This all stems from the fact that CCDs and CMOS are not colour sensitive. Therefore, in order to define the colour coming out of a CCD/CMOS pixel, you have to know before hand what colours it can see.

If you use a 3-chip setup (which can be done for both technologies) you split the light into red, green and blue, and you have a whole CCD for each colour. You use some patterned filter (Bayer, etc.) on a 1-chip design.

The problem is, when the information is stored to disk or tape, it is compressed. The easiest way to compress something is to simply throw away the information... so that's what they do. The whole YUV scheme says that you calculate the luminosity (how bright) each pixel is in the image and you store that in a grid. You then do two additional calculations to produce 2 chroma channels, which can be used to re-construct the colour. Up to this point the procedure can be pretty much lossless.

But then people say that they are more sensitive to brightness than they are to colour (we have both brightness and colour sensors in our eyes. In general, we have more brightness ones)... so they reduce the resolution of the chroma channels by a factor of 2 (4:2:2) or a factor of 4 (4:1:1 and 4:2:0). Specifically they reduce the information in the horizontal dimension. The claim is that we are less sensitive to horizontal things than vertical things, but I expect it has more to do with the horizontal scanning nature of CRTs.

If you ever have the chance to look at some truly uncompressed 4:4:4 video, you'll be amazed at how bad chroma compression really is.

-Steve

color bits are how many different colors there can be in each RGB channel. For example 8 bit color has 256 shades per channel. There are 3 channels of color to equal 24bit. 36bit means 12bits of color per channel. That means each channel can have 4096 shades per channel instead of just 256. This can reduce color banding in color gradients. Spreading 256 shades of blue across a 1920x1080 pixel image can cause color banding because the color gets spread in chunks because there are not anough samples to soread throughout the whole image. With 4096 samples there are more then enough samples per channel to fit the 1920x1080 image.

This has nothing to do with chroma compression however. A 36bit image can still be compressed as 4:2:0. What this means is that each color has a wider range of possible colors but the detail is still reduced.

The only chance any of us ever have of being able to shoot 36bit color is with some form of mpeg4. Mpeg2 cannot use anything other then 8 bit color so no HDV camera will ever have 36bit color.

Isn't the potential limiting factor on the HV20 (vis a vis three chip camcorders or pro-single chip) the size of the chip itself? I'm sure the images are great in good light but isn't noise a real issue esp. in low light on smaller chips? The other thing is the unlimited DOF you get from such a small chip. It's very difficult to throw the background out of focus. Whereas a single full size (35mm) chip gives you a chance to get a very narrow DOF. But single 35mm-size HD-rez chips are still very expensive to manufacture and also require lots of bandwidth in camera to process. So as an aspiring filmmaker, the large DOF on an HV20 would be my biggest hesitation.

Well now that is what makes this a consumer camera. It isn't designed for professionals. It can be used and give very pro results but it isn't meant for that.

We must remember this is a consumer camera. It is what it is and we should not expect it to be a pro camera. If you want a pro camera you should get a pro camera. If you are concerend about DOF even a 1/3" pro camera can be hard to work with. That is why there is all talk of 35mm adapters for cameras. Even a $9,000.00 XLH1 has DOF field issues. A 1/2.7" chip is actually a tiny bit larger then a 1/3" pro chip so at least the HV20 is starting to move in a better direction.

Sorry if I was stating the obvious :-( but it seemed like low light noise and DOF were bigger issues than how a single chip handles colorspace.

It all depends on the person. While lowlight noise may be a concern for some people, it is something that can be overcome by using lights. Color space however is something we have no control over. DOF is also something that not everybody is going to be concerned about. A consumer will want the opposite effects of DOF then what a filmmaker would want. For DOF it is another one of those things that is subjective and can be worked on by the user if they so choose. A 35mm adapter can be used if you must have a certain style of DOF. Color space and overall quality from the camera, chip and DSP is something that none of us have control over so it is nice to find a camera that works for us.

Besides the HV20 should be better at DOF then any other 1/3" camera since it has a 1/2.7" chip. This is the first camera other then 1/2" or 2/3" cameras to have a chip larger then 1/3". I know it isn't that much larger but it is a step in the right direction. You will get the same type of results for DOF as you would with any highend 1/3" camera and maybe even slightly better. As far as I know this is the only HD camera under roughly $25,000.00 that has a chip larger then 1/3".

Just to clarify -- it is the first HD camera other then 1/2" or 2/3" cameras to have a chip larger then 1/3". It's been done before in standard definition (re: the last models from the Canon Optura series DV camcorders).

wow thanks for the easy to understand answers!!! as usual, this place rocks. thanks thomas&steven.

ya'll really are "experts" when it comes to this stuff.

In milimetres (diagonal), how large is the surface area of 1/3 and 1/2.7"? How much larger is the HV20's sensor.

In all actuallity it is easier than it looks, although I am not sure who decied to do a decimal fraction to represent size, but a 1/3ccd is .33 inches (although most ccds can be misrepresented by size) and the 1/2.7 Cmos sensor is .37 inches.

A pet peeve of mine... don't get me started.

I wrote a little bit about it here: http://www.dvinfo.net/canonoptura/ar...eage.php#opccd

The interesting thing about the HV10 is that it actually has LESS noise in reasonable to good light than many prosumer cams costing much more. The pristinity of its image is one of the great things about the HV10.

im going to be making a lot of short films,some event videography,will the hv20 be good enough or should i step up to the a1.thank you

I'm coming from a DVX100 (original) and am chomping at the bit to replace it. I do my creative stuff (short film projects, etc.) with it but also a lot of family video and some weddings. That latter two tend to be low light stuff (I usually light my film projects). The XH A1 is my first choice for a new video camera but I'm also going to replace my Canon Rebel XT DSLR this year with a higher end (hopefully full frame) camera. So to buy both a Canon 5D (or it's replacement) and the A1 means a bit of a hit financially ($6-7K). Now all of a sudden the HV20 looks like an appealing (i.e., low cost) HDV solution. I can't wait to see the image it produces.

Keep in mind that the HV10 is still not a stellar performer in low-light. HDV cams in general need plenty of light. With that said, cams such as the A1 or Sony FX7/V1/FX1 are still better in low-light than the HV10. But give the HV10 decent light, and you'll be hard pressed to tell it apart from far more expensive HDV cams.