View Full Version : PDX10's 16:9 implementation - need clarification


Graham Hickling
June 12th, 2004, 11:00 PM
Could someone please explain to me the difference, if any, between 16:9 mode as implemented by the PDX10 versus the much cheaper model like the TRV70 and 80?

I have the latter, which has enough pixels for a 2MP still image, and it definitely uses some of those extra pixels to implement 16:9 (the filed of view gets wider, etc).

Yet only the PDX10 gets marketed as having a "true" 16:9 chip. So I'm confused....

Boyd Ostroff
June 13th, 2004, 07:39 AM
I have read that these cameras offer high res 16:9 also, but don't really know anything about them. I think they use electronic image stabilization (EIS) instead of the PDX-10's optical stabilizer. If so then you probably can't use EIS and 16:9 simultaneously, right? And of course these are only single chip cameras so the quality is not going to be as good as a 3-chipper. Also remember that the PDX-10 is marketed as part of Sony's professional line of camcorders (http://bssc.sel.sony.com/Professional/webapp/SubCategory?m=0&p=2&sp=11&sm=0&s=&cpos=) which is separate from the consumer division.

But if you want a real answer to your question, it's time for some due diligence. Shoot the EIA 1956 test pattern (http://www.bealecorner.com/trv900/respat/#EIA1956) in both 4:3 and 16:9 mode and compare the vertical resolution.

Andre De Clercq
June 13th, 2004, 01:01 PM
Even with enough pixels available it's a matter of CCD chip readout strategies. The PDX10 has two different readout systems. One for 4:3 and one for 16:9. No vertical remapping of the readout is needed. the number of active lines remains 480. Other cams like pd150, DVX100A... keep the 4:3 readout in 16:9 mode, clip-off top and bottom and perform an upres (vertical remapping) starting from the reduced numbers of lines (360) to generate the 480 active pixels needed for the 16:9 anamorphic pictures. Some others just generate a letterbox output (no vert. upres)when set in 16:9 mode like the DVX100. The upres is the tricky step and needs a difficult trade-off between vert resolution loss(vertical filtering) and aliasing.

Graham Hickling
June 13th, 2004, 04:47 PM
Thanks for your replies. I've just had a look at this pdf...

http://bssc.sel.sony.com/Professional/docs/brochures/dsrpdx10final.pdf

...and from the diagram on the second column on page two, it seems to me that the PDX10 and the TRV70 (which is what I have) use essentially the same strategy to switch from 4:3 to 16:9. That is, they both maintain 480 lines of vertical resolution, and use extra pixels to the left and right of the 4:3 pixel area to achieve the wider aspect ratio. That's what the TRV70 manual says it does - it doesnt letterbox and it's angle of view widens when 16:9 is selected.

The PDX10 will cost more given that it's a 3-chipper with DVCAM, has OIS, etc. (Maybe its the only prosumer 3-chipper that doesn't letterbox and up-res to get 16:9, hence the marketing focus on that feature???).

By the way, Boyd, the TRV70 does have EIS in 16:9 mode. It has a 2MP sensor (cf. the PDX10 which has 1.1MP) so I guess it has enough extra surplus pixels to keep EIS working even in wide mode.

Tommy Haupfear
June 13th, 2004, 06:54 PM
By the way, Boyd, the TRV70 does have EIS in 16:9 mode.

Graham, the TRV70 employs electronics sensors (not CCD pixels) for its image stabilization which is why it can be active in its 16:9 mode.

One note about the TRV70 is that while it does have a high quality 16:9 mode that widens the angle of view the widening is lessensed as you zoom and at full zoom its the same width as using 4:3. Try it out. Place your cam at full wide and toggle 4:3 and 16:9 via the menu to see the changes in width. Now zoom 100% and try the same toggle. Not sure why this happens but effects most of the HQ 16:9 Sony single chippers.

