Native 1080p (DVCPRO HD)?

[Rocketeer]

Would it be fair to speculate that based on the HVX200, a 2/3'' P2 Varicam that does 1080p can't be to far away? maybe with a 4 card array?

 

[Barry_Green]

Well, before encoding the image has to pass the DSP, so it can't be overkill as by definition.

Yes it is, because the extra pixels are costing you surface area that could have been better employed for sensitivity and latitude...

Better to have a good head-section and a mediocre format, then a mediocre head-section and an exellent recording format...

Absolutely agreed.

 

[Barry_Green]

CMOS is coming (or supposed to be)... JVC's GY-HD7000U is supposed to use three 2/3" CMOS sensors, and supposed to be on the market around June of 2006.

Regarding the HC1 -- what's the deal with the "rolling shutter"... I haven't kept up on that, but apparently you can get some weird banding or stretching artifacts depending on the type of shot... is that going to affect all CMOS sensors? Or is that specific to the HC1 design?

 

[Spiff]

I saw a 3 CMOS Sony the other day... anyone know anything about it?

It's probably their mid-range consumer DV cam. So far as I can recall from the CCI.com review, the CMOS sensor was the only thing about it that was particularly interesting... other than that it behaved much like the other camcorders from Sony in that price bracket.

I was under the impression that 1 CMOS was enough... can anyone explain?

Regular CMOS chips are intensity sensitive only, and require a primary colour Bayer filter or a prism just like a CCD system in order to get colour information... So a 3 CMOS has the same advantages over 1 CMOS as 3 CCD does over 1 CCD. Expect to see 3 CMOS on the more expensive camcorders as the technology gets more common.

EDIT II:
Apparantly the Foveon chips are also a version of CMOS. When I read about these in Scientific American a few years ago, they were talking about how this design is senstive to the depth deposition of energy - and since blue photons are more energetic than red photons, the Foveon chips are actually are single-chip but colour sensitive at the same time. The primary advantage of this method is that you don't have the optical issues associated with aligning CCDs or prisms... but the read-out times of the chips were (then) too slow to get good video. I guess they're working their way to that.

You can be assured however, that if a camera does have a Foveon chip (or a Foveon-type-chip) that the marketing department will have a firm hold of that fact, and be pumping the camera release with a suitable amount of "GROUNDBREAKING REVOLUTIONARY SUPER-AWESOMENESS" hype, some kind of fancy trademarked name, and everything else that goes with nifty new tech.


EDIT:
Hey Barry, here's a nifty thread on the rolling shutter thing:
http://www.sonyhdvinfo.com/showthread.php?t=2688
From what I can tell, it's an implementation-of-CMOS issue - I see no technical reason why it couldn't be avoided with appropriate circuitry.


-Spiff

 

[mmm]

It's probably their mid-range consumer DV cam. So far as I can recall from the CCI.com review, the CMOS sensor was the only thing about it that was particularly interesting... other than that it behaved much like the other camcorders from Sony in that price bracket.



CMOS requires a primary colour Bayer filter or a prism just like a CCD system in order to get colour information... So a 3 CMOS has the same advantages over 1 CMOS as 3 CCD does over 1 CCD. Expect to see 3 CMOS on the more expensive camcorders as the technology gets more common.

-Spiff

Thank you Spiff...2

Another question - why do the D20 and Genesis only use a single CMOS sensor?

 

[Spiff]

why do the D20 and Genesis only use a single CMOS sensor

Interpreted from this posting:
http://www.cinematography.com/forum2004/lofiversion/index.php?t6178.html

It looks like the D20 and Genesis sensors are just really high resolution and physically large CMOS sensors that use Bayer-like patterns/methods. If the resolution of the sensor is large enough, you don't need to worry about the Bayer interpolation and can just down-sample to 4:4:4 colour. The sensor size will give them good low-light performance and everything else we associate with large CCD cameras.

-Spiff

 

[Barry_Green]

Would it be fair to speculate that based on the HVX200, a 2/3'' P2 Varicam that does 1080p can't be to far away? maybe with a 4 card array?

I'm pretty sure exactly such a product is in the works. At NAB 2005, during the "HD in 3D" presentation, they teased a forthcoming shoulder-mount 2/3" 3-CCD P2 camera. They didn't call it a VariCam at the time, I don't think, but the handwriting is all over that wall!

 

[harlan]

Would it be fair to speculate that based on the HVX200, a 2/3'' P2 Varicam that does 1080p can't be to far away? maybe with a 4 card array?

There's one in development which will offer far more than just the inclusion of 1080.

 

[mmm]

I'm pretty sure exactly such a product is in the works. At NAB 2005, during the "HD in 3D" presentation, they teased a forthcoming shoulder-mount 2/3" 3-CCD P2 camera. They didn't call it a VariCam at the time, I don't think, but the handwriting is all over that wall!

Isn't it time to update DVCPROHD too... it is a little, well, old fashioned.

 

[evinsky]

I thought that the D20 and for sure the Genisis are RGB sensors and not Bayer. IE they have dedicated color photosites that require no de-mosaic process. This is essentially the same as three chip it just means the sensor has so much densitiy it can dedicate three photosites to every recorded pixel.

 

[mezelf27]

Would it be fair to speculate that based on the HVX200, a 2/3'' P2 Varicam that does 1080p can't be to far away? maybe with a 4 card array?

I thought I saw a concept drawing of such a cam already (from pana, obviously)

 

[mezelf27]

Yes it is, because the extra pixels are costing you surface area that could have been better employed for sensitivity and latitude...

That is true, but I'm guessing a personal preference. I know from experience (writing the image compression software) it is a CHOICE, rather than a calculation, as it is impossible to calculate witch sacrafice will end up to be visible and witch ones won't. The perfect balance should be as much pixels that it is virtually impossible to detect they're not there and the maximum of latitude that comes with using the minimum of pixels.
Problem is it isn't possible to exactly calculate that 'minimum of pixels' as it is correlated with the imager passing through DSP.
However; should the pixelcount only be 960x720 (I'm not saying it is, but should it be so...), well then: rest assured, that's too little pixels to get a decent 1920x1080 frame (you can only do so much... in the case of pixel shift: give the green a mathematical weight of 2 (compared to 1 for the other 2 colors - as it is approximately in the human eye as far as sensitivity is concerned)... what do you get? (960 x 720) x 4/3 = 921 600 as artificial average pixels per CCD, where you're trying to create an image that needs 2 073 600 pixels.... - by the way 921 600 is exacly the native resolution of 1280x720)

 

[mezelf27]

Isn't it time to update DVCPROHD too... it is a little, well, old fashioned.

One should say that, indeed, but that usually leads to a lot of compatibility issues that isn't quite good for a product for the broadcast-market (where stability is a bigger issue). the 960 vs. 1280 e.g. would otherwise a thing to consider...