[David Mullen ASC]

It's too early to say what the rental would be, but I doubt cheaper -- smaller 35mm sync-sound cameras aren't any cheaper to rent than the larger 35mm ones afterall.

 

The 4K workflow is still a complex issue which basically right now, Dalsa handles in-house for the client... but I overheard a suggestion from a DP about how Dalsa could get "out of the post business" by partnering maybe with the right company.

 

The camera does an internal downconversion to 2K which apparently looks great due to oversampling, and a 2K post is quite common in Los Angeles.

 

The first feature to use the Dalsa Origin is just starting, and it's mainly a Viper shoot with certain parts being shot on the Dalsa Origin.

 

My feeling is that they should buy some RED cameras, some Phantoms (which they have one there) and create a bigger rental inventory.

[Andrew Ray]

Thank you for taking the time Karl, Lance, Max, Brian, and explaining it here, I really appreciate. I hope that we are not hijacking the thread, but it is Dalsa that decided to go custom made lenses that triggers this discussion.

Max, I din't know that ARRI 65mm are rehoused Zeiss medium format!

Is Master Prime 65mm rehoused as well?

 

Karl, last month I read a lot of material along your lines of reasoning.

Dalsa has interesting article there.

http://www.dalsa.com/dc/documents/Image_Re...PTE_37_2003.pdf

 

I am still not clear on the combined resolving power of film/sensor and lenses.

From their calculations it looks that as long as the lenses have double resolving power as compared with sensor/film you should get close to 90% of the rated resolution at 50% modulation.

Also I hope that the new Dalsa camera will have more then 8MP sensor on it.

 

David, did they said anything about the sensor on the new camera during the presentation yesterday?

 

What we did here, we tested lenses with the 100% modulation and once selected we use D20 to get some test shots at 2MP (ARRI D20 is 6MP Bayer sensor) at 50 and 10% modulation. Test charts are B&W. In color, the performance is almost the same providing that we have green content in the test charts. Shooting test without the green brings all tests significantly down.

We used this charts for 50% and 10% modulation with different colors.

http://www.normankoren.com/Tutorials/Lenst...86p_15g_0is.png

 

Everything looks acceptable for 2K format. When we extrapolate the tests to 4K nothing makes sense, unless lenses get better rating in Lp/mm.

 

Anamorphic lenses in terms of resolving power, give you bit better performance and extra light in the frame but it is like squeezing the already squeezed lemon again in desperation.

I do not discuss artistic performance of anamorphic lenses here.

Somehow anamorphic lenses it is not elegant way to go especially from engineering point of view.

Edited April 11, 2007 by AndrewR

[Brian Drysdale]

Perhaps you should factor in that the resolution of a 4k Bayer pattern sensor isn't actually 4k. From memory, in theory, it's 0.7 of that figure: 2.8k.

 

If I'm wrong, no doubt, I'll be corrected on that figure.

[Max Jacoby]

Max, I din't know that ARRI 65mm are rehoused Zeiss medium format!

Is Master Prime 65mm rehoused as well?

I'm not sure I get your question, but the lenses Arri has for the 765, their 65mm format camera, are rehoused medium format lenses that come in a Maxi-PL mount. Like I said they are optically not as good as 35mm lenses (and are slower too, with stops between T2.1 and T4.2), but they don't have to be, because, as Karl has explained, the bigger negative compensates for that.

 

The Master Primes are designed from the ground up and are 35mm spherical lenses. Technically speaking they are the best lenses available anywhere. They include all the latest advances in lens design and as tests have shown are sharper at T1.3 than any other lens at T2.8. If you are not happy with them, too bad, you won't find better ones.

 

Also I hope that the new Dalsa camera will have more then 8MP sensor on it.

That sensor still gives you a 4K Bayer image, is that not big enough for you? The chip itself is already bigger than 35mm neg, cramming more pixels in there would either increase the size even more or reduce the performance.

 

Everything looks acceptable for 2K format. When we extrapolate the tests to 4K nothing makes sense, unless lenses get better rating in Lp/mm.

What do you mean by extrapolate to 4k? The D20 has a 3K Bayer chip, I don't see how you can get a 4K image out of that.

 

I honestly don't see where your problem with current lenses is. 35mm film has more than 4K resolution and it does just fine with the existing lenses, so it should be good enough for a 4K Bayer chip as well.

