EX and HVX200

[Steven Thomas]

Originally Posted by Steve Mullen

Remember it was my math model's backward prediction of the HVX's pixel resolution that finally forced Panasonic to admit it was using SD chips. There was a VERY good reason they tried for months to hide this fact.

Steve, I remember this well.
I realize that companies must retain some confidential engineering deisgns, but I thought something was up when they evaded this question right until the camera was released.

I was on the preorder list for this camera. A coworker ended up getting his and we ran it through the paces. It's easy to get wrapped up with expectations. This stands true for eveything, but we were really let down by
the soft image and low light ability/noise.

At that time, we ended up staying with a couple JVC HD100.
Sure the HD100 had its faults, but we liked the overall image and that's what counts.

[Thomas Smet]

I know exactly how pixel shift works.

Is the HVX200 HD? Yes
does pixel shift work to make it HD? Yes
Is it as good as a native 1920x1080 chip? No
Does it really matter? Not really

Look the HVX200 is a great camera but it isn't as good as it could be. I personally think high detail is very overrated and I wish people would see that. Pixel shift is a great way to create HD but it isn't the only way nor is it the best way. Although that depends on your definition on what makes a better image. To detail junkies pixel shift just doesn't cut it. A single chip bayer pattern will give you more detail. To those who like a natural looking organic image the pixel shift works great. The reason I do not like pixel shift has nothing to do with the detail but other image issues which I will not go into because it is based 100% on my opinion and view.

[Gabe Strong]

Originally Posted by Steve Mullen

That's simply not true.
Sony brings each under-sampled frames into the EIP. The EIP then performs an INTERPOLATION on these pixels to GENERATE A WHOLE NEW FRAME.

This is an ACTIVE process whereas Pixel-shift is a PASSIVE process.

Don't try to claim interpolation offer the "same benefits of Pixel Shift under a different name." The TX Shoot-out showed huge resolution differences between the HVX and the V1. The MEASURED V1 rez about 800x800. (My math model indicates it is about 775x775 -- which is damn close.)

The HVX measured exactly what my math model predicts: 540x540.

Remember it was my math model's backward prediction of the HVX's pixel resolution that finally forced Panasonic to admit it was using SD chips. There was a VERY good reason they tried for months to hide this fact.

Thomas clearly shows why pixel shift doesn't work in SD. There is a word for DV cameras that had 270K chips. They UNDERSAMPLED. It takes sophisticated technology to CHEAPLY build hi-rez chips PLUS the DSP needed to increase their lower sensitivity. Sony has it. And, once again they have leaped over the competition. (JVC also builds exactly the CCDs they need for 720p.) It's time to accept that 1080i requires FullHD chips.

The next TX shoot-off will show the EX1 to deliver 1000x1000 lines. (My math model predicts 1048x1048.) That is 4X the rez of the HVX!

To hold to your belief in the face of both theory and data only makes it look like you are defending Panasonic and/or Canon.

I'm with you here Steve! Not bashing on the HVX, it is what it is, but technology moves on, sorry but a person can't always have the latest and greatest. Shoot, I'm still using a PD-150!

[Steven Thomas]

Originally Posted by Thomas Smet

does pixel shift work to make it HD? Yes
Is it as good as a native 1920x1080 chip? No
Does it really matter? Not really

Hmmm.
It sure matters to me.
I see a large difference, at least with the certain "pixel shift" camera we were using.
If one can not, I guess it does not.

[Craig Seeman]

It's not that pixels shift is a "bad" thing. Certainly the images look better than from those same cameras had they not employed pixel shift. The better way to look at it is, does pixel shift to a specific resolution look better than something shot at that specific resolution with "native" pixels. The answer is most likely no.

Pixel shift improves the resultant image. Native pixels would improve it more.

Does THAT matter? It will to those who can see the difference.

When the time comes, I hope there's a Texas shoot out between the HVX200 and the EX1 at 720p and 1080p.

I suspect the next hand held 3 chip HD camera from Panasonic will not be using 960x540 chips.

[Steven Thomas]

Originally Posted by Craig Seeman

Does THAT matter? It will to those who can see the difference.

My point exactly.

[Thomas Smet]

When it comes down to it the HVX200 will have one advantage and only one advantage and that is 4:2:2 recording to the cards.

