View Full Version : HVX200 CCD specs?


Steve Mullen
January 3rd, 2006, 10:39 AM
Does anyone know the native aspect-ratio of the HVX200 CCDs?

The pixel aspect-ratio?

The number of H and V CCD elements without playing the Green-shift game?

Chris Hurd
January 3rd, 2006, 11:10 AM
Panasonic chooses not to release that information at this time, just like Canon chooses not to reveal the bit depth of the DSP on the XL H1. I don't think the U.S. marketing departments have any say on this policy. I believe these decisions, like the cameras, are made in Japan.

Shannon Rawls
January 3rd, 2006, 12:10 PM
I thought Canon freely said at the DVEXPO that the DSP is 8bit.

No?

- ShannonRawls.com

Chris Hurd
January 3rd, 2006, 12:54 PM
I thought that was the SDI ouptut, not the processor.

?

Martin Doppelbauer
January 3rd, 2006, 01:13 PM
My guess is Panasonic is keeping the specs a secret for a very good reason.
HVCPRO HD only records 1280 columns in 60i so probably the chip is doing just 1280x720. But we'll know for sure as soon as the first resolution chart shots are uploaded.
Martin

Robert Graf
January 3rd, 2006, 01:16 PM
The light sensitivity has already been announced.

The pixel count doesn't matter much because the *real* question is how well the camera resolves detail. While CCD resolution does have an impact on this ability, there are other factors that will degrade the camera's ability to resolve detail (and potentially in a big way). My advice is for people to shoot a resolution chart (properly), and see how the camera manages under actual stress tests.

Cheers,
Bob

Steven Thomas
January 3rd, 2006, 01:28 PM
Martin,
do a bit of searching here, DVXUSER and cow. This topic has been discussed heavily, not to say it's not important.

Myself, I could care less how they are acheiving it, just as long as it looks good. Jan from Panasonic has mentioned the in-house resolution chart numbers; although, this might of been a preproduction model.

I can't remember the numbers, although I believe it was stated as good as the Sony Z1, or better.

Steve

Barlow Elton
January 3rd, 2006, 02:01 PM
My guess is Panasonic is keeping the specs a secret for a very good reason.
HVCPRO HD only records 1280 columns in 60i so probably the chip is doing just 1280x720. But we'll know for sure as soon as the first resolution chart shots are uploaded.
Martin

After puzzling over footage and uprezzing 720p to 1080p, and looking at native 1080p, I think this is spot on. This camera is a Varicam lite, which is AWESOME, but it's pretty obvious to a lot of folks that the 1080p looks an awful lot like uprezzed 720p.

All that said, I am deeply in love with 60p. It's so crisp and lucid.

The low light noise does worry me a bit. I cranked the gain on the H1 and it was amazingly clean, and that camera is DEFINITELY better in low light, but at what cost? Is the HDV tradeoff worth it? One thing I hope to test in the H1 is it's noise suppression feature. It will be interesting to record this camera's SDI into a Decklink, trying lots of codecs.

The jury's still out...

Petr Marusek
January 3rd, 2006, 02:25 PM
Pixel shift generally increases resolution about 1.5x. DVCPRO HD 1080-60i records 1280x1080 pixels. That means that 853x720 pixel CCDs could be used with vertical and horizontal shift. Panasonic's 10x as expensive, when we add cost of lens, etc., model 400 is using 1280x720 chips for 1080-25p.

Progressive scan CCDs need 2x as much light. Assuming the same pixel light efficiency is achievable by Sony and Panasonic, the 853x720 progressive CCDs would produce nearly as good low light performance as Sony's Z1 960x1080 interlaced chips. Both cameras are rated 3 lux.

My guess is that Panasonic did the most to keep the low light performance high and used something like 853x720 pixel chips, which is fine. Until someone looks at the chips under a microscope, we may never know for sure.

Although most of us are courious, the only thing that really matters is performance. If Panasonic achieved it with native 1080p chips or with lower pixel count chips with pixel shift is not that important.