 

Also Andrew, please use your full name, as per forum rules.

[Mitch Gross]

the lenses Arri has for the 765, their 65mm format camera, are rehoused medium format lenses that come in a Maxi-PL mount.

 

I believe it is referred to as the Super-PL mount.

[K Borowski]

Karl, last month I read a lot of material along your lines of reasoning.

Dalsa has interesting article there.

http://www.dalsa.com/dc/documents/Image_Re...PTE_37_2003.pdf

 

THat's actually an even more-detailed description of lense resolution than I dwelt on. It even takes into account the vibrational tendancies of light because it is a wave not actually a beam following a perfectly straight line. Basically though, even in the article they state that they're assuming an "ideal" 5.6 lens, meaning one that probably has image quality loss. Frankly that's unrealistic.

 

You guys are all correct: when you try to get up past say 10 MP of image resolution, especially if you're using something smaller than 65mm, you basically need a master lens craftsman to custom-mold you a lens, and not use any filters on it. I was crunching the numbers, for fun one day when I was trying to figure out a way to get a 16MP file from scanned slide film. Even with the cripsest of Zeiss primes, I just couldn't figure out a system that would break the barrier.

 

Anyway, as far as the inverse proportion rule I've given you for an optical image chain, I'm not saying that there aren't ways of improving lenses so that they can get very very near to lossless, but we won't be there for a long time, at least to the point where these lenses will be commercially, affordably (relatively speaking of course) available.

 

Haha, and people bitch about grain. They oughta be lamenting using an outdated material like glass ;-)

 

I actually wish there were information out there on the popular DSLRs available today. I'm sure that 16 megapixel sensor is not resolving anywhere near that much after you take into account lens distortion.

 

Of course, keep in mind that, even with subpar lenses, most people don't see the difference. These are things only filmmakers and (good) photographers would even talk about. As long as your lens isn't made of a piece of plastic chances are it will do a good job at producing photographs in a wide array of environments.

 

Regards,

 

~Karl

[Andrew Ray]

What do you mean by extrapolate to 4k? The D20 has a 3K Bayer chip, I don't see how you can get a 4K image out of that.

 

I honestly don't see where your problem with current lenses is. 35mm film has more than 4K resolution and it does just fine with the existing lenses, so it should be good enough for a 4K Bayer chip as well.

 

Also Andrew, please use your full name, as per forum rules.

 

Max, 65 mm I was referring to the Master Prime 65mm focal length ARRI/Zeiss lenses.

I think Master Prime are aspherical not spherical.

See here:

http://www.arri.com/prod/cam/master_primes/mp_articles.htm

 

Are these lenses re-build?

Maybe I expressed myself ambiguously so we have to differentiate between 4K format and Mega Pixels needed for 4K format.

 

4K format is derived from the fact that it has 4k horizontal pixel count.

Now in 2,35:1 screen size it will need about 12MP on Bayer sensor in good color and pattern proportion in the content of the picture and about 20MP in adverse content, like lack of green colors or fine repeating patterns.

So we can have many different quality grades of 4K Bayer sensor, depends from the MPs the 4K Bayer sensor have. Also Dalsa 4K camera has only 8MP.

 

I agree with you that 35mm film has more than 4K resolution as a master copy and only if it is done using highest quality film that is normally not used for most of the shoots.

 

When it comes to the final copy that is watched by the viewers in the cinema, it doesn't come even close to 4K since the DI copy process is based on 4K sensors (ARRI) and the process require the light to pass 3 times through the lenses.

Once when we are capturing the image (filming) second when we are scanning, third when we are projecting in the cinema. So out of 3 passes via 4K quality lenses/copy/display you end up with hardly 2K full quality. Other not DI processes are even more loss introducing processes.

 

Now in digital 4K camera/presentation process, the lenses are used only once, during the acquisition.

From this point on, especially when the material is watched on the non projecting digital screen, the quality of the material is mathematically lossless. (zero degradation)

 

I understand that only few cinemas has ability to show full 4K material, both analog or digital way now, but our material is not news/broadcast category, so it may have more then 10 years presentation life. We know that full resolution 4K format storage of our material will have much higher value 5 years from now then 2K one. How much you will pay for 10 years old, low res good story/acting films versus, high res high quality good story/acting?

