Jump to content

ccd size


Ravi

Recommended Posts

Hi guys, I am shooting my next feature on LKD7500 viper which is a wonderful camera.But I am wondering why would engineers at thomson gv would opt for 2/3' ccd sensor and not a super 35 size or bigger one.Afterall it is a state of the art ,professional camera, it will be a marginal addition to overall price.Your thought on this? Thanks in advance Ravi

Link to comment
Share on other sites

Hi guys, I am shooting my next feature on LKD7500 viper which is a wonderful camera.But I am wondering why would engineers at thomson gv would opt for 2/3' ccd sensor and not a super 35 size or bigger one.Afterall it is a state of the art ,professional camera, it will be a marginal addition to overall price.Your thought on this? Thanks in advance Ravi

Sorry I forgot to mentioned my complete name its ravi sudashan,my apologies..

Link to comment
Share on other sites

You should change to your full name using the my controls.

 

The viper is the first cameras intended to record the data from the sensors. At that time none of the HD cameras were full 35mm sensor and since Thompson have a background in high end broadcast cameras it would make sense for them to use proven technology in an unproven market on what could be a limited production run.

Link to comment
Share on other sites

  • Premium Member

It's easier to get 1920 x 1080 pixels per color channel with a 3-CCD set-up than a single 35mm sensor with a color filter array like a Bayer pattern, which then requires processing into RGB at some point, even for monitoring.

 

Plus Thomson's LDK camera is primarily made for TV broadcast work and not cinema work, so it makes sense for them to stick to a 2/3" 3-CCD design. The extra depth of field of a 2/3" sensor makes more sense for a camera designed to shoot rock concerts, sporting events, news shows, etc.

Link to comment
Share on other sites

  • 3 weeks later...
It's easier to get 1920 x 1080 pixels per color channel with a 3-CCD set-up than a single 35mm sensor with a color filter array like a Bayer pattern, which then requires processing into RGB at some point, even for monitoring.

 

Plus Thomson's LDK camera is primarily made for TV broadcast work and not cinema work, so it makes sense for them to stick to a 2/3" 3-CCD design. The extra depth of field of a 2/3" sensor makes more sense for a camera designed to shoot rock concerts, sporting events, news shows, etc.

 

Its the only 3 x 9Mpixels camera that Thomson builds, its the only 2,35 :1 native HD camera... it has 27.5Mpixels CCD to deliver a 1080p image, thats x4 oversampling... its 2.5 times the pixel count of RED... it has a real mechanical shutter...

 

So no David, I disagree, its not designed for TV... its a real Digital Cinema camera...

 

In CinemaScope mode with standard lenses, it delivers the same resolution as RED does... with the difference that the same pixels are derived from 2x source pixels as RED has...

 

I don't thing a regular broadcast camera gets 13 Oscar nominations...

 

F23 its not even close...

 

Of coarse an S35mm ccd on a new Viper would not heart us...

Link to comment
Share on other sites

  • Premium Member

Even Thomson files it under "Broadcast Products"

http://www.thomsongrassvalley.com/products_all/

 

Sure, the Viper is a good digital cinema camera. But Thomson as a company is primarily interested in making broadcast cameras, which is why I doubt you'll see a 35mm sensor Viper in the near future unless someone there changes their mind. I tried to get the head of Thomson once to talk to me about the future of the Viper and he seemed totally uninterested in the digital cinema market. The Viper, and FilmStream, was an experiment for them.

 

But my main point is to answer the question why the Viper has 2/3" sensors, which is the same reason the F23 has 2/3" sensors -- these are broadcast video companies who primarily sell to broadcast video people and rental houses who have invested millions of dollars over the years on 2/3" B4 optics. The fact that these 2/3" cameras also make decent digital cinema pictures is a nice secondary market for these companies, though I don't think the Viper was successful enough for Thomson to get them excited or committed to the idea of making more cameras for that market. Where is the Viper 2? Is the Infinity the next generation of Viper?

 

And believe me, when I say something is a broadcast video camera, that's not meant to be an insult. Those things can be state of the art and produce super high-quality HD images, good enough for cinema applications.

