After two months with the A1

[Mikko Lopponen]

And still the hv20 looks noticeably better than the z1. :)

[Matthew Nayman]

can I see some of this footage? I own an A1, and don't think I see what you folks are looking at.... and I have been looking at video professionally for years.

Granted, it's always in a cinematic setting, so nice slow pans, dollies, some hand held stuff... but I see no breakup in complex scenes.

[Mikko Lopponen]

I did a fight scene last week with the hc1 & hv20 and both held up compression wise very nicely. It seems strange that even panning would completely break up the compression on the a1. Maybe something else??

[Richard Hunter]

Originally Posted by Chris Hurd

All things *not* being equal -- say we we have two 3-chip systems, pixel counts on the chips are *not* identical in each. One system has chips with a lower pixel count than the other. Chip size is identical in each, but the system with fewer pixels on the chips also uses Pixel Shift. Let's say the resolution boost provided by Pixel Shift equals the native resolution of the other system. The advantage still goes to the system using Pixel Shift, this time because of greater sensitivity and better low light performance. Fewer pixels on the chip means the pixels are physically larger and therefore gather more light than a chip of equal size containing more pixels.

Hi Chris, thanks for the detailed reply. The problem I have in accepting this is that from other things I have read, and also what I have seen for myself, some of this is not borne out in real life cameras.

I have no experience with HDCAM and UHD, so let's take a couple of examples based on the XL1S that I know you are very familiar with. Maybe you can explain where I am going wrong with this.

First, there's sensitivity. The sensor block on the XL1S has a lower sensor count than that on a VX2000, and performs pixel shift to increase the resolution. Yet the VX2000 which is also 1/3" performs so much better in low light. If pixel shift really has the sensitivity advantages mentioned, this should not be the case.

Second, image sharpness. When the XL-2 was introduced, with its higher count sensor block, one of the big improvements over the XL1S was the sharpness of the image. It seems to me that while pixel shift can allow a lower resolution sensor block to create the same number of data samples as a native resolution sensor block, it does not necessarily produce an equally sharp image.

I do understand that cramming more and more sensors onto small sensor blocks, such as is happening with small consumer cams, brings its own problems and does not always lead to improved video pictures just by having high pixel count. I also appreciate that pixel shifting allows us to get good results at more affordable prices. I just find it hard to accept that everything about pixel shift is as marvellous as the manufacturers would like us to believe.

Richard

[Chris Hurd]

Originally Posted by Richard Hunter

The sensor block on the XL1S has a lower sensor count than that on a VX2000, and performs pixel shift to increase the resolution. Yet the VX2000 which is also 1/3" performs so much better in low light. If pixel shift really has the sensitivity advantages mentioned, this should not be the case.

Well, not only the VX2000, but that entire Sony family of camcorders: the VX2000, VX2100, PD150, PD170 and DSR250. Here's my take on what's going on. First, I've never been able to determine for certain whether or not these Sony camcorders are using a pixel offset technology. There's no mention of it in their marketing material, but that doesn't mean it might not be there. However, there's a reason why these particular cameras are so darn good in low light situations: they utilize a DSP (that is, an image processor) that's newer, better and more efficient than the one in the Canon XL1 / XL1S and most other contemporary video cameras. There's some significant amplification going in on that DSP that's clean and pumps out a very bright image. That's why those cameras are so phenomenal in low light.

I certainly didn't intend to imply that Pixel Shift was the *only* way to increase sensitivity... the DSP plays an enormous role as well, as the VX/PD series of Sony camcorders readily bear out.

Second, image sharpness. When the XL-2 was introduced, with its higher count sensor block, one of the big improvements over the XL1S was the sharpness of the image. It seems to me that while pixel shift can allow a lower resolution sensor block to create the same number of data samples as a native resolution sensor block, it does not necessarily produce an equally sharp image.

Sorry... not quite following you here. Both cameras, the XL2 and XL1 (also XL1S) are utilizing Pixel Shift technology. The XL1 / XL1S had to rely on double-axis Pixel Shift (that is, horizontal as well as vertical) to boost the resolution of its chips (250K effective pixels) to meet the spec for standard definition DV.

Meanwhile the XL2 uses H-axis Pixel Shift with 460.8K effective pixels in 16:9 mode and 345.6K effective pixels in 4:3 mode. Now without Pixel Shift, this would represent a 1:1 ratio of discrete sensor points for each pixel. But added to this is Pixel Shift in the horizontal axis, providing an increasing in resolution which is most helpful considering that the 16:9 DV mode on this camera (960 x 480) must be written to tape as 720 x 480.

Anyway, I think I lost your question. Basically the XL2 is considerably sharper than the XL1 / XL1S, because even though it uses Pixel Shift on only one axis, it still has substantially more pixels to start with than the older XL1 / XL1S.

I just find it hard to accept that everything about pixel shift is as marvelous as the manufacturers would like us to believe.

Pixel Shift is a perfectly legitimate, effective way to boost resolution on a three-chip camcorder. It is so widely and commonly used among all manufacturers that it's rare to find a three-chip camcorder not incorporating one form of pixel offset or another (as previously mentioned, the JVC Pro HD series seems to be the sole standout exception). With internet chit-chat full of measurebators (not you or me of course) that tend to fixate solely on the pixel count of an image sensor, the manufacturers pretty much have to make a point that they're using it. Double-axis for the old Canon XL and the recent Panasonic HVX and consumer JVC Everio; single-axis for the current Canon XL and XH as well as Sony 3-chip HDV camcorders, most likely single-axis for the older Sony VX & PD.