Compared to 35mbits the whole I frame only thing doesn't matter as much anymore unless you like shooting strobe lights for a living.

* Low light will be better
* natural detail will be better
* quality will be better
* You get more video per card then with P2.
* You get SDI outputs and not just component outputs.
* You get better depth of field because of the 1/2" chips
* You get better transfer speed not to mention the cards transfer faster then the cards fill up with recorded video.
* The recording format itself has more resolution
* You can use the camera in a HDV type workflow if all your other equpiment is HDV.
* You get a real lens with real controls
* You "may" get 10bit video from live SDI. There are still conflicting reports on this one so until somebody from Cineform or somebody tests it we will not know for sure.
* You get 720p 24p at quality much higher that is virtually as good as Intraframe video because it has so many bits to work with. The thing with 720p is it can be as good as 1440x1080i video btu sadly it never was with DVCPROHD since it only used 960x720 pixels. 720p is already low enough without making it any softer and it really needs to be at 1280x720 to be able to go up against 1080i HD.
* The cards may not be cheaper but you get 3 times the amount of video on the same sized card which equals a huge cost savings. If somebody is used to shooting with 3 16GB P2 cards they could really get by with one SxS card.

Going back to the 4:2:2 thing. Sure the HVX200 records 4:2:2 to the card but if you really are serious about visual effects work chances are you wouldn't record your footage to the cards anyway. In this case the Ex1 with it's 100% digital SDI output and native higher resolution will give you much much better 4:2:2 quality material to work with. So really the 4:2:2 to card becomes more of a budget shooting sort of thing because serious FX work would capture uncompressed anyway.

[Kyle Self]

Originally Posted by Thomas Smet

I know exactly how pixel shift works.

Is the HVX200 HD? Yes
does pixel shift work to make it HD? Yes
Is it as good as a native 1920x1080 chip? No
Does it really matter? Not really

I beg to differ, it most certainly matters.

Go to the HVX forum and take a look at some post that were made not long ago. There was much dismay from the masses there when a couple of early adopters (who were the reason some of these people referenced as buying their cameras) coughed up the fact that they could NOT use their HVX 200 for a large number of projects they shot. Reason being, poor resolution.

One person who just recently bought a 200 for a project went and got a Cannon H1 because for one the resolution of the 200 wasn't up to the standard needed.

If it works for you great, but there are a lot of situations where it has not worked for people. One side effect of the EX is that it is probably going to force Panasonic to put real chips in the 200 and eventually the 500.

K

[Steven Thomas]

Yes, panasonic will have to step it up.
They have a large fan club. If they come up with a "true" high rez cam that offers something other than DVCPRO-HD.

I'm not saying you can not get decent footage from a DVCPRO-HD cam, of course you can. The format can not yeild full 1080P HD.
If that does not bother you, so be it - live with it for as long as you like.

Having said that, their higher end cams such as the HPX500 produce some real nice stuff.

[Chris Hurd]

Originally Posted by Steve Mullen

ClearVid... is an ACTIVE process whereas Pixel-shift is a PASSIVE process. Don't try to claim interpolation offer the "same benefits of Pixel Shift under a different name."

Your contention that one is active and the other passive takes nothing away from my original point, which you know to be true: both processes... Pixel Shift and ClearVid... both produce more sampling points per pixel, and both processes are indeed forms of interpolation which effectively boost the resolution of the image sensors. Sony's marketing emphasizes ClearVid's differences from "conventional spatial offset" (Pixel Shift), but my original point remains clear and valid: it begins with creating more sampling points per individual photosite. Both processes do this -- Pixel Shift and ClearVid. Both interpolate.

The TX Shoot-out showed huge resolution differences between the HVX and the V1.

If you're referring to the 2006 Texas HD Shootout -- the one everybody knows about, the one which Adam Wilt supervised -- then I have some bad news for you. The V1 most definitely was *not* part of that test. It wasn't even there. I'm the one who produced that event so I know exactly what we had on hand.

It's time to accept that 1080i requires FullHD chips.

It's time to accept that the most popular 1080i HD acquisition formats of the day (Sony HDCAM and XDCAM HD) have produced mountains of stunning program content *without* FullHD chips... thanks in no small part to Pixel Shift.