Martin Doppelbauer
January 3rd, 2006, 10:57 PM
The well respected German magazine "Videoaktiv Digital" has recently compared the Canon XL H1 (1440x1080 CCDs) and Sony Z1 (960x1080 CCDs). In their december/january 2006 issue they also published enlarged crops of some of the test charts they shot.

They found both vertical and horizontal resolution of the H1 around 800 lines per picture height (lph).

It is interesting to see that the horizontal resolution is already beyond the theoretical limit of the DVCPRO HD system used in Panasonic's HVX200 (which would be 1280/16x9 = 720 lph - not regarding the influence of the lens and picture postprocessing).

The Z1 was tested 610 lph horizontal and 750 lph vertical.
To me this shows that pixel shifting can hardly compensate for lack of native CCD resolution although the results must have been even lower without it (960/16x9 = 540 lph at most).

But then the Z1 is just half the price of an H1 (and street prices of the FX1 are just one third) so overall that seems pretty fair.

The first resolution figures published of the HVX200 are 700 lph horizontal and 750 lph vertical (see http://zinner.org/) which I personally find a little disappointing considering the HVX200 in working condition (i.e. including a set of 2 times 8 minutes P2 cards) carries about the same price tag as the XL H1. But there's no reference where the figures came from (official Panasonic ?).

Michael Pappas
January 3rd, 2006, 11:13 PM
Where are these res tests at? I looked at Zinner, but no luck in finding them...

Tom Roper
January 4th, 2006, 12:02 AM
Martin, there's reason to be skeptical.

1.) Resolution gives a different result at any combination of zoom and aperture.

2.) But moreover, 800 lines horizontal implies a somewhat remarkable 98.8% efficiency since only 810 lines are possible for a 1440 sensor.

3.) Yet in the vertical, 800 lines only represents 74% efficiency for the same lens.

Unless you accept that the glass somehow suffers astigmatism in the vertical, it's a better idea to put your trust in the picture.

Shannon Rawls
January 4th, 2006, 03:26 AM
Don't be disappointed Martin. I have a feeling the HVX-200 will do much better then Zinner claims it will.

- ShannonRawls.com

Martin Doppelbauer
January 4th, 2006, 12:14 PM
Martin, there's reason to be skeptical.

1.) Resolution gives a different result at any combination of zoom and aperture.

2.) But moreover, 800 lines horizontal implies a somewhat remarkable 98.8% efficiency since only 810 lines are possible for a 1440 sensor.

3.) Yet in the vertical, 800 lines only represents 74% efficiency for the same lens.

Unless you accept that the glass somehow suffers astigmatism in the vertical, it's a better idea to put your trust in the picture.

1.) Yep - absolutely true. In another test with different apertures (http://www.repaire.net/site/tournage/compar_Z1Sony_HD100JVC/comparatif_Z1_Sony_&_GYHD100_JVC.php ) the Z1 performed constant between full open and two third closed. Only apertures below f8 were somewhat reduced. You will find that the author of this test rates the Z1 much higher than the German magazine. But I have no similar tests of the XL H1 and the mag says nothing about the focal length and aperture they used in the tests.

2. and 3.) The drop in vertical resolution can be explained by the fact that the Canon records interlaced. Usually vertical filtering is needed on interlaced material to reduce flickering effects. 700 to 800 lph seems the typical vertical resolution for 1080i.

Here is where the magic of the HVX200 was ment to come into play. Panasonic says it's the first true progressive 1080p Handycam on the market. If that is true, vertical resolution should be in the area of 1000 lph.
But if it's only around 700 to 800 lph (or just uprezzed 720p) then the same resolution and the 25p frame rate could be achived from the Z1/FX1/XLH1 by simply deinterlacing 1080i material in postproduction.

So we are still waiting impatiently for the first published resolution chart shots ;-)

Steve Mullen
January 4th, 2006, 02:02 PM
Pixel shift generally increases resolution about 1.5x.