 

10 years from now most American households will have 4K projectors/screens competing with the cinemas and cinemas will be fully digital.

Who would want to watch these soft 2K formats then, it will look old.

 

Andrew Ray

Edited April 11, 2007 by AndrewR

[Max Jacoby]

Max, 65 mm I was referring to the Master Prime 65mm focal length ARRI/Zeiss lenses.

I think Master Prime are aspherical not spherical.

 

Are these lenses re-build?

In the case of cinema lenses 'spherical' is used to designate a type of lens that, unlike 'anamorphic' does not squeeze the image. In lens design terms, 'aspherical' and 'spherical' refer to types of glass elements and most modern lenses contain a combination of both.

 

Also when you are messuring the lines/mm of lenses you will find that longer lenses have more resolution. That is due to the fact that the longer the lens the narrower a field of view it needs to cover, so the rays of light do not need to be bent as much before they reach the filmplane/sensor. Wide angle lenses are harder to design on the other hand and a focus-fall off towards the edges it expected. It's just the laws of physics.

 

The Master Primes are not rebuild, they are designed from the ground up, as your link clearly states.

[David Mullen ASC]

4K format is derived from the fact that it has 4k horizontal pixel count.

Now in 2,35:1 screen size it will need about 12MP on Bayer sensor in good color and pattern proportion in the content of the picture and about 20MP in adverse content, like lack of green colors or fine repeating patterns.

So we can have many different quality grades of 4K Bayer sensor, depends from the MPs the 4K Bayer sensor have. Also Dalsa 4K camera has only 8MP.

 

I'm a little confused as to what you're saying because you seem to be mixing up a couple of concepts. As for the "is 4K Bayer-filtered truly 4K" question, if you believe that it isn't, you'd need a sensor with more than 4000 pixels horizontally.

 

But if that wasn't what you are saying, then you only derive 12MP from 4K when it's a 4x3 image (4000 x 3000 pixels), not a 2.39 image, unless you are talking about a 4K scan of 2X anamorphic photography.

 

A 4K scan of a Super-35 frame that is cropped to 2.39 for film-out to 35mm anamorphic (or for digital projection at 4K probably) is 4096 x 1713, which is 7.02 megapixels.

 

A 4K scan of a 35mm 2X anamorphic frame would probably be 3656 x 3054 (that's a guess, based on the fact that in a 2K scan, it's only 1828 pixels across for anamorphic/Academy/1.85). That's a 9.3 MP frame.

 

So the Dalsa Bayer-filtered CCD, being around 4000 pixels across and 2000 pixels vertical, 8 MP, is certain in the same ballpark as those figures. But again, that's assuming that you don't want to increase pixel count in both directions to compensate for de-Bayering resolution loss. But that's a separate issue. I was just trying to figure out where you got 12MP for a 4K scope image. Usually that figure is tossed out for 4x3 images.

[Gary McClurg]

I hope you guys cheat and don't keep all these numbers in your head... :D

[Max Jacoby]

I hope you guys cheat and don't keep all these numbers in your head... :D

I am sure David doesn't cheat.

[K Borowski]

Max, 65 mm I was referring to the Master Prime 65mm focal length ARRI/Zeiss lenses.

I think Master Prime are aspherical not spherical.

See here:

http://www.arri.com/prod/cam/master_primes/mp_articles.htm

 

Now in digital 4K camera/presentation process, the lenses are used only once, during the acquisition.

From this point on, especially when the material is watched on the non projecting digital screen, the quality of the material is mathematically lossless. (zero degradation)

 

I understand that only few cinemas has ability to show full 4K material, both analog or digital way now, but our material is not news/broadcast category, so it may have more then 10 years presentation life. We know that full resolution 4K format storage of our material will have much higher value 5 years from now then 2K one. How much you will pay for 10 years old, low res good story/acting films versus, high res high quality good story/acting?

 

10 years from now most American households will have 4K projectors/screens competing with the cinemas and cinemas will be fully digital.

Who would want to watch these soft 2K formats then, it will look old.