Tom Hardwick
June 22nd, 2004, 10:45 AM
That's an interesting observation Tommy. It suggests that if you have a 10x zoom in the 4:3 mode you then have a 11 or 12x zoom in the 16:9 mode. This cannot be, but it's interesting nontheless.

tom.

Boyd Ostroff
June 22nd, 2004, 04:21 PM
<<<-- Originally posted by Tommy Haupfear : Graham, the TRV70 employs electronics sensors (not CCD pixels) for its image stabilization -->>>

I don't understand how it can do this unless it uses optical image stabilization. With EIS the sensors would tell the camera how much movement there is, but in order to compensate for that movement you will need to utilize some portion of the image outside the "normal" field of view, and that could only come from "extra" CCD pixels. But maybe the CCD has so many extra pixels that it can still handle 16:9 and EIS?

Tommy Haupfear
June 22nd, 2004, 07:52 PM
Boyd, I've never been 100% sure about the motion sensors but after doing a Google search I came up with the link below that pretty much says that Sony's Steadyshot uses both motion sensors and excess CCD pixels. The link looks pretty dated and I'm not sure what the now "Super" in new Super Steadyshot adds to the equation..

Click here (http://www.digitalvideoforless.com/steadyshot.htm)

Tommy Haupfear
June 22nd, 2004, 08:01 PM
Tom, now that I've done a little digging on the Super Steadyshot, I wonder if it plays a part in the odd behavior of a TRV70 (and others) when in 16:9 mode. It would seem that a TRV70 uses both horizontal and vertical motions sensors along with excess CCD pixels for its Super Steadyshot. At full wide I would think that the image stabilization would not be nearly as active as when you zoom to 10x optical therefore giving you the full 16:9 width. Then as you zoom closer to 10x the system becomes more agressive with its pixel use and 16:9 is forced to give up some real estate.

Just an idle thought and definitely not to be taken too seriously! :)

Boyd Ostroff
June 22nd, 2004, 08:03 PM
I suspect some may disagree with the statement:Only 270K pixels are required to deliver excellent picture qualityI believe "Super Steadyshot" is what Sony calls its optical image stabilizer (using the motion sensors and moving optics). Anyway, the distinction that article makes is that of actually measuring camera movement as opposed to estimating it based on frame comparisons. So perhaps the TRV-70 is only using 270,000 pixels to produce its "excellent picture quality" in 16:9 mode? ;-)

Tommy Haupfear
June 22nd, 2004, 08:10 PM
I suspect some may disagree with the statement:

Agreed, I believe you need at least 340k pixels. That explains the Super in the new Super Steady Shot. With single CCD Sonys starting out at 680k and going up they probably have overcame the 270k limitations of the original Steady Shot.

Sony refers to their OIS as Optical Super Steady Shot.

Boyd Ostroff
June 22nd, 2004, 08:35 PM
Interesting. On my VX-2000 it says:Super SteadyShot
Digital HandycamI didn't realize they also applied that term to EIS.

Graham Hickling
June 22nd, 2004, 11:03 PM
Interesting!

My take on this is ..... the highend Sony's use opto-mechanical systems that Sony sometimes calls "optical super steady shot". Sony cameras like my TRV70 have "super steady shot" whereby sensors detect camera movement and adjust the pixel-reading accordingly. Some other manufacturers detect movement of the image on the chip and process the image accordingly (which Sony claims results in poorer quality than their method).

I guess most people in casual conversation refer to the first as OIS and both the second and third as EIS.

Tommy Haupfear
June 22nd, 2004, 11:14 PM
I guess most people in casual conversation refer to the first as OIS and both the second and third as EIS.

I've also seen DIS (digital image stabilization) refer to non-Sony type of CCD based stabilization that lacks the motion sensors. Then you have Panasonic with their occassional DEIS..

So much for standardization!