My guess is that Panasonic did the most to keep the low light performance high and used something like 853x720 pixel chips, which is fine.

Green pixel shift can give UP TO 1.5X only on static images that are of certain colors. That, of course, is a rez. chart and not real motion video. For this reason, Green-shift is a good way to slightly increase detail when the the CCDs are of the SAME resolution as the recording format.

It is not a good way to try to obtain the resolution for a format that has a much high rez. than the CCDs.


I believe the CCDs are 960x720 which matches the 720P DVCPRO HD codec. However, both the CCDs and the codec are under-sampling in terms of the ATSC 720p format which is 1280x720 and 720p HDV.

When it comes to the 1080i ATSC spec. of 1920x1080 -- it's obvious that both HDV and DVCPRO HD under-sample horizontally as does HDCAM. However, DVCPRO with a 1280 rather than 1440 spec. under-samples more. And, if the CCDs are only 960x720 then they are under-sampling even for an under sampled codec.

This WILL NOT be apparent on rez. charts because if both H and V green-shift are used, the 920 will increase enough for 1280 and the 720 will increase enough for 1080.

Thus rez charts will look fine, but the real test is either a non B&W rez chart or real video.

Shannon Rawls
January 4th, 2006, 02:35 PM
I believe the CCDs are 960x720 which matches the 720P DVCPRO HD codec. However, both the CCDs and the codec are under-sampling in terms of the ATSC 720p format which is 1280x720 and 720p HDV.

If that's the case, is it safe to call the Panasonic HVX-200 and the Sony Z1U a "FAKE HD" camera since they come up short on both versions of HD?
Especially since everybody is so hellbent on calling things TRUE & FAKE based on what's happening at the sensors and not the final recording.

Just curious.

- ShannonRawls.com

Keith Wakeham
January 4th, 2006, 02:36 PM
There is are a couple of ways to find out for sure the pixel could, unfortunately both of these mean dissassembly of a HVX200.

Hook an osscillascope up to all pins and record waveform off a ccd. Eventually you will figure our what is the horizontal clock, vertical clock, and the reset and shutter clocks. From that you can figure out all the pixels, link up to the ccd front end and find when the output is active. This will give you an indication of active, but not as accurate as the method to figure out the number of total pixels driven.

The other way is to rip off the lens and use a metal microscope (reflective not transpartent) and measure pixel size. But since your already their, you could just start counting.

Either way a scarifice could be made. Anyone got one? I'll do it, I promise i'll be very careful!

Les Dit
January 4th, 2006, 02:55 PM
Shannon,
I'd say yes to the 'Fake' label.

To me, when I think of HD from the old days, it was 1920 across. As in 20 years ago. NHK et al. Now they have taken the HD logo and watered it down to all kinds of flavors. To compensate for this, they apply highly excessive sharpening to make it look sharp to first glance prosumers. Other than chopped off high lites and shadows, this sharpening is a dead giveaway to the 'video look'. And yet they talk out of the other side of their mouths about film makers. It's funny. It's all about selling.

If the last couple of years results in a resolution barely better than the old JVC HD10, it will be pretty sad. They would be taking the prosumers for a HD label ride. Like the H2 Hummer = Chevy Tahoe with a new skin. ( candy coated ).

-Les

If that's true, is it safe to call the HVX-200 and the Z1U a "FAKE HD" camera since they come up short on both versions of HD?
Especially since everybody is so hellbent on what to call stuff because of what's happening at the sensors.

Just curious.

- ShannonRawls.com

Petr Marusek
January 4th, 2006, 02:58 PM
Green pixel shift can give UP TO 1.5X only on static images that are of certain colors. That, of course, is a rez. chart and not real motion video.

I believe the CCDs are 960x720 which matches the 720P DVCPRO HD codec.

Thus rez charts will look fine, but the real test is either a non B&W rez chart or real video.