 

Andrew Ray

 

Woah, you're starting to speak deleteriously here. It took over *50* years to finally surpass just standard definition televion with HD. The number of households that have HD is still astronomically low. Care to wager 10 years for HD, since it still hasn't taken over yet? Howabout 20? Don't think digital's going to come in and fu**ing turn things upside down like it did in still photography. Remember that making a film is totally different than shooting senior photos or a football game in terms of planning. Also money is obvioulsy not an object. Frankly, not being compatible with fast 35mm lenses is a HUGE disadvantage for this camera, IMHO. This one aspect could easily relegate it to something that is only used for art films, or omnimax films. Filmmakers, especially nowadays, desire flexibility above all else. Tell them they can only use your two hand-crafted lenses that just bearly eek 90% of the resolution that goes through them onto the film without distortion, and they'd probalby laugh at you. This is to me basically a digital equivalent of VistaVision. Now sure, there are plenty of 35mm still lenses that would almost certainly cover the frame of this camera if it's only 8MP, but what about the whole megapixel barrier? There's no way you're breaking through 16 if it's only an 8 MP chip to begin with, and again without glass that basically doesn't exist within the realms of reality for most people, you'd probalby be lucky if you broke a true 6. That would certainly rival 500T film (It'd certainly outdo Techniscope ;-) ).

 

Hell, even the GRAIN aesthetic is still really popular on TV. If anything, by your thinking, they should take The Shield off the air because it deliberately reduced the quality of its imagery by inadequately lighting and not utilizing tripods effectively, and using ZOOMS?! Das ist tottally verbotten.

 

Just remember you can make utter sh it on regular 8mm as easily as you can with omnimax. If you're a quality nut (take this in the light-hearted manner I intend, I'm one too, just on the photochemical side of the war), now is NOT the time to be getting into cinematography, because frankly, using one prime and a tripod in every shot would get you laughed at and get you nowhere in the boxoffice. Watch what they have on MTV for a day and then imagine what that particular audience values as opposed to what you do.

 

Take care.

~Karl

[jan von krogh]

The number of households that have HD is still astronomically low. Care to wager 10 years for HD, since it still hasn't taken over yet? Howabout 20?

i do agree that 10-20 years might be a realistic timeframe for the last SD-TV sets to be turned off in the industrialized countries, rather 15-20 imho.

 

the broadcast stations & productionhouses who want to produce certainly have less than 10 years, the ones who dont have hd yet already have to catch up.

 

however, new sets and new sales is a different story - i cannot predict the future, but i would be surprised to still see sd-home TVs in 2010 as mayority, especially because the price difference in flatpanels is basicly nullified. also, i assume that trendleading movies won´t be reaching #1 sales anymore, when only available in sd within very few years from here. dvd just recently showed, how fast consumers buying behaviour change those years.

 

Don't think digital's going to come in and fu**ing turn things upside down like it did in still photography.

i try always to be very careful, when predicting the future... steenbeck/moviola->nle, 33lp->CD, typewriter->wordprocessor, print->dtp, multitrackaudiotape->daw, telex->web, 8mm->vhs->dv & dvd, crt->lcd, slr->dslr....

 

once digital systems reach a critical price/performance level and a certain quality level which is as good or better than their analogue precessor, they have the nasty habit to dominate down their markets quite fast..

 

the only exceptions to this rule we have seen since the beginnings of the digital era 20/30 years ago were, and will be for a longer time, fine mechanics which then became, speaking in digital terms, a/ds...

such as microphones or loudspeakers, certainly lenses.

 

but film vs. sensor..... as mentioned i would be really careful making predicitions here, i remember quite precise what just recently happened to the still photography market. especially if red (or another player in the biz) really fullfills, the change might be quite faster than even digital snobs expected.

 

Also money is obvioulsy not an object.

here i have to disagree - typically we will find more than 10 businesspeople on 1 artist in the movie, especially in the broadcastindustries. sadly, the huge mayority of movies & tv meanwhile is mostly done only by money as motivation.

 

Frankly, not being compatible with fast 35mm lenses is a HUGE disadvantage for this camera, IMHO.

i agree.

 

Hell, even the GRAIN aesthetic is still really popular on TV.

personally i do agree, but i am surprised that the audiences perception is changing quite quickly.

what is really pondering me, and i am really against these developments, is that broadcaster as the BBC now begin to ban 16mm film - i didn´t see that coming.

