HC1/A1 HD video frame resolution?
 

According to most reports the A1 and HC1 capture video using a 1440x1080 window. Unfortunately, the only test results for the HC1 measures its vertical resolution at only 480-lines. This number is way too low for a 1080-row window.

I have a computer model that let's me predict (with near perfect accuracy) the resolution that will be measured given a CCD's resolution. It can also work backwards from test data. It shows that the HC1 is using a 1440x810 (16:9) window.

If this number is correct, then the EIS area is 480 horizontally and 630 vertically. For a total of 302,400 pixels within a 1920x1440 CCD that is 4:3. (Actually, perhaps, only 480 of the 630 is for EIS -- the other is simply centering of the 16:9 window on the 4:3 CMOS.)

Now comes the fun. A 1440x810 window has 1,166,400 pixels. Let's assume the new HC3 has the same HD video resolution window. Let's also assume that Sony has decided to use the same size buffer area which is 302,400 pixels.

The sum of these two values is 1,468,800 pixels. And, camcorderinfo.com says the HC3 CMOS chip has 1,434,000 effective pixels.

Which tells me three things:

1) It seems very much to confirm my model's estimate that the HC1 is working with a 1440x810 window.

2) The HC1's full 1920x1440 and 1440x1080 CMOS is only for 4:3 and 16:9 stills.

3) The HC3 is only giving-up still pix resolution, not HD video rez. The gain is greater sensitivity.

Does all this make sense to HC1 and A1 owners?

mmmmmmmmmmmmmmm. repetition.

I hope its your keyboard that's broken and not your mind that's breaking down trying to answer my question.

Sounds feasible. I wonder if it has anything to do with making an allowance top and bottom for steadyshot corrections?

It's completely wrong. I have HC1 and doing shot , using whole CMOS sensor in picture mode (1920x1440) and then going to video mode and grabing frame from video I'm getting around 1800x1000 window visible from (1920x1440). Of course it's squeezed anamorphicaly to 1440x1080.

But it will be using interpoletion to get more pixels. Sony says it can get 2 times more. So if Sony has 4 Mpx from 2Mpx clear-vid CMOS, then in video it will get 2,8 MPx from effective 1,4 Mpx.

But HC1 is 1 sensor unit (interpolation occurs + interlacing). 3 sensor cameras also are getting only 700-lines from 1080 (only interlacing occurs).

1) With a STILL you can bring it into Photoshop and see its resolution (1920x1440). But how do you KNOW -- with video -- you have "around 1800x1000 window?" All you know is what the video size is after it has been recorded -- and we know that must be 1440x1080.

2) How do you KNOW its from 1920x1440?

3) If it were "squeezed anamorphicaly" -- ONLY the horizontal number would get smaller.

I'm not clear what you are saying.

Now about the HC3 -- the idea that you can get more information than you have pixels is a marketing concept. A computer can guess what might lie between pixels, but that's all it is -- a guess. So CLEARVID, or whatever Sony may call, it is not what I'm interested in.

Moreover, I don't think Sony has claimed CLEARVID is used on the HC3.

Actually, I don't think I've seen a resolution chart from the HC1/A1U. Anyone got links? I'd be interested.

-Steve

There's a partial one here - the link is in my post headed:
HC1 'Cinema mode' resolution test
http://www.dvinfo.net/conf/showthrea...t=56590&page=2

It is just of near-center resolution, with and without the HC1's 'Cinema mode' activated.

I've thought it would be straightforward. Focus on CMOS sensor. It has 1920x1440 and have 4:3 aspect ratio. HDV is 16:9 so it uses some portion of the sensor. In ideal situation, with optical image stabilization video frames would be created from 1920x1080 and squeezed to 1440x1080.
What I've done is take shot in still mode (1920x1440). Then go to video mode and capture video exactly pointing camcorder at the same place (or more preciously camcorder has not been moved even by 1 mm when changing mode from still to video ;) ). On computer I've taken frame grab from video (1440x1080), unsqueze it to 1920x1080. Now the most important part - compare view area seen in still mode and in video mode. In video mode I see the area of about 1800x1000 from the still mode.

It has not been claimed, but it's really probable.
CLEARVID uses quite similar technic to get more information than the number of pixels as 3CCD shifting (e.g. Panasonic consumer line). It's not quite marketing concept. It somehow works. Image at 4Mpx from 2 Mpx will look really bad (as bad as GS400 stills at 4Mpx ;) ), but it will be better than simple interpolation from 2 Mpx to 4Mpx without CLEARVID.

THANK YOU!

Looks like camcorderinfo.com had the camera in Cinema Mode when thy ran there tests as their reported 480-lines of V rez is exactly what one gets in Cinema Mode. The 1080i V rez is about 700-lines, very little less than the FX1/Z1. H. rez is about 700-lines which is much better than the FX1/Z1's 550-lines.

Hey Steve, Just out of interest, what's the correct thing to do when shooting a resolution chart with a camera that's doing the non-square pixel 'thing'?

Should I rescale the bitmap image back to 16:9, so that the chart proportions are correct? Or leave it as is?

I wondered too -- then decided any rescale might alter the data and just looked at the centered wedges.