That is very interesting. What is amazing are the frame grabs that the Italian guy posted that show how much sharper H1 is compared to Z1 on color images. Looking at the resolution charts, which are B/W, the differences not that big.

Graeme Nattress
January 4th, 2006, 03:05 PM
Early HD was analogue, so it didn't have any pixels across. HDCAM is 1440x1080, so not even that is full raster. Recently we got the terribly expensive HDCAMsr which is 1920x1080 - finally.

To call HD fake or not fake is a joke. Just look at the specs and decide yourself what you think, and look at the picture too and decide on that.

Personally, I think RED has the right idea - skip all this HD nonsense and go straight to Ultra-Definition.

Graeme

Shannon Rawls
January 4th, 2006, 03:12 PM
To call HD fake or not fake is a joke.

Graeme, that's the same song I was trying to sing about HDV being real HD. Also about the Canon XL-H1 and its ability to record 24p.
But I got slammed for it.
It didn't matter to them that it 'records' 24p, everybody kept pointing at the sensors and told me to use that as the definitive answer.

OK, well, hell......don't stop looking at the XL-H1's sensors.....look at EVERYBODYS sensors! if you define "24p" by what's happening at the sensors, shouldn't we define "HD" by what's happening at the sensors as well?

HDCAM may not meet the mark for 1920x1080, but the sensors clearly meet the 1280x720 mark, so it gets a "TRUE HD" pass, don't it?

For the record, I think it's all stupid. I define a camera by what it delivers me, not by how it creates it. I'm just playing devils advocate since I got ripped a new one for calling the XL-H1 a 24p camera.

- ShannonRawls.com

Les Dit
January 4th, 2006, 03:14 PM
Absolutely. In those 'video' days, it was Mhz not pixels.

I agree, Matsushita and Sony need to be dealt with, with companys that have no history in selling tape and $80K tape decks. An Arri D20 looking thing is the right direction, in many ways.
-Les


Early HD was analogue, so it didn't have any pixels across. HDCAM is 1440x1080, so not even that is full raster. Recently we got the terribly expensive HDCAMsr which is 1920x1080 - finally.

To call HD fake or not fake is a joke. Just look at the specs and decide yourself what you think, and look at the picture too and decide on that.

Personally, I think RED has the right idea - skip all this HD nonsense and go straight to Ultra-Definition.

Graeme

Petr Marusek
January 4th, 2006, 03:17 PM
HDCAM may not meet the mark for 1920x1080, but the sensors clearly meet the 1280x720 mark, so it gets a "TRUE HD" pass, don't it?

HDCAM CCD's are 1920x1080.

Graeme Nattress
January 4th, 2006, 03:25 PM
To say it's "the sensors" whether it be real 24p or not, or real HD or not is close, but it's a touch over-simplisitic. I think you have to take a holistic approach and look at the entire image process.

With the canon, it looks like you've got 24p from de-interlaced 48i, which to me, is not 24p, but very close. Certainly closer than CF24. So is it true 24p - no, but is it fake 24p, only a little :-) The end results look good, and we can read between the lines to figure out what it's doing. It would have been better if Canon had just said outright what they're doing and not kept quiet, but hey... What do you expect. So the answer is not TRUE or FALSE, but somehwere inbetween.

As for the Panasonic, we know it's using some pixelshift, which can in most cases give real resolution. DVCProHD rightly or wrongly trades some of it's luma rez for chroma rez, over and above what the other formats are doing. Does this make it any less real HD - no, but it does give a different quality, and the human eye is best to judge that, not a spec sheet.

Graeme

Chris Hurd
January 4th, 2006, 03:38 PM
HDCAM CCD's are 1920x1080.Incorrect. HDCAM CCD's are 1440 x 1080, with a PAR (pixel aspect ratio) of 1.333:1. HDCAM SR is 1920 x 1080 with a PAR of 1:1. But this is all... so very pointless.

