I'm not sure why Chris seems to think pixel shifting works.Well I'm not the only one who seems to think pixel shifting works. Apparently the engineering and design teams of nearly every major CCD manufacturer on the planet seem to think so too. Look, if spatial offset didn't work, then it wouldn't be such an integral part of almost every three-chip camera system ever made. Like I said -- it's there for a reason.

Well I'm not the only one who seems to think pixel shifting works. Apparently the engineering and design teams of nearly every major CCD manufacturer on the planet seem to think so too.
It does bring benefits, it does work, no doubt.

BUT I'm concerned that a belief has unfortunately arisen among some that it is a magic bullet - "pixel shift will cure everything". And that is not true. The best write up on the subject I've seen is from Panasonic - ftp://ftp.panasonic.com/pub/Panasonic/Drivers/PBTS/papers/AG-HVX200.CCD-WP.pdf - who claim that it will enhance native luminance resolutions by 1.5x. Though they do admit this to be a best case scenario, and more independent studies generally talk of 1.2-1.3x to maintain a decent mtf.

They claim an "effective" resolution for the HVX of up to 1440x810 (though admit that to be a best case scenario) - I'd reckon around 1200x675 to be nearer the mark, though it depends at what mtf measurement you call it a day.

Consequently, pixel shift does work - it does give an improvement over the results if it wasn't used - but let's all be realistic. Even Panasonic don't claim it will achieve anything like 1920x1080 from the HVX chipset.

BUT I'm concerned that a belief has unfortunately arisen among some that it is a magic bullet - "pixel shift will cure everything".No, no, no. I certainly hope I'm not giving that kind of impression. My long-standing annoyance has only been with those who have argued out of ignorance that Pixel Shift and the spatial offset process is some kind of gimmick, or worse, some sort of "marketing trick." It's neither. It's real, it works and it's there for a reason... but the technology is well over a decade old, and there are newer and perhaps better ways to boost resolution, as we've seen from Sony's ClearVid interpolation process.

I think we're eventually going to see an increasing rate of change to single-chip designs in the pro video market (as already clearly evident within the realm of Ultra-High Definition digital cinema acquisition systems, and as already happened long ago with digital still photography), so why should it even matter anymore.

Well I'm not the only one who seems to think pixel shifting works. Apparently the engineering and design teams of nearly every major CCD manufacturer on the planet seem to think so too.

Well, looking at the AG-HPX500, it sure seems to be working well.
I think part of this problem is a lot if us (including myself at time. LOL),
are comparing the "higher" native cams against the HVX200.

My understanding (and from some stuff I've seen) the HPX500 sure looks like its capable of producing some detailed clean images.

My long-standing annoyance has only been with those who have argued out of ignorance that Pixel Shift and the spatial offset process is some kind of gimmick, or worse, some sort of "marketing trick." It's neither. It's real, it works and it's there for a reason...

But the results of the Texas shootout showed that pixel shifting doesn't succeed at increasing the useful resolution of a camera, so what's the practical benefit to users compared to capturing at the native sensor resolution and upscaling later?

The other practical benefits are increased sensitivity (as previously explained) plus a lower manufacturing cost per CCD (which leads to greater availability and a less expensive camera). Resolution isn't everything... a fundamental concept to grasp.

Resolution isn't everything... a fundamental concept to grasp.

Of course not, but you haven't said how pixel shifting in the camera is beneficial compared to upscaling sensor data in post. I won't say more about this because you obviously feel strongly about it, but it sure seems like pixel shifting is mainly useful for marketing purposes.

Why do it in post when it can be done in the camera.

If spatial offset technology was "mainly useful for marketing purposes," then Panasonic would simply have announced the HVX200 as using double axis Pixel Shift right from the very beginning instead of keeping it a secret for so long. If it's there just "for marketing" then why don't Canon and Sony make a bigger marketing splash with it and inform more customers that H-axis Pixel Shift is used in the entire Sony XDCAM HD line and XL and XH series Canon camcorders? Why is it hardly ever mentioned by Sony, despite having used spatial offset for years in their full range of DVCAM and HDV camcorders?

JVC has been the only company I'm aware recently to make a point of mentioning spatial offset in their Everio camcorder product marketing (Panasonic doesn't count; they had to be prodded into disclosing their HVX200 specs).

The notion that Pixel Shift is mainly useful for "marketing purposes" is totally absurd, considering how seldom it's ever referred to in product marketing material. Prior to JVC, the only time I've ever seen it "marketed" was for the original Canon XL1 back in 1997 (despite the fact that all three-chip Canon camcorders to follow have used it in one form or another).