(btw there are many petitions online to join who intend to make the bbc overthink the decision to ban 16mm film, i strongly recommend to sign them soon! good starting point is http://www.dff-dk.dk/dff/newssinglepage52.aspx?NID=65 )

 

If anything, by your thinking, they should take The Shield off the air because it deliberately reduced the quality of its imagery by inadequately lighting and not utilizing tripods effectively, and using ZOOMS?! Das ist tottally verbotten.

lol for the content and lol for the german :)

[Andrew Ray]

i would be really careful making predicitions here

Thanks! to all, I really appreciate your comments.

Jan, I will try to be more realistic, though I have trouble with it when I smell big changes in technology.

 

Yes, I am judging the changes in the cinema/TV in a bit egocentric way.

Maybe changes will not be so fast, certainly HDTV didn?t came quick.

I hope though, that progress is more like 2,4,8,16,32 rather then 1,2,3,4,5.

 

David, you touched interesting subject.

Yes, there is a lot of confusion about 1080p 2K and 4k.

I think it is due to the fact, that it is new standard and everybody sees it and does it a bit different way.

Again ?everybody? is just handful of the companies.

 

I just try to think about 4K like a fixed 4096 display elements on the screen.

Some are saying that 4K it is 4000 pixel elements, and yet some say 3,800 pixels, like Sony 4K projector.

I hope that 4K will be 4096 and 2K is 2048 and 1080p is 1080x1920

Again, we are talking about digital displays, LCD screens, projectors.

 

Now we have different aspect ratios of the screens, 4:3, 16:9, 1,85:1, 2.35:1, 2.4:1 etc.

http://en.wikipedia.org/wiki/Aspect_ratio_%28image%29

 

For simplicity we can achieve any of these screen ratios, just by putting proper number of vertical pixels in the frame that has 4096 pixels as a horizontal base.

We are talking screen pixels, not camera pixels and RGB nomenclature, so one pixel has one R one G and one B pixel.

Now when we are talking about the camera sensor, it is different story all together.

 

To cover all the aspect ratios we should have sensor shaped like 4:3, but 4:3 will be out after 2010 by FTC decision, so no need for this one. So let?s say we start with 16:9 ratio.

 

?Poor man? solution sensor will have less then 4096 horizontally and less whatever we need for vertical to meet the ratio of the screen.

Then in post we will uprase it to 4096 X ???? for the final presentation.

 

This one will not meet the film quality requirement but somehow Star Wars was shown in the cinemas based on this formula. Hey, first experiment, let?s be forgiving here.

 

Now if we want to compete with film, then our 4K sensor better have more then 4096 RGB pixels horizontally. In old days we were saying you need 3 times more digital samples for the frequency of one analog interval.

But in the modern days of mathematics conversions and extrapolations and side effect filtering, we are getting closer to 2 times instead 3 times.

 

Now the RGB pixels in 3xCCD prism based system we will count differently then pixels in single CMOS Bayer pattern sensor and yet different in 3XCCD systems with pixel shift by ½ of the width and height of the pixel and yet different from two Bayer pattern sensors shifted by one full length of pixel right and down. And to confuse it even more we will count differently pixels on the single chip Foveon RGB sensor.

 

Dalsa, ARRI D20, and RED all have Bayer pattern camera sensors.

So to call themselves 4K cameras RECORDING cameras they could have even 2048 horizontal pixels and RECORD full 4096 pixels, what?s the problem to record in whatever you want resolution after upresing.

 

But to call the camera full 4K capturing system, then IMHO we better see 8192 horizontal pixels in the Bayer sensor. Some will argue that 6144 will do.

Well, we have all this mathematicians working on it, so I will not argue here.

Sometimes less mean more. Seriously.

 

Karl, I agree with you that 8mm or OMNIMAX can equally impress people as a good/story/acting material, but if you are in the nature/science stories most of the time, you can impress the people much easier, showing tons of details on the screen so your home theater screen looks more like a window to the outside world not like a screen and people go wowww! Though you need wide angle lenses most of the time for this wowww!!

 

Andrew Ray

[David Mullen ASC]

Arri does not call the D20 a 4K camera -- they consider it a 2K/HD camera. It has about 3K horizontal but Arri recognizes that after de-Bayering, it's more like a 2K image, resolution-wise.