Link to comment
Share on other sites

Even Thomson files it under "Broadcast Products"

http://www.thomsongrassvalley.com/products_all/

But my main point is to answer the question why the Viper has 2/3" sensors, which is the same reason the F23 has 2/3" sensors -- these are broadcast video companies who primarily sell to broadcast video people and rental houses who have invested millions of dollars over the years on 2/3" B4 optics. The fact that these 2/3" cameras also make decent digital cinema pictures is a nice secondary market for these companies, though I don't think the Viper was successful enough for Thomson to get them excited or committed to the idea of making more cameras for that market. Where is the Viper 2? Is the Infinity the next generation of Viper?

 

And believe me, when I say something is a broadcast video camera, that's not meant to be an insult. Those things can be state of the art and produce super high-quality HD images, good enough for cinema applications.

 

I don't thing that the design behind this sensor has to do with the 2/3 manufacturing easiness... because is the only 3 x 9Mpixels CCD sensor that Thomson is using, and its being designed from scratch to do scope...

 

I thing David, a better reasoning is the 2/3 "culture" that they had... R&D to a bigger CCD wafer is much easier than a smaller... the breakthrough in microelectronics is to make things smaller... so toooo many broadcast engineers around that project... and they have Fu%$#$@ it...

 

The performance of the camera today has nothing to fear from F23... and a lot to fear from F35... RED is another story...

 

DoF is the number one difference... WHICH IS BIG, and in favor of RED...

Dynamic range which is in favor of Viper by at least 2,5 stops...

Motion artifacts like skew and global shutter are non exist in Viper since it is using a true mechanical shutter.

The uncompressed nature of Viper gives much better results in all aspects of image, but its harder to work with than RED...

 

Resolution is another subject... lets compare... in case of CinemaScope feture, like "Benjamin Button"...

 

With RED we have at 4K in the center of an ISO card 1450 lines, with Viper in scope mode, we have 1300 lines.

 

When we film out we using, 2K film recording almost 99%...

 

So with RED we get from a 4096 x 1742 bayer, thats 7.1 Mpixels source, we get 2048 x 871 image.

 

With Viper we get from a 1920 x 3240 RGB sensor, thats 3 x (1920 x 3240)= 18,66 Mpixels source, a 1920 x 1080 anamorphic scope image...

 

RED = 2048 x 871 = 1.78Mpixels in the theatrical CinemaScope projection

F23 = 1920 x 817 = 1.56Mpixels in the theatrical CinemaScope projection

F35 = the same as F23...

 

Viper = 1920 x 1080 anamorphic pixels = 2.07Mpixels in the theatrical CinemaScope projection

 

So with Viper against RED, F23, F35, we get a better resolution by at least 25% when projecting in CinemaScope...

 

Viper has a bigger resolution than all the others and it gets that because of the 27Mpixels sensor...

 

If the CinemaScope its not Cinema and DoF is what matters, then its not a Digital Cinema camera...

 

But to judge a camera like Viper with how its being list in a web site... for me, the marketing guys in Thomson, they have mess it up big time...

 

For me Viper is way better than all Sony cameras apart from F35...

 

As i see it is

 

EPIC FF35, D21, F35, Viper and all the rest

Link to comment
Share on other sites

  • Premium Member

Ease of manufacture doesn't clearly favor either design. They're both very difficult to make. With a single chip camera, your chip fab has to be able to align and apply the Bayer or vertical stripe color filter material. With a three chip camera, you have to make a complex and precise glass prism and dichroic filter assembly, and you have to stick the three chips onto it so they line up correctly. That ain't easy. With a single chip camera, you can use pretty much any lens design you want. There's nothing in the way until you get to the OLPF, just a few millimeters from the focal plane. With a three chip camera, you can't make a lens faster than f/1.45, because the glass prism block is in the way. You can, though, tweak chromatic abberation in the block design. The data from the three chips comes out in proper pixel form, three numbers, RGB, for each place. With a single chip, it takes some computing to deal with the fact that you have only one color per physical location. That both designs have plenty of pluses and minuses is borne out by the fact that both are out there in the marketplace and doing just fine. There's no clear winner.

 

Little chips are easier to make than big ones because of the yield problem. Any tiny dust particle ruins whichever chip it hits. If you get two dust particles on a wafer containing 10 chips, you get 8 good ones, or maybe if you're lucky 9. If there are 100 chips on the wafer and the exact same dust, worst case 98 will be OK. (Those numbers are just to explain the principle, they're not the actual numbers for any particular chip.) As chip makers get better at handling the dust problem, bigger chips become economically feasible.