Again, it's in most every three-chip system ever made and it's there for a reason. Hope this helps,

[Richard Hunter]

Originally Posted by Chris Hurd

Anyway, I think I lost your question. Basically the XL2 is considerably sharper than the XL1 / XL1S, because even though it uses Pixel Shift on only one axis, it still has substantially more pixels to start with than the older XL1 / XL1S.

Hi Chris. I think that IS my point, that more pixels should give a sharper image than fewer pixels plus pixel shift.

And thanks for the info on the XL2, didn't know it was pixel shifting too.

Richard

[Matthew Nayman]

Richard...

Let's also not forget that the XL2 had a better processor (Digic II) and more refined sensor blocks than the XL1s

A pixel is not always a pixel.

[Ken Wozniak]

Originally Posted by Larry Secrest

At one point in the story I have a woman running on a background of trees, it's windy and the leaves are very agitated. The shot is :woman running+ panning+moving leaves on tree.
No way to film this. The HDV codec fails miserably. There is absolutely no way to film this in a decent way.
Larry.

I haven't noticed any breakdown, although my hi-speed footage has been limited to panning during a football game. Still, following the ball carrier with all the other players (and fans) moving in the background, the background does get very busy. I'll take a more critical look at it. Plus, I'm going to try a "torture test" with circumstances similar to what Larry was shooting.

I'll agree that the HDV codec is less than ideal. I'd love to have uncompressed HD, but my wallet is not that deep.

[Wayne Dupuis]

Thank you for the informative info here. I am curious why LCD/Plasma panels haven't been brought into the big picture. C'mon folks the output on these panels has to be mentioned. Compared with my Sony Wega, images at high def looks better on the LCD panel, the same footage panned will blur, fuzz, pixelate on plasma/lcd, but the CRT image looks great all the way across and into the bokeh of the moving image.
So I guess I am saying that the delivery method must be partly to blame, and not so much the codec if lit for the scene you wish to capture.
LCD must go... We've been fed a television that will not hold up; it's a throw away technology that has us buying into the next best thing. CRT worked well. I hope from what I see of new technology; combining HD and CRT, we get a standard output device that actually delivers what our camera's are able to produce; before we raise a bunch of shooters that are compromised by this inferior delivery system.
IMHO

[Denise Wall]

Not sure if this question even fits in here but from the standpoint of a still photographer, would not opening the iris as much as possible, thus decreasing resolution of the background while panning action, decrease the problem? If still too much in focus BG then maybe use more zoom to isolate the subject/blur background more.

I do action shooting outside with BG of trees and grass with lots of fast pans. I've been working on ways of getting around the limitations of the HDV codec in these action shots while still shooting the way I want to, i.e., following the subject(s) framed tightly versus letting them move through or around in a still as possible field. It would seem to me that the less resolution the BG is given, the less trouble it causes in such situations. Then again, you all are probably all doing that to start with. I just didn't see it mentioned in this thread.

[Bill Pryor]

Your point is very good and valid, and I think most of us try to do that most of the time. I haven't had any of that artifacting problem some talk about, and I've shot waves in the water, trees blowing in the breeze, kids riding around on bicycles, etc. The only things I've shot that have been projected on a big theater screen are mainly interviews, and they look great. I may check out some of this footage with more and faster motion on a big screen to see what it looks like under theatrical projection conditions.

[Matt OBrien]

I would love to see footage of these artifacts. I've never had any pixelation or artifact problems with my A1. Unless the scene is improperly lit, I don't have any complaints. I understand HDV is limited, but so is my budget and I accept these limits. With price limitations you must accept compromises. If you don't want to compromise with imagine quality, you better not have budget restrictions. If you don't have budget restrictions build a Spielberg Ranch...

Just my two cents.

[Daniel Browning]

Originally Posted by Matt OBrien

I would love to see footage of these artifacts.

To really illustrate the artifacts correctly, one must shoot the same footage with a comparison codec and Canon's compression simultaneously. Anyone with a AJA IO HD and a G1 could do it. Seeing the footage side-by-side removes as much subjectivity as possible.

Originally Posted by Matt OBrien

I've never had any pixelation or artifact problems with my A1.

That's all that matters. It's all a sliding scale of circumstance and subjective preference. For some people and source material, 1 Mbps quicktime HD h.264 is good enough. For most of the viewing public, 8 Mbps Satellite HD is fine. For you, 25 Mbps HDV is sufficient. Others require 144 Mbps HDCAM.

Originally Posted by Matt OBrien

I understand HDV is limited, but so is my budget and I accept these limits. With price limitations you must accept compromises.

I don't think anyone would argue with you there. I really dislike the many image quality compromises in my XH-A1, but it's the most I can afford, so I have to accept that. Until Scarlet is out, that is. :-)

[Jerome Marot]

Originally Posted by Matt OBrien

I would love to see footage of these artifacts.

It is fairly easy to film subjects which will show the artifacts. Take your camera and go outside by daylight. Film trees moving in the wind. Zoom in and out slowly.

When the leaves have a certain size (not too big and not too small), the picture breaks down and the leaves look unnatural.

You will need a full HD screen to see the HDV artifacts.

[Alex Plank]

Larry, I see you live in Bethesda. If you ever need someone with another A1 in the area, let me know and I may be able to help out.