Chris Hurd
January 4th, 2006, 03:43 PM
I'm just playing devils advocate since I got ripped a new one for calling the XL-H1 a 24p camera.You're supposed to be playing devil's advocate on other websites. I thought that's what I was paying you for. I'd really rather not have devil's advocates running around this place. They burn the soles of my sneakers when I stamp 'em out.

You did not get "ripped a new one" for calling the XL H1 a 24P camera. It pretty much *is* a 24P camera for all intents and purposes. Where you got into trouble was not in what you said but *how* you're saying it. And there's the problem. Hope this helps,

Graeme Nattress
January 4th, 2006, 03:44 PM
Incorrect. HDCAM CCD's are 1440 x 1080, with a PAR (pixel aspect ratio) of 1.333:1. HDCAM SR is 1920 x 1080 with a PAR of 1:1. But this is all... so very pointless.

The CCDs on the HDCAM are 1920x1080 with a PAR of 1. The image is subsampled to 1440x1080 by the HDCAM codec, but you can see the full resolution and detail out of the SDI.

Indeed it's pointless - you've got to look with your eyes, and not just look at one spec, but all the specs and fully understand how those specs interact.

All video cameras are a compromise - but different people like different features, and different manufacturers compromise in different ways - this is good.

Graeme

Chris Hurd
January 4th, 2006, 03:46 PM
I always look to you to get me squared away, Graeme... much appreciated! My apologies to Petr Marusek.

Graeme Nattress
January 4th, 2006, 03:51 PM
No probs Chris. I often wonder how I keep it all straight in my head - all those utterly useless factoids...

Graeme

Shannon Rawls
January 4th, 2006, 03:54 PM
You're supposed to be playing devil's advocate on other websites. I thought that's what I was paying you for.

WELL I AIN'T GET MY CHECK YET!!!! You keep telling me to wait til next week, but you've been saying that for 6 years now. DAMMIT I WANT MY PAYCHECK!!!

and can I get my "well there you have it" post back that you just deleted since YOU WERE WRONG and graeme just checked your butt!

- ShannonRawls.com

Chris Hurd
January 4th, 2006, 03:57 PM
can I get my "well there you have it" post back since YOU WERE WRONG and graeme just checked your butt!

When Graeme or anyone else for that matter checks my butt, it's note-worthy, fun, and definitely adds to the conversation in way or another. Pointless one-liners don't add anything though. And there you have it.

Barlow Elton
January 4th, 2006, 05:23 PM
Indeed it's pointless - you've got to look with your eyes, and not just look at one spec, but all the specs and fully understand how those specs interact.

I have a challenge for anyone near an Apple store or who owns a Mac with that gigantic 2560x1650 monster. Well....first of all...if you think it's reasonable to say that Kaku's raw clips are fairly indicative of the resolution and quality of the shipping camera...then please read on.

Take the "HVXcity108024p" and "HVXcity72060p" clips. Open them both in QT player side-by-side. Drag the 72060p clip with the "show movie info" window also open. Drag the lower right handle of the 60p clip until it becomes a 1920x1080 frame. The movie info window should update while you're dragging and you can follow the numbers until they reach 1920x1080.

Reposition both clips side by side or on top of each other if you can.

Now look at the fine details. Viewing these clips with such an unforgiving, full-rez monitor seemed to make both shots about as naked as they could be, at least in terms of perceiving resolution.

Can anyone honestly say that both clips don't look nearly identical? I had two store employees (unbiased curiosity) look closely at both clips, and thought they looked the same, in terms of sharpness and fine details.

Somebody else should try this...to my eyes the 1080p looks quite obviously like an uprez. I think it still looked good, just fairly apparent that there weren't many more pixels acquired originally in 1080p mode.

Mike Marriage
January 4th, 2006, 05:43 PM
Can anyone honestly say that both clips don't look nearly identical? I had two store employees (unbiased curiosity) look closely at both clips, and thought they looked the same, in terms of sharpness and fine details.