 

As for whether the 4K horizontal sensor cameras like the Dalsa and RED are "4K" for all color channels after de-Bayering, that argument has been going on for years, ever since Dalsa came out with the Origin camera.

 

Their argument is that they start out with a 4K Bayer-filtered RAW file and process it into 4K RGB files, and that the final resolution is a factor of the quality of the de-Bayering algorithm, how "smart" it is at reconstructing missing color information, since eaach filtered sensor site is not 100% green, red, or blue, but contains some information of the other colors, which can lead to a fairly good reconstruction. Even most critics of Bayer filters would admit that it's not as bad as only getting 50% of the green resolution and 25% of the red and blue resolution. But to me, it also seems as unlikely that they are perfectly reconstructing 100% of green, red, and blue either. Hence why I tend to feel that 4K Bayer-filtered cameras are really "3K" cameras.

 

Now in the world of filmmaking, there are so many factors that affect resolution that whether there is a visible, practical difference between (effectively) 3K digital images and 4K scans of 35mm film, especially when you've got bigger visual differences between a digital image and a film negative image than just pixel resolution (i.e. film has a grain structure), I think it's within a realm of uncertainty and imprecision that 4K Bayer-filtered sensor images, de-Bayered into RGB, certainly compete with 4K RGB scans of 35mm film negative for sharpness. So at some point, you drift into theory instead of practical reality. All I really care about is what I can see with my eyes looking at the image projected on a big screen.

 

However, I'm all for over-sampling and certainly there would be image improvements from using a Bayer-filtered sensor with more than 4K horizontal resolution (in point of fact, the RED sensor is more like 5K across). However, the downside would probably be a loss of sensitivity from cramming smaller sensor sites into a 35mm-sized area, and these current 4K cameras are already slower in sensitivty at their zero-noise baseline than 500 ASA film stocks -- the Dalsa is around 250 ASA, the Arri D20 is more like 200 ASA, etc.

 

Of course, with such a low-noise image in the Dalsa, I saw some material pushed to an effective 1000 ASA and the noise was minimal.

 

As for needing projector chips with a 4x3 aspect ratio, that would be ideal since anamorphic projection could be accomodated, plus Academy, but the truth is that most movies are widescreen (1.77 to 2.39) and most 2K digital projection at movie theaters, and probably 4K projection, generally opt to use spherical projector optics and just use fewer vertical pixels, rather than deal with anamorphic projector lenses.

[Andrew Ray]

As for needing projector chips with a 4x3 aspect ratio, that would be ideal since anamorphic projection could be accomodated, plus Academy, but the truth is that most movies are widescreen (1.77 to 2.39) and most 2K digital projection at movie theaters, and probably 4K projection, generally opt to use spherical projector optics and just use fewer vertical pixels, rather than deal with anamorphic projector lenses.

 

 

Dave, you are just on.

It is important to separate 4K screen from 4K sensors terminology.

I didn?t know that we can have such nice use of 4:3 frame aspect format.

Just didn?t think about it.

 

Yes D20 is not 4K capturing or even recording camera but I just use it as an example that in post even D20 could be considered as a 4K recording camera.

I am just bit sarcastic here because there are some companies out there marketing this way.

If RED has 5K horizontal resolution then hmmmm?. I hope that the de-Bayer algorithm does high end math to get unique horizontal 4096 pixels out of it.

There was one guy there, I think his name was Nipkov, he said that there is impossible to transmit more data bits per second then the channel bandwidth and there was even the Nipkov law of max speed transmission.

http://en.wikipedia.org/wiki/Paul_Gottlieb_Nipkov

 

He is dead now and guess what, we are transmitting the 56k over plain telephone channel that is only 3K (3000Hz wide) According to his theory and Nipkov law we can transmit only 3K

Now we have Cable TV channels transmitting in old 6MGHz old single NTSC channel 30MBits/sec bit rate of HDTV channel.

 

So I learn one thing in my life, never say never.

 

I guess if you take 2X2 , 4 adjacent pixels of the Bayer pattern and you scan whole sensor left and right and up and down with only one bit offset you can get 5K of unique measurement per horizontal line in the RED 4K sensor.

Then you can repeate the same using 3X2, and 2X3, adjacent pixels and get two more measurements again. Put it in some smart algorithm and maybe you can get 4K out of it.