 

Viper doesn't gain any resolution at all over 1920 x 1080 from its 1920 x 4320 photosite design. The physical photosites are four times wider than they are high, and they're not individually readable. What the Viper does is it reads them out adding up groups of four to make 1920 x 1080 1.78:1 square pixel HDTV, or it reads them out adding up groups of three vertically to get an anamorphic non-square-pixel image. In both cases it outputs a 1920 x 1080 data set. It's strictly a clever way of doing scope on the chips, no more, no less.

 

Whether you have rolling shutter jell-o-vision issues depends entirely on how fast you can read out your chips. Viper's mechanical shutter is there to save them the covered storage array that Sony uses on their CCD's, cutting the size of the chips in half. With the shutter closed, Viper can pull the data off the edge without streaking bright points such as practical lights in frame.

 

 

 

 

 

-- J.S.

Link to comment
Share on other sites

Viper doesn't gain any resolution at all over 1920 x 1080 from its 1920 x 4320 photosite design. The physical photosites are four times wider than they are high, and they're not individually readable. What the Viper does is it reads them out adding up groups of four to make 1920 x 1080 1.78:1 square pixel HDTV, or it reads them out adding up groups of three vertically to get an anamorphic non-square-pixel image. In both cases it outputs a 1920 x 1080 data set. It's strictly a clever way of doing scope on the chips, no more, no less.

-- J.S.

John,

 

Even if its on-chip averaging, that doesn't mean that light is not falling in physically different photosites, that have different noise patterns and light levels, in the same point in time... that is exactly the same as if it was, on a later step inside a DSP done an average processing step... the result will always be, lower noise floor and quieter resolution with no ringing effects... the same thing like getting 4K RGB frame and downres to HD res... the only problem is that we don't have access to the original 4K image...

 

And yes you get more resolution, when shoot in 1:2.35 scope an ISO res chart we get 1300 lines... instead of 980... and that's because in scope the camera records in 1440 vertical resolution... but records only the center portion of it that is 1080 in anamorphic scope...

 

So its not only clever, it is also very effective idea...

Link to comment
Share on other sites

  • Premium Member

Actually viper is pretty noisy.

 

Don't get hung up on the mechanical shutter, either. It isn't anything to do with skew, it's there to allow the offload mechanics of the sensor to function properly. F23 has no mechanical shutter (and is enormously resistant to blooming)

 

P

Link to comment
Share on other sites

  • Premium Member
John,

 

Even if its on-chip averaging, that doesn't mean that light is not falling in physically different photosites, that have different noise patterns and light levels, in the same point in time... that is exactly the same as if it was, on a later step inside a DSP done an average processing step... the result will always be, lower noise floor and quieter resolution with no ringing effects... the same thing like getting 4K RGB frame and downres to HD res... the only problem is that we don't have access to the original 4K image...

 

And yes you get more resolution, when shoot in 1:2.35 scope an ISO res chart we get 1300 lines... instead of 980... and that's because in scope the camera records in 1440 vertical resolution... but records only the center portion of it that is 1080 in anamorphic scope...

 

So its not only clever, it is also very effective idea...

 

Downconversion from 4K to HD isn't done by simple averaging. That wouldn't gain you any advantage in top octave resolution by making the Nyquist limit with a steeper slope low pass filter curve. It's done with multi-tap filters that can make that steeper cut.

 

I discussed this at length with Mark Chiolis when the Viper was new. It doesn't do multi-tap filtration on the data from its vertical sub-sites. It just adds up three or four of them, just a simple sum. It adds up the signal, it adds up the noise. As far as resolution is concerned, it's exactly the same as if the chip had bigger photosites, equal in area to 3 or 4 of the actual sub-sites.

 

It would be entirely pointless to do fancy filtration in the vertical direction only. The horizontal sampling is the same 1920 across for the Viper and the Sony cameras. Human perception of images with a mis-match between vertical and horizontal resolution is that we see them as having the lower of the two resolutions. (Unless the difference is too big, over 2:1. Then we see the mis-match as such.)

 

I don't doubt that you're seeing higher resolution from the Viper when you shoot a chart. It's just that the four by one photosite design isn't the reason for it. The real reason is something else, likely a less aggressive OLPF.

