XL2 resolution
 

all the specs say XL2's CCD produces a 960x480 image (in 16:9 mode for NTSC or 960x576 for PAL). Eventually the image is recorded in DV format which allows only 720x480 (or 720x576 for PAL) for 16:9 or 4:3 aspect ratios. Now, my question is this: does it scale it down before it records the image on tape? is there a way to preserve it at its original size of 960x480?

also, where do you choose between 2:3 and 2:3:3:2 modes?

thank you

16:9 is resized to 720x480 as it goes to tape (it has to be). Most non-linear editing programs will recognize the 16:9 flag and convert it back to 960x480 (my four-year-old Canopus DV Rex RT does this).

Go to the Camera Setup menu, choose 24p Mode Sel., then choose 2:3 or 2:3:3:2. Hope this helps,

When downsampled for recording as a DV signal the video must first be low pass filtered. Were this step skipped Nuyquist criterion would be violated and Moire (aliasing) would be created. This low pass filtering throws away the highest frequency information (fine detail) captured by 960 wide CCD. This information in not recoverable so when your NLE or Quicktime or video card upsamples back to 960 x 480 the resolution of the picture is no longer 960 x 480 pixels but rather 720 x 480 displayed in however many pixels you use for the display (1920 x 1080 in the case of my monitor). There are tricks that can give the appearance of increased sharpness (such as USM) but it is important to understand that basic picture information has been irrecoverably lost when down sampling. OTOH, if it looks good, who cares?

To answer the question specifically: Yes, it is possible to preserve the image at its original 940x480 size (or any other size you like in that aspect ratio). No, it is not possible to preserve it at its original resolution of 940x480.

Now this all begs the question "If they can only put out 720 pixels per line why do they oversample at 960?". Without sitting down for a long chat with the engineers that designed this thing I can't begin to answer that question. One possibility is that it is a means of dealing with Moire. A sensor which oversamples followed by a digital low pass filter in the electronics gives similar results as a sensor with the desired end resolution covered with an optical low pass (blur) filter.

All NTSC DV is 720 x 480. There is no way to capture 960x480 in DV, it must be downsampled to 720x480.

The 720x480 is treated as having a pixel aspect ratio of 1.2, meaning that the pixels are interpreted as being 20% wider than they are tall.

Your editing program doesn't work in 960x480 mode, it works in 720x480 with a 1.2 aspect ratio. All DV is 720x480, the thing that changes (between 4:3 and 16:9) is the aspect ratio of the pixels, not the number of pixels.

If you're making graphics in a photo program for use in your editing program, and your graphics program only supports "square" pixels (i.e., it doesn't know about "pixel aspect ratios") then you'd want to make your graphic at 853x480, with the understanding that your editing program is going to downsample it to 720x480.