Also if the camera will pan even so slightly (one pixel width) then you get whole more extrapolated pixels. Now take first frame then second frame, compare/extrapolate, separate again in to two frames and voila, 60fps full 4K, but I want to see it first.

Seeing is believing.

As to the oversampling, you see Dave, for the dynamic range it really doesn?t matter if you have 8MP or 22MP in the S35 frame. We will be combining these pixels back together to form 4096X???? so the light and dynamic range will combine as well.

It is like two TV antennas are giving you 3dB more signal (one stop in terms of light) then single one.

What does matter is the coverage of the sensor surface to total sensor surface.

http://www.dalsa.com/dc/documents/Image_Se...18-00_03-70.pdf

 

For CCD it is I think 90% for CMOS is 50%

CMOS though can go with all these 50% of the other circuitry surface under the surface of the photosensing layer. Kind of multilayer design, once they do it then even CMOS will have 90% surface coverage and much higher speed.

See above article about the CMOS and CCD.

 

Andrew

[Illya Friedman]

Frankly, not being compatible with fast 35mm lenses is a HUGE disadvantage for this camera, IMHO. \\

 

Okay Karl you lost me, what camera (the one not compatible with fast 35mm lenses) are you referring to?

 

I.

 

Illya Friedman

Dalsa

[David Mullen ASC]

Okay Karl you lost me, what camera (the one not compatible with fast 35mm lenses) are you referring to?

 

Yes, I'm kinda lost here too. You can put T/1.3 lenses on all the 35mm-sized single-sensor cameras on the market (Genesis, D20, Dalsa, I assume RED, etc.) For example:

http://www.dalsa.com/dc/equipment_Rentals/....asp?cat=lenses

 

In fact, I believe some of Dave Stump's tests of the Dalsa used an Arri Master Prime.

 

Now I recall some sort of limitation with prism-block 3-CCD cameras that makes T/1.6 as the fastest lens that can be made, not that T/1.6 is slow by anyone's book.

[K Borowski]

Well, I had assumed from all the talk I've seen of using Zeiss medium format lenses that the size of this sensor is bigger than 35mm academy, I assumed the size of Vistavision. So you're saying it's a 4Kx2K in an academy-sized chip?

[David Mullen ASC]

Well, I had assumed from all the talk I've seen of using Zeiss medium format lenses that the size of this sensor is bigger than 35mm academy, I assumed the size of Vistavision. So you're saying it's a 4Kx2K in an academy-sized chip?

 

VistaVision is the same size as the 35mm 8-perf still camera format.

 

The sensor to the Dalsa is just slightly bigger than the 4-perf Super-35 format (which is slightly bigger in turn than Academy), but smaller than the 35mm 8-perf format.

 

So one solution has been to adapt elements from 35mm still camera lenses, some from the medium format world, although that has really only been necessary with the Phantom 65 digital camera, which does have a 65mm-wide sensor, so medium-format lenses are being used for that camera just like with 65mm cine cameras.

 

Also, any focal-length longer than 32mm doesn't vignette on the Dalsa so you can use regular Zeiss cine lenses above that -- and even if you had some vignetting below 32mm, it could be cropped out once you change the 1.96 : 1 full sensor image, whether you crop on the sides to get 1.85 or top and bottom to get 2.39.

 

I just saw some stunning 5-perf 65mm photography by Bill Bennet projected at the Egyptian Theater as a test, to show how good film can look. 50 ASA and some 250 ASA stocks, Zeiss Hasselblad lenses, Vision Premier print... looked amazingly sharp.

[Brian Drysdale]

In practise, are many people shooting with a 2:1 aspect ratio? I know it has been floated, but Vittorio Stararo seems to be the only person that seems to be using it.

[David Mullen ASC]

In practise, are many people shooting with a 2:1 aspect ratio? I know it has been floated, but Vittorio Stararo seems to be the only person that seems to be using it.

 

No, it's not being used much. But then, 4-perf 35mm is 4x3 and that's being used less and less. There has been some advantages from having your capture area be a different dimension than the final aspect ratio in terms of ability to reframe, etc.

[K Borowski]

VistaVision is the same size as the 35mm 8-perf still camera format.

 

The sensor to the Dalsa is just slightly bigger than the 4-perf Super-35 format (which is slightly bigger in turn than Academy), but smaller than the 35mm 8-perf format.