To hold to your belief in the face of both theory and data only makes it look like you are defending Panasonic and/or Canon.

Nonsense! As I stated before and will happily repeat once again -- and what you also know to be true -- is that Pixel Shift is found in almost every three-chip system in one form or another among *all* manufacturers -- Sony, JVC, Panasonic and Canon -- with exceptions so very rare as to be a cause of concern for product marketing when it's *not* used, such as the notable case of JVC's Pro HD line. JVC wisely chose to note in their Pro HD marketing material that Pixel Shift was not implemented in their CCD block; an excellent example of selling the difference (in other words, it's a feature, not a bug).

Steve, you know as well as I do that Pixel Shift is incoporated by *all* major manufacturers of three-chip camera systems both past and present. Sony's entire HDCAM and XDCAM HD camera lineup, for instance. Formerly in JVC's popular DV500 & DV5000 series camcorders, because they used Sony CCDs. And while JVC avoided it in their Pro HD line, they continue to use it in a big way (double-axis) in their Everio series.

I mentioned those Canon and Pansonic antiques (XL1, GL1, AG-EZ1) because somebody mistakenly referred to the HVX200 as establishing some sort of "precedent" for H+V Pixel Shift, when the fact is that double-axis spatial offset has been in use for more than a decade... and continues on in the JVC Everio camcorders. So obviously I am not "defending" any particular brand over another... just pointing out that double-axis H+V is more common than most folks might realize.

Originally Posted by Craig Seeman

...does pixel shift to a specific resolution look better than something shot at that specific resolution with "native" pixels. The answer is most likely no. Pixel shift improves the resultant image. Native pixels would improve it more.

Incorrect, if sensitivity and low-light performance are as crucial as many people here seem to think. If other factors are equal (sensor size, DSP etc.) then a lower pixel count with Pixel Shift will yield greater sensitivity than a higher pixel count without it, since fewer pixels on the same size chip means they're physically larger and therefore gather more light. So-called "native" pixels might improve sharpness, but at the cost of sensitivity which can be just as important.

But of course the DSPs (or EIP, whatever you want to call the processor) are seldom equal and make a *big* difference, as previously pointed out for the Sony V1 as well as Canon's Digic DV II and others. Sharpness and sensitivity are mutually exclusive terms, so it becomes a matter of trade-off. Which would you rather have interpolated... resolution or sensitivity? To some extent you can have a balance of both, thanks to resolution boosters such as Pixel Shift and ClearVid, and amplified brightness from the processor. That's an interpolation as well, and there's nothing wrong with it when it's done right... for example, how many Sony VX2000's and PD150's have wedding videographers purchased over the years specifically because of the artificially enhanced low-light performance of those camcorders?

Perhaps some folks tend to forget that CCD and CMOS image sensors are *analog* devices whose output must be converted to digital, so there's no reason why the pixel count of the sensor must be 1:1 with the pixel count of the recording format (so-called "native," seldom used for good reason). It doesn't matter if it's less, equal or more, because the signal must be converted from analog to digital anyway. The best camcorders within a certain class (HDV), the Sony V1 and Canon XH A1, produce superb 1920 x 1080 images from lower resolution chips, resolution interpolation and sensitivity interpolation: in the V1 it's 960 x 1080 with ClearVid and a state-of-the-art EIP; in the XH A1 it's 1440 x 1080 with H-axis Pixel Shift and Canon's proprietary Digic DV II. It's highly doubtful that "native" pixels would improve the image from either camcorder, but they would definitely take a hit on low light performance.

[Gabe Strong]

Well, I've worked with the Sony 350 XDcam and 330 XDcam and the HVX.....I know which I think looks better......and it's really not that close to my eyes, even with the cheap lens on the 330. Obviously the XDcam EX is not the 350 or 330 (The EX actually has a HIGHER resolution native chip) but maybe the lens isn't as good, I'm not sure about that since I haven't seen one. There's also the fact that the the XDcam has bigger chips. But if Panasonics contention is correct, this should surely be made up by the fact that XDcam is using the 'inferior' Long GOP 4:2:0 right? I guess after all is said and done, I'm going to believe my eyes......and my eyes say that the higher NATIVE resolution chips look better than the pixel shifted ones. So what is the reason for that? I can't say for certain (although I will continue to hold my belief that lower resolution pixel shifted chips don't look as good as HIGHER resolution pixel shifted chips).... But I can say which one looks better to me, and it's really not even a contest in that department.

