hvx200 pixel count revealed

[Earl Thurston]

Originally Posted by David Heath

Having read over all the above, I feel a lot depends on definitions.

I think you've hit the nail on the head. The ratios we all see for 4:2:2, 4:1:1, 4:4:4 etc. are specifically definitions for the digital side of things. They're not really meant to define the analog aspects of thing like the CCDs, though people have been using them for that purpose.

Thus, when Jan defends the HVX as being fully 4:2:2, she is absolutely 100% correct in that regard, but specifically as it defines the digital side of things from the DSP onward. What happens on the analog side, from the CCDs to the DSP, however, is a different issue and the digital definitions aren't really meant for this.

[Barry Green]

Originally Posted by Brian Luce

so now we're saying not even dvcpro50 is 4:2:2? wow, the news gets worse all the time. i thought dvcpro50 was already confirmed at 4:2:2.

This is preposterous. DVCPRO50 delivers full 4:2:2 performance, and all you have to do is look at the footage. Have you looked at the 4:1:1 vs. 4:2:2 footage I posted?

All the talk of specs may be interesting for the technogeeks among us, but don't forget that the images are what count. When you put the package together and turn it into actual video, the HVX performs spectacularly.

[Stephen L. Noe]

Originally Posted by Earl Thurston

Thus, when Jan defends the HVX as being fully 4:2:2, she is absolutely 100% correct in that regard, but specifically as it defines the digital side of things from the DSP onward. What happens on the analog side, from the CCDs to the DSP, however, is a different issue and the digital definitions aren't really meant for this.

Not exactly, the CCD's pixel count is what defines what is fed to the DSP. With DVCProHD 720p where it pertains to the Varicam we already know how they've done the math. This camera is a different animal. At HD resolution something is missing from the CCD's and there must be some interpolation method that is making up the extra row. That extra information is being displayed when the camera is stationary but when the camera is in motion the information is not there and the default CCD rez is revealed.

There is no question that the format itself is 422 (DVCProHD), the question (to me) is how are they getting information that does not exist at 1080p? I'm going to read the supplied information over again and try to find if I'm missing something.

[Steve Mullen]

Originally Posted by Stephen L. Noe

There is no question that the format itself is 422 (DVCProHD), the question (to me) is how are they getting information that does not exist at 1080p?

You are correct. A format has V & H specification. For example, NTSC defines 480-lines. But, the actual "effective" resolution is not 480 TVL. It's much lower.

So, of course, the DVCPRO50 and DVCPRO HD "format" is defined to be 4:2:2. But, it seems very clear that there is less chroma vertical resolution from the HVX200 CCDs than from a Varicam CCDs.

So we know the luma resolution varies. We know there is no way the camera records resolution equal to other 1080 camcorders. And, now we know the effective vertical chroma resolution is low for 720p and very low for 1080.

Does this really matter? Barry says no.

For the most part I agree completely.

Except where our understanding allows us to understand things folks are seeing when they critically look at the video.

And, except when Panasonic and reviewers neglect to explain bothe postive and negatives of their technology. But, there is nothing new here. Sony leads you to believe -- and reviewers continue to claim -- that PS enables 960 pixels to become 1440.

What we are doing with the HVX200 is exactly what we did with the FX1/Z1 and the HD100 -- subjecting marketing claims to a very, very close examination.

[Earl Thurston]

Originally Posted by Stephen L. Noe

Not exactly, the CCD's pixel count is what defines what is fed to the DSP...At HD resolution something is missing from the CCD's and there must be some interpolation method that is making up the extra row.

Actually I'm in agreement. Although the digital side is 4:2:2, the CCDs are not producing the full resolution that could be carried within that data. I'm just saying that "4:2:2" is a definition that doesn't apply to the analog side, thus people can't say Panasonic lied and the camera isn't really 4:2:2, because technically it is.

