View Full Version : Pixel shifting? What is it exactly?


David Delaney
March 4th, 2007, 09:25 AM
What does Pixel shifting really mean? I know that the HVX has it, but how does it differ from other hd cameras like the Z1u or the FX1?

Peter Jefferson
March 4th, 2007, 09:33 AM
from my understanding of it, its an in-cam interpolation algorythm..

John Bosco Jr.
March 5th, 2007, 09:34 PM
Basically it's shifting the green CCD 1/2 pixel biaxially, both H and V. This increases the 960 x 540 progressive chips to HD resolution. Panasonic's thought is larger pixels equal greater sensitivity, producing adequate results in low light from the 1/3rd inch CCDs.

As far as other cameras go, most achieve or are close to achieving hd resolutions. To compare, HVX200 has 500,000 pixels and canon XHA1 has over 1,600,000 pixels. The canon achieves the necessary 1440 x 1080i resolution for HDV. Right off you would assume the camera from canon to be the sharper camera. However, HDV compresses the video 25mb/s; whereas, the HVX200 uses DVCPROHD compression at 100mb/s. All this means is that the HDV camcorders suffer more when there is fast motion in the image because of compression artifects (usually in the form of macroblocking). However, if you were to put these cameras side by side, you would be hardpress to pick an outright winner. The canon looks a bit more washed out but sharper than the Panny. Each has its pros and cons. Some prefer the soft look and richer colors of the HVX200, while others like the sharper image from the XHA1. Of course, HDV camcorders compress audio as well; whereas, the HVX200's DVCPROHD codec produces uncompressed audio. Another advantage of the HVX200 is variable frame rates. Of course, the canon has a longer lens.

Robert Lane
March 5th, 2007, 11:48 PM
Barry Green has a very good technical overview of what it really is and how it's accomplished - I just can't find it anywhere...

David Heath
March 7th, 2007, 06:30 PM
Basically it's shifting the green CCD 1/2 pixel biaxially, both H and V. This increases the 960 x 540 progressive chips to HD resolution.
It's worth noting that this resolution increase applies mainly to luminance, and is most effective for a subject with low colour saturation. For a highly saturated subject, pure red lit, say, all the output will be from the red sensors, none from the green or blue, so the pixel shift technology can't have any effect.

Since the human eye is more sensitive to luminance resolution than chrominance, and most TV systems have more limited chrominance resolution anyway, it should generally be seen as a good thing.

Chris Hurd
March 7th, 2007, 06:38 PM
Pixel Shift is a resolution boost that has been around for a very long time, and yes it is a Good Thing (beware of various marketing tactics that try to portray it as anything else). If I recall correctly, Panasonic invented it, but most all three-chip camcorder manufacturers use it these days. It is the rare exception for a three-chip camcorder *not* to use Pixel Shift, such as the JVC Pro HD series GY-HDxxx. I've described it several times before in the past, but there's a good technical write-up from Panasonic floating around somewhere... stand by while I dig it up.

Chris Hurd
March 7th, 2007, 06:52 PM
Okay, here it is; the best description of Pixel Shift that I know of.

Go to this page: Learn about Panasonic's AG-HVX200 (http://catalog2.panasonic.com/webapp/wcs/stores/servlet/ModelDetail?displayTab=R&storeId=11201&catalogId=13051&itemId=93120&catGroupId=34401&surfModel=AG-HVX200)

Under the "Support and Resources" tab, click the link at the bottom of the left column on that page, entitled "Tech Paper - AG-HVX200's Advanced Progressive CCDs" -- your description of Pixel Shift is found in that PDF document.

As I said, most all three-chip camcorders use it. Everything by Canon, everything by Sony, everything by Panasonic. Some earlier products from JVC. Hope this helps,

John Bosco Jr.
March 8th, 2007, 05:00 AM
Are you sure "everything by Canon?" The XHA1, G1 and XLH1 all have native HDV resolution chips. I believe Canon abandoned pixel shift with the XL2. Unless I'm missing something. It's funny that they did that, though, because they used pixel shift with success in the XL1/1S cameras.

Peter Jefferson
March 8th, 2007, 07:53 AM
from my understanding, Canon used Panasonic sensors for their XL/GL series cameras

Barlow Elton
March 8th, 2007, 10:51 AM
Are you sure "everything by Canon?" The XHA1, G1 and XLH1 all have native HDV resolution chips. I believe Canon abandoned pixel shift with the XL2. Unless I'm missing something. It's funny that they did that, though, because they used pixel shift with success in the XL1/1S cameras.

The XL-H1, XH-A1 and XH-G1 employ horizontal pixel shift to achieve a 1920x1080 image internally before scrunching back down to the HDV 1440x1080 format.

These cameras do have all the native pixels needed for the actual recording format, but rely on less aggressive pixel shift in order to achieve full raster processing in-camera.

Chris Hurd
March 8th, 2007, 10:02 PM
Are you sure "everything by Canon?"Yes, I am sure, everything by Canon.
I believe Canon abandoned pixel shift with the XL2. Incorrect. As Barlow points out above, all three Canon 3-chip HDV camcorders (XL H1, XH G1, XH A1) have the exact same CCD block, and it does in fact employ the Pixel Shift process in the horizontal axis. Since each CCD is already 1440 x 1080 to begin with, the resulting boost in resolution puts them in the ballpark of 1920 x 1080 before going to tape.

John Bosco Jr.
March 9th, 2007, 04:10 AM
Yes, I am sure, everything by Canon.
Incorrect. As Barlow points out above, all three Canon 3-chip HDV camcorders (XL H1, XH G1, XH A1) have the exact same CCD block, and it does in fact employ the Pixel Shift process in the horizontal axis. Since each CCD is already 1440 x 1080 to begin with, the resulting boost in resolution puts them in the ballpark of 1920 x 1080 before going to tape.

Got U... Thanks for the added information. I stand corrected.

John Bosco Jr.
March 9th, 2007, 04:23 AM
I have a question about sensitivity. Maybe it has been addressed before, but here it goes.

It is Panasonic's theory that full HD resolution on 1/3rd inch chips would not make the camera sensitive enough. So how does the HVX200 compare with the XHA1/G1/XLH1 in low light situations?

Peter Jefferson
March 9th, 2007, 09:45 AM
As Barlow points out above, all three Canon 3-chip HDV camcorders (XL H1, XH G1, XH A1) have the exact same CCD block, and it does in fact employ the Pixel Shift process in the horizontal axis. Since each CCD is already 1440 x 1080 to begin with, the resulting boost in resolution puts them in the ballpark of 1920 x 1080 before going to tape.

this is why the SDI out of the G and H1 is so popular.. full res straight out of the box

Barlow Elton
March 10th, 2007, 02:26 PM
I have a question about sensitivity. Maybe it has been addressed before, but here it goes.

It is Panasonic's theory that full HD resolution on 1/3rd inch chips would not make the camera sensitive enough. So how does the HVX200 compare with the XHA1/G1/XLH1 in low light situations?

The HVX and A1 have similar sensitivity, but a different rendition of noise. Basically, the A1's noise is finer, grittier, more luma whereas the HVX has more swimming, larger-sized chroma noise.

As far as Panasonic's theory goes, it's surprising that the A1 has just over 1.5 MP and the HVX has just over 500K, yet they are similarly sensitive. (according to Adam Wilt's tests) By all rights, the HVX should have much better sensitivity and less noise because its pixels are physically larger. I guess this is where theory can go out the window.

I know that the A1's interlace CCD's give it a bit of a sensitivity advantage over progressive, but I'm sure it's not completely apples to apples. They're simply different sensors with different capabilities.