 

 

 

 

 

 

-- J.S.

Link to comment
Share on other sites

I don't doubt that you're seeing higher resolution from the Viper when you shoot a chart. It's just that the four by one photosite design isn't the reason for it. The real reason is something else, likely a less aggressive OLPF.

 

-- J.S.

 

Averege is going half way to what a sophisicated algorithm will do...

 

So the point is, that five years after Thomson introduce Viper, Sony build an F23 that is a lot less of a cinema camera than Viper is...

 

In all aspects... DR, resolution, scope capable etc...

 

So F23 can be a good B-camera on a Viper project...

Link to comment
Share on other sites

  • Premium Member
Averege is going half way to what a sophisicated algorithm will do...

 

Well, OK as a subjective opinion.... Not sure how rigorously we could quantify that.

 

So the point is, that five years after Thomson introduce Viper, Sony build an F23 that is a lot less of a cinema camera than Viper is...

 

Is it really only five years? I remember Mark showing us stuff a lot longer ago than that. I was surprised back then that they had sunk so much into a camera which at the time had absolutely no way to record its output. But they won that bet.

 

 

 

-- J.S.

Link to comment
Share on other sites

Well, OK as a subjective opinion.... Not sure how rigorously we could quantify that.

 

Is it really only five years? I remember Mark showing us stuff a lot longer ago than that. I was surprised back then that they had sunk so much into a camera which at the time had absolutely no way to record its output. But they won that bet.

 

-- J.S.

 

The question John is 1) why someone would buy an F23 and not a Viper? and 2) why Vipers are vanished? Only Fincher can understand the difference?

Link to comment
Share on other sites

  • Premium Member
The question John is 1) why someone would buy an F23 and not a Viper? and 2) why Vipers are vanished? Only Fincher can understand the difference?

 

I've heard that Fincher won't be using the Viper again -- he'd rather use the F23 or something else. The Viper had some sort of fan noise issues that Thomson wouldn't fix, so he had to use the F23 for the hospital scenes in "Benjamin Button" just to be able to record dialogue, and since then, I think he did a commercial on the RED. Basically the problem is that Thomson is not really supporting the Viper anymore. Not to mention that a lot of people find it a bit noisy compared to the F23.

Link to comment
Share on other sites

  • Premium Member
The question John is 1) why someone would buy an F23 and not a Viper? and 2) why Vipers are vanished? Only Fincher can understand the difference?

 

Hi Evangelos,

 

Band Pro was doing a very good part exchange deal for Vipers traded in for F23's, apparently they collected more than half the Vipers that were ever made, these will probably be destroyed so they can get the tax deduction.

 

Stephen

Link to comment
Share on other sites

  • Premium Member
.... apparently they collected more than half the Vipers that were ever made, these will probably be destroyed so they can get the tax deduction.

 

I've never heard of that -- what government gives you a tax deduction for destroying cameras?

 

 

 

 

 

-- J.S.

Link to comment
Share on other sites

  • Premium Member

Seriously, though, Sony isn't a government. It isn't owned by the government as would be the case under communism, nor is it controlled by the government as it would be under fascism. If there really is a govenment somewhere giving tax incentives for the destruction of Vipers or any cameras, that would be interesting news.

 

 

 

 

-- J.S.

Link to comment
Share on other sites

  • Premium Member

At least here in the States, you have to own something for a while, like 3 - 5 years or more, before you can depreciate it for federal taxes. If they didn't do that, businesses could take truly worthless items as trade-ins, and claim too much for them.

 

 

 

 

-- J.S.

Link to comment
Share on other sites

  • Premium Member
I've never heard of that -- what government gives you a tax deduction for destroying cameras?

 

-- J.S.

 

Hi,

 

When stock clearly has no value this must be reflected in the accounts. Stock is usually held at the lower of cost or net re-saleable value.

 

Stephen

Link to comment
Share on other sites

Hi I have shot for 20 days with viper now,I love easy to use approach of the system.What I did not like is its electronic viewfinder and lack of balance when one is doing handheld work.Another problem that we are facing is time code breaks.The workflow that we are following is record on venum flash then transfer the data on HD tape at HQ mode.I have shot alot in low light,hope my footage is not going to be grainy....Thanks Ravi sudarshan

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...