[Bob Grant]

Let's look at this another way around and take an extreme example.
DigiBeta records 720x576 (in PAL) at 4:2:2. One can upscale that and record it to HDCAM SRW at 1920x1080 at 4:2:2. No one surely is going to try to convince us that what we end up with is the same color resolving ability in say the blue channel as native HDCAM SRW?

Thing is yes, DVCProHD is 4:2:2, the point is though that it's lower resolution than HDV so to say that the difference between the two formats in terms of color resolution is as simple as the difference between 4:2:0 and 4:2:2 is invalid.

I agree that all these numbers don't mean a lot, what matters is how the image looks on the BIG screen relative to the original image in front of the lens in the first place. The aim of any imaging system be it film, still or video should be to do as accurate a job of that as possible, for creative reasons the image can then be bent in whatever direction one desires, starting with a pre bent, fudged image simply because in its native form it is more pleasing is limiting creativity.

Certainly the eye resolves differences in color poorly but many things done to images in post today rely on differences in color, if that resolution is not there then the resulting artifacts are very noticable to the eye.

So why not a simple test, use the standard resolution charts but instead of black and white use red and white. Then see how the various cameras stack up.

[David Heath]

Originally Posted by Barry Green

All the talk of specs may be interesting for the technogeeks among us, but don't forget that the images are what count. When you put the package together and turn it into actual video, the HVX performs spectacularly.

None of that can seriously be disagreed with, it's all common sense. Though 'performs spectacularly' may need qualification - compared to what? In reality, as far as I can tell, ALL of the small form HD cameras perform spectacularly, and it's comparisons between them that are interesting.

What does trouble me about all of this is that before the HVX200 was released I read numerous times of it's specs, all pointing to how much better than the opposition they would make it. (Though, of course, some specs were never released in advance......) Equally, all sorts of numbers were paraded to rubbish rival cameras and formats.

What's happening now is real world tests are starting to be done, and more numbers are coming into the public domain. Whilst none of them rubbish the HVX, they do call into question some of the wild assertions made about how much *better* than all the others it would be.

It's fair enough to say specs are just for the tecnogeeks. It's fair enough to pore over them if you're interested - and aren't these forums the place to do just that? What nobody should do is quote and make a big deal over the specs that suit them, but downplay the ones that don't. Or even worse, exagerate the bad issues surrounding rival cameras, whilst not acknowledging their good points. And as Steve so rightly says, whilst Panasonic may be the ones under the microscope here, it's not always that way.

[Bob Grant]

I'll go further than that, as a geek I'll say that numbers matter an awefull lot but they can be wrong, meaningless or not telling us the whole story. One only has to look at the history of the quest for better audio reproduction.

The engineers back in the 1950s kept building systems with better numbers but in double blind tests the audience kept saying they sounded worse. After nearly a decade of very deep investigation it was realised that they were measuring the wrong parameters, sadly most audio gear is still speced using numbers that don't tell us a whole lot, hence we still have people using terms like 'warm' and 'spacious'.

Now Panasonic could well be onto something, if double blind tests show that the viewing public prefers the image of camera A over camera B yet the numbers tell us camera B is better then we need some serious investigation, perhaps our metrics are wrong or not telling us the whole story.

[John Jay]

moving from the 50's to the 60's, I remember buying LP records clearly marked as 'Electronically Reprocessed for Stereo' - in truth they just messed with the frequency and phasing of a Mono record.

It seems we are now in the era of 'Electronically Reprocessed for HD' ...

[Stephen L. Noe]

Originally Posted by Steve Mullen

Does this really matter? Barry says no.

For the most part I agree completely.

It defines where the camera is best used and what scenario's to avoid. With the HVX there is the possibility of noise and low rez. With the Z1 there is a possibility of marcroblock, with the HD-100 there is the possibility of split screen. We know why the Z1 macroblocks and we know why the HD-100 split screens, it's been defined. Now why does the HVX have low rez (sometimes) and excessive noise (in gain)? I think figuring these things out is defining it's boundaries to stay within to get the best overall picture the camera can offer. From my experiments, putting the HVX in motion does result in a soft picture.