 

So one solution has been to adapt elements from 35mm still camera lenses, some from the medium format world, although that has really only been necessary with the Phantom 65 digital camera, which does have a 65mm-wide sensor, so medium-format lenses are being used for that camera just like with 65mm cine cameras.

 

Also, any focal-length longer than 32mm doesn't vignette on the Dalsa so you can use regular Zeiss cine lenses above that -- and even if you had some vignetting below 32mm, it could be cropped out once you change the 1.96 : 1 full sensor image, whether you crop on the sides to get 1.85 or top and bottom to get 2.39.

 

I just saw some stunning 5-perf 65mm photography by Bill Bennet projected at the Egyptian Theater as a test, to show how good film can look. 50 ASA and some 250 ASA stocks, Zeiss Hasselblad lenses, Vision Premier print... looked amazingly sharp.

 

Yeah, I'm sorry, I misspoke. I basically thought the DALSA sensor was a 24x36mm (8 perf, Vistavision 35mm still etc. . .) sensor. You're basically saying it's 4 perf S35? Well, in that case, it would probably be compatible with Zeiss's line of 35mm lenses. Haha, I'm kind of surprised at you Mr. Mullen, for poking fun at me for suggesting a modern Holllywood film would go through the hastle of shooting and lighting for real technicolor film, when limiting yourseslf to '70s era modified medium format still photography lenses or 65mm lenses of similar age would probalby be just as frustrating the the modern breed of lensers. Some of them probably get puzzled why zooms are even made anymore. I get funny looks for shooting with prime lenses at weddings. In the commercial still photography world primes and glass that's slower than 2.8 is viewed as being as backwards as shooting 35mm over a DSLR.

 

So, in other words, if there are significant limitations in the availability of modern fast zoom lenses to this camera, or if it runs into that distorted digital falloff issue when used with 35mm or Vistavision lenses, it may not ever catch on due to those problems. I'd say it all depends on whether DALSA was smart and designed their camera so that it is compatible with enough modern glass to make not enough of a difference to prompt moviemakers to stay away from it solely for not being compatible enough with modern 35mm glass, which is really the BEST glass made short of custom built scientific lenses.

 

I heard all the talk about Zeiss "70mm" format lenses and assumed that it was 8-perf size or bigger. There shouldn't be any problem with using 35mm still lenses (as long as they're the good ones of course) since they have coverage that is good enough for at least 36x36mm, maybe 30x30, which still leaves plenty or room for, what 18x24 mm sensor with twice the density going across the sensor as the vertical pixel density?

 

If that is the case, then no, that doesn't seem to pose any limitation, except with some of the extreme zooms that you can only get for 35mm flat. I mean, practically speaking you don't have the largest amount of possible lenses at your disposal, which would definitely turn off some directors and cinematographers that want to be able to shoot low-lilght wide open at a moment's notice.

 

That may make the Dalsa unsuitable just as the current 2.35:1 is going out of fashion due to competition from S35 blown up via a 2K intermediate. I assume the lenses it'll have available for non-anamorphic shooting should be able to, at least technnically, rival the performance of flat lenses then.

 

I got the initial impression that this was like the large format digital movie camera that was being tested by someone who posted on here a few mos. ago. Really, that stuff is never going to challenge 35mm because of the constant technical improvements of 35mm glass, whether you're looking at still photography or moviemaking, the 35mm field sees improvements first, and sometimes those improvements make it to MF or LF, and sometimes forget it. Large format/medium format lenses are still hugely limited in their zooming ability. There simply wasn't enough market demand to necessitate them constructing larger-barrelled lenses for the market, so you could be stuck using still zooms which aren't as good, older Vistavision stuff, or shooting with 35mm scope or flat glass and just eating the vignette and cropping it out in post.

 

What I'm basically getting at here is that a super-high resolutino D cine camera with the same lens selection available to it as to a 15-perf. 65mm movie camera, is going to have a hard time rivalling the 35mm market, and the really incredible lenses made for the format these days. Rather it'll be cameras like the Genesis or Sony's successor to the F950, whatever that'll be. taht will probalby have more success than the more "high-end" resolution-wise cameras.

 

I hope that makes more sens.

 

Regards,

 

~KB