What I feel strongly about is the ludicrous notion that an integral component of three-chip camera design is there "just for marketing." I can hear engineers groaning over that nonsense all the way from Japan.

Of course not, but you haven't said how pixel shifting in the camera is beneficial compared to upscaling sensor data in post.
The diagram in the Panasonic paper gives a good description - though it's important to realise that the process works less well if the image is saturated in colour. Fundamentally, it enhances luminance resolution only - but since colour is normally subsampled, this shouldn't matter.

It relies on the green pixels being spatially offset from red and blue, simple upscaling assumes r,g,b to represent the same point.
I won't say more about this because you obviously feel strongly about it, but it sure seems like pixel shifting is mainly useful for marketing purposes.
Hmmm, well, here we're back to the "magic bullet" statements, aren't we? I won't disagree that it's benefits have been overstated in the past, whether by ignorance or design I wouldn't like to speculate. But that is not the same as saying it is of no use at all. (Chris - you weren't in my mind when I made the "magic bullet" comment.)

Returning to the Panasonic link, it's also worth noting one reason WHY they give for using pixel shift.
Knowing that the camera needed to do SD and HD, Panasonic surveyed the technologies and their respective “trade-offs”. If we used a 1/3" native resolution 1920 x1080 CCD, it wouldn’t have enough sensitivity to be practical in low light because the pixels would be too small to collect sufficient light.
Well - the answers simple when you know it, isn't it? Just use 1/2" sensors!!! The trick is then building the lens and camera small enough!

The notion that Pixel Shift is mainly useful for "marketing purposes" is totally absurd, considering how seldom it's ever referred to in product marketing material.

It's mentioned all the time in the form of the advertised recording resolutions of the cameras. 'Nuff said before I get censored.

It's mentioned all the time in the form of the advertised recording resolutions of the cameras. No it ain't.

The recording resolution is quite often a different thing from the resolution of the image sensor, which may or may not involve Pixel Shift (case in point: the VariCam and DVCPRO HD, see Thomas Smet above). I don't see how you can possibly draw a connection there. The recording resolutions are tied to the format and they are what they are. As I said, spatial offset is seldom mentioned.

The recording resolution is quite often a different thing from the resolution of the image sensor, which may or may not involve Pixel Shift...I don't see how you can possibly draw a connection there.

It's an obvious connection to me: would anyone have pre-ordered the HVX200 if it was marketed as a "540p" video camera? But enough of this: let's enjoy the fact that the EX1 has both a sensor and recording format with full 1080p resolution. That's a good thing which will matter to some users.

Pixel shift Smixel shift. The real question was how the EX-1 is going to handle fast moving action! Can't wait to get some real world reviews on this issue as now it is all speculation.

Man, I wish we all lived closer to get together and have a few beers. I bet we'd all start agreeing more. LOL

Sometimes it real hard to get your thoughts across via the internet.
I can easily see both sides on this one.

Pixel shift Smixel shift. The real question was how the EX-1 is going to handle fast moving action! Can't wait to get some real world reviews on this issue as now it is all speculation.

I'll tell you what. if it manitains the clarity I'm seeing in the frame grabs posted this morning, it's going to turn a lot of heads.

Man, I wish we all lived closer to get together and have a few beers. I bet we'd all start agreeing more. LOL

Sometimes it real hard to get your thoughts across via the internet.
I can easily see both sides on this one.

But could we all agree on what type of beer to buy? Thats the real question here. Does one camera taste great while the other may be less filling?

But could we all agree on what type of beer to buy? Probably not, but assuming we could all manage to get together, I'll gladly provide a round of Shiner Bock (http://www.dvinfo.net/conf/showthread.php?p=526561) for everybody.

Sounds good to me!

http://www.shiner.com/beers/images/Bock.gif

I love these discussions. Look. I've used the XHA1 quite awhile. I've used the HD110, hv10, blah blah blah...

The XHa1 produced some amazing footage. I loved the imagery. I just got the HVX a few weeks ago and I can tell you that I prefer the HVX HD over the canon.

And when the EX comes out, I'll have one. But I don't expect it to be vastly superior to the HVX or Canon for that matter. I said it before, the HVX is the best video tool out there because it does everything. It doesn't do it all better, but it does everything. The EX1 will do some things better, but it doesn't do everything.

I, like many other Canon users were trying to set our cameras up to "look like HVX" footage in the area of color.

I'm excited about the EX1 but I expect it will do that one thing slightly better. And that's resolution. But I'm more than pleased with the final output of the HVX.

Okay, everybody drink two beers and then vote: is the EX1 camera body uglier than the HVX200 or about the same? :-)

Thanks to everyone for a friendly and amiable way to conclude a hot thread.