[Thomas Smet]

Originally Posted by Kyle Self

I beg to differ, it most certainly matters.

Go to the HVX forum and take a look at some post that were made not long ago. There was much dismay from the masses there when a couple of early adopters (who were the reason some of these people referenced as buying their cameras) coughed up the fact that they could NOT use their HVX 200 for a large number of projects they shot. Reason being, poor resolution.

One person who just recently bought a 200 for a project went and got a Cannon H1 because for one the resolution of the 200 wasn't up to the standard needed.

If it works for you great, but there are a lot of situations where it has not worked for people. One side effect of the EX is that it is probably going to force Panasonic to put real chips in the 200 and eventually the 500.

K

Dude I hear you. When I said it doesn't matter I meant that to those who like the softer look it isn't as big of a deal. Some people like crisp video while some like softer video. I don't like the HVX200 but it is still a nice camera and for every person who thinks it is a SD toy there is another person who will swear by the quality and use it for a film out project with great success and make a lot of money from it.

These are all just tools and sure they all have different looks to them. It is all about the look you like and thats it.

[Chris Hurd]

Originally Posted by Gabe Strong

Well, I've worked with the Sony 350 XDcam and 330 XDcam and the HVX... and my eyes say that the higher NATIVE resolution chips look better than the pixel shifted ones.

None of these cameras... the Panasonic HVX200, the Sony PDW-F330 or F350... none of them have "native" resolution chips. The HVX200 employs Pixel Shift in both H and V axes. The Sony XDCAM cameras use Horizontal-axis Pixel Shift, just like all Sony HDCAM cameras and nearly every other three-chip camcorder system ever made.

[Gabe Strong]

Originally Posted by Chris Hurd

None of these cameras... the Panasonic HVX200, the Sony PDW-F330 or F350... none of them have "native" resolution chips. The HVX200 employs Pixel Shift in both H and V axes. The Sony XDCAM cameras use Horizontal-axis Pixel Shift, just like all Sony HDCAM cameras and nearly every other three-chip camcorder system ever made.

Chris,

I realize the Sony PDW-F330 and F350 don't have "native" 1920x1080 resolution chips, and that they too use some pixel shift,.....what I AM saying, is that they have higher "native" resolution than the HVX.....they are "Pixel shifted" from 1440x1080 to 1920x1080. The HVX on the other hand is "Pixel shifted" from a MUCH lower 'native resolution' to achieve the 1920x1080 pixels. So yes, you are right, neither one is "native 1920x1080" but the PDW F-330 and F350 are a HECK of a lot closer than the HVX.....And the fact also remains that the EX does look like it WILL have native 1920x1080 chips. I am saying that I can see the difference in the 1080 video from the 330 and 350 and the HVX.....and my belief is because the 330 and 350 have a 'higher resolution chip'.....but some claim it isn't higher resolution (because of the pixel shifting of the HVX it too puts out 1920x1080) so I was saying higher NATIVE resolution, and possibly using the term 'native' in error. What I meant to say, was it had more real pixels, or less pixel shifting.....sheesh, now I don't even know if I said that right, but hopefully someone understands what I am trying to say and can clarify.

[Craig Seeman]

Originally Posted by Chris Hurd

Incorrect, if sensitivity and low-light performance are as crucial as many people here seem to think. If other factors are equal (sensor size, DSP etc.) then a lower pixel count with Pixel Shift will yield greater sensitivity than a higher pixel count without it, since fewer pixels on the same size chip means they're physically larger and therefore gather more light. So-called "native" pixels might improve sharpness, but at the cost of sensitivity which can be just as important.

It depends on how the chip is designed. Have you read Juan Martinez' article in TVB about CCD compared to CMOS? He talks about what Sony has done to improve the sensitivity of the CMOS chips in the EX1.

I suspect the EX1 will have both greater resolution and better low light sensitivity (probably taking into account noise too) than the HVX200 under matching conditions. In short, it is possible for a chip with smaller pixels at 1920x1080 to have comparable or better sensitivity than a pixel shifted 960x540 chip and we may soon be able to verify that.