Standard Definition MiniDV can capture at at *true* 16:9 Pixel Aspect Ratio?

[Jon McGuffin]

I'm asking a question here I'm almost certain I know the answer to but it's just one of those questiosn you have to ask.

By very definition, isn't it impossible to have a SD camera that records in the MiniDV format to record in a true 16:9 pixel aspect ratio? The resolution of MiniDV is fixed and wouldn't make that possible correct?

Jon

[Glenn Chan]

There's no such thing as a 16:9 pixel aspect ratio.

16:9 refers to the aspect ratio of the picture, not the aspect ratio of the pixel. With a particular # of pixels horizontally and vertically (e.g. 720x480), you can change the shape of each pixel / change the pixel aspect ratio to change the picture aspect ratio. You can adjust how wide each pixel is to change how wide the whole picture is.

You can record 16:9 images into 4:3 formats (formats that are normally 4:3) by squeezing/stretching the image to fit... that's called anamorphic.

[John Miller]

I have a number of Sony miniDV cams that can record true 16:9. In fact, that's all I ever use...

[Glenn Chan]

There's another sense of the word true in that some implementations don't give very high resolution in 16:9 mode (e.g. compared to its 4:3 mode, compared to cameras designed with 16:9 chips).

[Kevin Shaw]

Technically, the DV format records 720x480 pixels regardless of the desired image format. Hence widescreen images recorded in DV have 720x480 non-square pixels, with each pixel representing more picture area horizontally than vertically. To produce a proper widescreen image during playback that data has to be interpreted to cover about 853 x 480 pixels on a square-pixel monitor, such as a computer screen or many HDTVs. So you could say that the DV format doesn't offer 'true' widescreen video in the sense that each recorded pixel represents one displayed pixel of the final image. However, the same is true for standard-definition DVDs so it's not a big deal, and a good widescreen DV image can yield a decent widescreen SD DVD.

As Glenn noted there are also issues with how some DV cameras capture a widescreen image. Most throw away some data in order to generate a widescreen image instead of a 4:3 one, plus you'll lose some vertical field of view when running a 4:3 camera in widescreen mode.

[Boyd Ostroff]

Originally Posted by Kevin Shaw

Hence widescreen images recorded in DV have 720x480 non-square pixels, with each pixel representing more picture area horizontally than vertically. To produce a proper widescreen image during playback that data has to be interpreted to cover about 853 x 480 pixels on a square-pixel monitor, such as a computer screen

In fact, the DV format never uses square pixels. Shooting 4:3 DV with square pixels would require 640x480 and not 720x480. So no matter which format you shoot - 4:3 or 16:9 - the image needs to be either stretched or squashed horizontally to be in the correct proportions...

[Jon McGuffin]

Thank you all for the feedback..

I'm still a little confused...

#1) If the MiniDV format is in fact "locked" into exactly 720x480 pixes (square or non-square), how can this be considered 4:3. Why in the world would any standardizing organization intentionally create a "standard" resolution that would have to be modified to fit correct, at the time, on any regular old television?

#2) I've seen some camera manufactures report that they offer a MiniDV camera that offers "true" 16:9 recording mode with higher resolution than standard definition. How is this possible?!?
http://www.camcorderinfo.com/content...der-Review.htm <- see the Performance button, then under video resolution.

[Ron Evans]

The image sensor in a camera gathers light to represent the image through the lens. This is never what gets recorded directly to tape or other medium. The output from the sensor inputs to a DSP that uses this data to create the image data to be recorded( or output and this can be different). The imager can have more or less individual sensors( pixels) than represented in the data being recorded. The more individual sensors the easier it is for the DSP to create a higher resolution image( and cover for defective sensor pixels). However the smaller the individual sensors( pixels) the lower the light collected and thus low light performance suffers. Hence the goal of having just the right number of individual sensors to meet the resolution required with the low light performance. Not all the individual sensors collect the same colour of light either so the DSP is both creating a luminance image and colour image from the sensors. Even for 3CCD camera where one might think that each sensor feeds one colour that isn't quite true, the green is often offset 0.5 pixels so that the DSP can create higher luminance resolution using all three sensors!!!!
A lot of newer camcorders have a lot more individual sensors than they need for video so that they can take still photos too!!
So the sensors gather light( intensity and colour), the DSP processes this input and formats for tape/medium as requested SD, HD, 4x3, 16x9 or its output connectors. Now one can see how important the DSP is in a camera and how the same sensor can have apparently different performance in different cameras not just because of the lens but also the DSP electronics. Just like 16x9 TV's where the scalar is really what makes the difference in the picture.

Ron Evans

[Boyd Ostroff]

Jon:

#1 - with an analog TV set (which has been the norm until recently), I don't think there was any advantage to using square pixels, so the more you could cram on a single scan line the better. A bit OT, but I remember when Apple introduced the first Mac with its 9" black and white screen. One of their innovations was designing the screen with square pixels which allowed much faster integer calculations by QuickDraw.

#2 - Comparing two camera's I own... The Sony VX-2000 has a sensor with 720x480 pixels. When you switch to 16:9 mode the only way to capture video in the correct proportion is to use a letterboxed area of 720x360 pixels. When recording to tape, the camera stretches this to 720x480 to create a proper anamorphic 16:9 image.

Now my Sony PDX-10 has a sensor with 1152x864 pixels. When you shoot in 16:9 mode it letterboxes an 1152x576 area and downscales it to 720x480. So with either camera the result is a 720x480 final image (345,600 pixels) however it was created from a 259,200 pixel image on the VX-2000 and a 663,552 image on the PDX-10. As the expression goes, garbage in, garbage out :-)

[Chris Hurd]

Also see my article at http://www.dvinfo.net/canonxl2/articles/article06.php for an explanation of how some DV camcorders get a native 16:9 image without losing resolution.

[Glenn Chan]

Why in the world would any standardizing organization intentionally create a "standard" resolution that would have to be modified to fit correct, at the time, on any regular old television?

With a CRT, it's easy to do non-square pixels. The electron beam scans across the tube horizontally, and each sample changes the intensity of that beam. You could fit more samples in there if you wanted. Though in practice, your overall resolution is limited by NTSC encoding (i.e. when the signal is transmitted).

Nowadays non-square pixels don't make that much sense.