Somebody else should try this...to my eyes the 1080p looks quite obviously like an uprez. I think it still looked good, just fairly apparent that there weren't many more pixels acquired originally in 1080p mode.

Maybe people who claim they can should sit a "blind" test.

Barry G said that res charts on a pre-production show a 25-30% increase for 1080p over 720p. I wonder if the increase is really so noticeable on a real-life COLOUR, motion pictures..?

Rob McCardle
January 4th, 2006, 05:48 PM
Interesting - I can see what you're saying.
As to the tech details - I haven't got a clue.

To me the 720 clip scaled looks sharper/better/ than the native 1080.

Also rather than drag - in qt player go Cmnd-J, click on the video track and enter 1920 in scale.
Also step thru the clips by using the <- & -> arrow keys to go frame by frame.

Barlow Elton
January 4th, 2006, 05:58 PM
To me the 720 clip scaled looks sharper/better/ than the native 1080.

That's exactly what one of the employees thought...and he teaches FCP and other apps at the store. He's familiar with HD in general.

It's an interesting comparison. I showed them HDCAM to DVCPROHD 24p clips, and 1080i clips from the Panasonic sampler DVD that came with FCP 4.5 for reference. It was easy to see the difference in perceived resolution.

Rob McCardle
January 4th, 2006, 06:01 PM
heh- well, when I get around to buying one of these, which will be when they release the PAL version, I'm going to be saving me some disc space !
cheers, thanks for that.

Barlow Elton
January 4th, 2006, 09:08 PM
heh- well, when I get around to buying one of these, which will be when they release the PAL version, I'm going to be saving me some disc space !
cheers, thanks for that.

I think that's one of the benefits of the camera. You really only need to shoot 720p, because the differences natively are miniscule, unless Kaku's clips are some kind of aberration.

I lean towards the H1 for other reasons, but if I were to get an HVX, it's because I LOVE 720p! P2 seems more manageable in this format. Also, I think with the new uprezzing algorithims coming onto the market, this format will look quite nice if needed at 1080 resolution. I think in reality, the HVX is a cool little Varicam. It's strength isn't raw resolution, but it's incredibly deep options for filmmakers. There's certainly a lot more to image quality than the pixel count. The noise issue on the other hand...well, that's an issue that will rear it's ugly head if there aren't any easy tweaks for it.

What I think is a bit disingenuous is the claim that the camera gives you a native 1080p. It's just a larger file.

Marc Olivier Chouinard
January 4th, 2006, 10:26 PM
If I trust the information in the FAQ page on panasonic website :
http://shop.panasonic.com/webapp/wcs/stores/servlet/prModelDetail?storeId=11301&catalogId=13251&itemId=95592&modelNo=Content12072005012903035&surfModel=Content12072005012903035

It says this :
What size is the 3-CCD imager?
The imagers are 1/3" CCD, 16:9 native aspect ratio. These are scanned and captured at 1080/60p, and the signal is then converted to 1080i, cross converted to 720p or down converted to 480p/480i, or cross converted for the many modes on this camera. This assures the highest quality of recording.

So to my knowleadge, they say the CCD scanned at 1920x1080 in progressive at 60 frame per seconds.

If they didnt say it scanned... Then ok maybe it not, but they did. Now it could be a mistake from Panasonic that they are saying this. Maybe the word scanned should'nt be there. But that is what they are saying to me ;)

Barlow Elton
January 4th, 2006, 10:32 PM
Do you just trust what a company claims in their FAQ? Try resizing that 720 clip and compare it to the native 1080. Let your eyes be the judge, not specs.

Pete Bauer
January 5th, 2006, 08:00 AM
Marc,

Fine point: the FAQ didn't explicitly say 1920x1080/60p, and no way are they building and marketing a $6K camera with a 1920x1080 photosite CCD block. Pending further information, l'd simply interpret that as "X x 1080/60p."