Michael Bott
June 26th, 2004, 03:46 AM
For the clearest explanation of why the PDX10's 16:9 mode is superior to that offered by its less expensive brothers see the following page on Creative Video's site.

http://www.creativevideo.co.uk/pages/cvp_dsr-pdx10p-16x9.htm

Tommy Haupfear
June 26th, 2004, 06:26 AM
For the clearest explanation of why the PDX10's 16:9 mode is superior to that offered by its less expensive brothers

Nice link, thanks Michael.

and here is another illustration..

Click here (http://img.villagephotos.com/p/2003-5/74415/PDX10.jpg)


- Idle thought - does the PC330 with its 3MP CCD do 16:9 better than the PDX10?

Graham Hickling
June 27th, 2004, 03:34 PM
That explanation is true for a TRV950/PDX10 comparison, but not for the TRV70 and TRV80, at least.

Both gain their 16:9 ratio in the same way as the PDX10 (i.e. expanding the sensor area sideways rather than cropping vertically). But they only single-chippers, of course.

Tom Hardwick
June 28th, 2004, 12:42 AM
Tommy's link is actually slightly more truthful than Michael's as the admission is given that in the 16:9 mode vertical resolution is indeed sacrificed when compared to the 4:3 rectangle.

My own filming tests show that the 16:9 uses 10 per cent fewer vertical CCD pixels than 4:3 mode but 20 per cent more horizontal pixels (see the diagram on Tommy's link). In a perfect world the vertical resolution would remain unaffected and the horizontal pixel count would increase by 25 per cent, but this PDX10 solution is certainly a step in the right direction.

tom.

Tommy Haupfear
June 28th, 2004, 08:08 AM
Good news for the upcoming Panasonic NV-GS400 as its claiming that no vertical zooming is necessary for its 16:9. Its also worth pointing out that the GS400 has almost an indentical CCD/pixel arrangement as the PDX10. 1/4.7" 3CCD with 1070k each

Boyd Ostroff
June 28th, 2004, 09:04 AM
<<<-- Originally posted by Tom Hardwick : Tommy's link is actually slightly more truthful -->>>

FWIW, "Tommy's link" is the same illustration found on page two of Sony's PDX-10 brochure.

Ignacio Rodriguez
June 29th, 2004, 11:21 AM
> Its also worth pointing out that the GS400 has almost
>an indentical CCD/pixel arrangement as the PDX10. 1/4.7"
> 3CCD with 1070k each

Hmmm. According to the Sony consumer web site, the TRV950 (same optics and sensors as the PDX10) actually has less pixels per chip.

"1/4.7" Advanced HAD™ 3 CCD imagers with 690k gross pixels each" and "1.0 megapixel gross" referring to the whole array, not to each CCD :-(

http://www.sonystyle.com/is-bin/INTERSHOP.enfinity/eCS/Store/en/-/USD/SY_DisplayProductInformation-Start;sid=qD3Zr6VkoL7ZnOWFCzDTpOp0N_9ldhiZtGI=?ProductSKU=DCRTRV950&Dept=dcc_DICamcorders&CategoryName=dcc_DICamcorders_MiniDVHandycamCamcorders

Confusingly, the Sony pro camcorder website lists the PDX10 as having "High Picture quality with three 1/4.7 inch CCD sensors with 1,070,000 pixels each"... then again it also says the camera has "reduced vertical smear" and 15 fps proscan which we all know it does not have, at least not in video mode:

http://bssc.sel.sony.com/Professional/webapp/ModelInfo?m=0&sm=0&p=2&sp=11&id=65263

Tommy Haupfear
June 29th, 2004, 11:45 AM
Hmmm. According to the Sony consumer web site, the TRV950 (same optics and sensors as the PDX10) actually has less pixels per chip.

Click on the specs of TRV950 and you'll see that it shares the same CCD/Pixel arrangement of the PDX10.

1.4.7" 3CCD with 1070k pixels per each CCD.

It is confusing how they state the 690k video gross pixels though.

Boyd Ostroff
June 29th, 2004, 12:40 PM
I believe if you read the fine print somewhere it states that figure is for 4:3 mode....