[Thomas Smet]

Originally Posted by Bob Grant

Let's look at this another way around and take an extreme example.
DigiBeta records 720x576 (in PAL) at 4:2:2. One can upscale that and record it to HDCAM SRW at 1920x1080 at 4:2:2. No one surely is going to try to convince us that what we end up with is the same color resolving ability in say the blue channel as native HDCAM SRW?

Thing is yes, DVCProHD is 4:2:2, the point is though that it's lower resolution than HDV so to say that the difference between the two formats in terms of color resolution is as simple as the difference between 4:2:0 and 4:2:2 is invalid.

I agree that all these numbers don't mean a lot, what matters is how the image looks on the BIG screen relative to the original image in front of the lens in the first place. The aim of any imaging system be it film, still or video should be to do as accurate a job of that as possible, for creative reasons the image can then be bent in whatever direction one desires, starting with a pre bent, fudged image simply because in its native form it is more pleasing is limiting creativity.

Certainly the eye resolves differences in color poorly but many things done to images in post today rely on differences in color, if that resolution is not there then the resulting artifacts are very noticable to the eye.

So why not a simple test, use the standard resolution charts but instead of black and white use red and white. Then see how the various cameras stack up.

The difference is that when you scale up that SD image even though the pixels are interpolated the vertical chroma and luma will be equal. With pixel shift there is no way ever even with an analog signal that the luma pixels can be the same as the chroma pixels in the vertical direction. This is why I actually think it would have been better to just do an upscale of the 960x540 instead of using pixel shift. Yes the luma would have been a little bit softer than it is now but at least the chroma and luma would be a perfect match. How much softer would it have been anyways?

[Thomas Smet]

Originally Posted by Barry Green

This is preposterous. DVCPRO50 delivers full 4:2:2 performance, and all you have to do is look at the footage. Have you looked at the 4:1:1 vs. 4:2:2 footage I posted?

All the talk of specs may be interesting for the technogeeks among us, but don't forget that the images are what count. When you put the package together and turn it into actual video, the HVX performs spectacularly.

If the numbers do not matter then why even use pixel shift. Since the HVX200 is already soft just forget about the slight gain in detail from pixel shift and just upsample the raw 960x540. A raw 960x540 will upscale very well but not have any issues of detail issues per channel because of the pixel shift. Yes in nature each pixel would be larger and softer than other cameras but it would have been a very clean high quality larger pixel.

[Barry Green]

Originally Posted by Thomas Smet

If the numbers do not matter then why even use pixel shift.

Because it's not about numbers, it's about definition in the image.

Since the HVX200 is already soft

Compared to what? Who says it's soft? In real world circumstances it's sharper than the Sony in 60i, way sharper than the Sony in 30F, and as sharp if not sharper than the JVC. Heck, even on a chart it's sharper:
http://www.fiftv.com/HVX200/3-Cams-Charts.JPG

[Giroud Francois]

this discussion is showing exactly what i was saying about the sad thing that happens to the HDV camera world.
You can take any CCD or CMOS of any resolution (after all any old CCD would fit), add a mumbo jumbo trick to magically multiply pixels, add a converter that will stretch to 1920x1080 and claim you have an HDV camera.
And then users will spend their time on forum to compare oranges and apples.
the good old time when a pixel was a pixel is gone....

[Robert Lane]

Barry,

This is just a curiosity but...

Why is the while balance in the HVX sample so much cooler than the others which are showing a more neutral response? Did they not balance white before doing the resolution test? I don't remember seeing this cool-white response from our HVX using the GMB color checker. I think.

You guys want to debate something that actually has real-world value? Figure out this color response thing in the resolution chart. Now THAT's something to ponder.