minutes on tape for 16:9?

[Ron Johnson]

If the # of pixels is higher by 2x in 16:9 mode, how is the extra data stored? Is there more compression used for this mode, or does the tape transport run faster?

[Boyd Ostroff]

All DV is the same, 720x480 (NTSC). The data written to tape is the same whether it's the XL-2 in 16:9 or a cheap one chip camcorder shooting 4:3. The higher resolution chips provide data that's used internally by the camera. The resulting image is "squeezed" horizontally to a 720 pixel width, and widescreen monitors know they have to stretch it back into the proper proportion.

So in other words, the extra data is not stored. It's thrown away during the downsampling process, however it assures that a higher quality image results because more pixels were sampled. The compression and tape speed are the same. It's still just DV...

[Scott Moody]

<<<-- Originally posted by Boyd Ostroff : All DV is the same, The resulting image is "squeezed" horizontally to a 720 pixel width, and widescreen monitors know they have to stretch it back into the proper proportion.-->>>

i haven't done the math, but does that mean one can get a higher-res 16:9 image using an anamorphic lens adaptor vs. the XL2's built-in 16x9 support since, with an anamorphic lens, one is using most (all?) of the 720x480 space vs. the XL2 using 720 by (less than 480) letterboxed space? I'm new to this. sorry if this is a silly question...


Thanks in advance,

Scott

[Boyd Ostroff]

No anamorphic adaptor is needed. The new design captures the full resolution of the 16:9 image and processes it in the camera to make it anamorphic. It isn't letterboxed except for the purpose of display in the viewfinder. There are a few other cameras which also do this, like the Sony PDX-10 for example.

[Ron Johnson]

From the info published so far, the 16:9 mode on the XL2 is not the usual squashed version as on XL1. The 4:3 mode covers a 720x480 pixel array while the 16:9 mode covers 962x480. So, natively, the XL2 is capturing more data (962/720 = 1.34x).

My original question was meant to gain understanding of how the extra 34% of data is managed - does it just get a higher level of compression, or does the tape transport run faster, thereby reducing recording time (on a given tape) compared to 4:3?

Thanks

[Barry Goyette]

Ron,

Boyd's first post was correct. The DV format only supports 720 x 480, so what happens is the data from the 960 pixels get squashed, squished or what ever you prefer to call it, down to 720, and then is compressed just like regular dv...at 5 to 1.

This method is preferable to the crop and stretch method used in most other cameras (including the xl1s), where the image is cropped to 16:9 and then the vertical resolution is stretched to fill the 4:3 format, as the capture resolution is significantly higher. It is also probably better than using an anamorphic adapter, as the distortions and aberrations added by the extra glass degrade the image quality somewhat. Take a look at simons clips here...all shot in 16:9.

http://www.xl1s.com/photos4.php


Barry

[Boyd Ostroff]

<<<-- Originally posted by Ron Johnson : From the info published so far, the 16:9 mode on the XL2 is not the usual squashed version as on XL1. -->>>

Don't confuse the format with the method when it comes to 16:9. The XL-2 does indeed "squash" the image just like the XL-1, GL-2, PD-150, etc. If you want to shoot 16:9 and use DV then the image is squashed, period. As Barry points out, the distinction that sets the XL-2 apart from the XL-1 is that this "squashed" image has been constructed from the full 960x480 pixels instead of only 720x360.

Here's an experiment you could try to help understand what happens. Start out with an image that is 960x480 pixels; that's what the XL-2 CCD's capture. Now use Photoshop to resize the image without constraining proportions to 720x480. This is what the XL-2 writes to tape. Now resize that image back to 960x480; that's what a widescreen TV does while displaying the recorded image (this is a bit of an oversimplification because it doesn't account for the non-square pixels, but the principle is still valid). Now in the process of going from 960x480 > 720x480 all of that "extra data" you keep talking about is just discarded. And when you stretch the image back from 720x480 > 960x480 it will be in the same proportions as the original image from the CCD's, but with less resolution because of this process.

But on camcorders that don't have high resolution CCD's, like the XL-1s, the same experiment would go like this: start out with an image that's only 720x360; this is the raw data from the CCD's. Now use Photoshop to resize without constraint to 720x480; that's what is written to tape. Finally perform the last step the same as above, and resize the image to 960x480. The quality will be noticeably worse because you started out with a smaller image (720x360).

I'm sorry if you find this hard to accept, but the fact is that none of that additional data from those extra pixels is ever compressed or written to tape. The distinction is that the 720x480 pixels which ARE written to tape contain more accurate data to represent the image because they were downsampled from 960x480.

[Ron Johnson]

Boyd - Thanks for filling in the details; what you have written makes sense.

[Rob Lohman]

Excellent explanation from Boyd. What you gain is higher sampling
and increased vertical resolution. You go from 360 lines to 480
and to a higher sampling horizontally.

[Barry Green]

Originally posted by Scott Moody :
i haven't done the math, but does that mean one can get a higher-res 16:9 image using an anamorphic lens adaptor vs. the XL2's built-in 16x9 support since, with an anamorphic lens, one is using most (all?) of the 720x480 space vs. the XL2 using 720 by (less than 480) letterboxed space?

On other cameras this would be a viable option. On the XL2, it's not such a good idea.

On other cameras the 16:9 area is extracted from a smaller subsection of the CCD (i.e., the DVX100 uses a 720x360 area of the 720x480 chip). So on cameras like that, the anamorphic adapter is a good idea, because it lets you use the full 720x480 surface of the chip.

On the XL2, it doesn't work that way. The largest area of the chip that can be used is the 16:9 patch. The 4:3 is actually a smaller subsection of the 16:9 area. So using an anamorphic adapter on the XL2 would significantly LOWER the amount of chip being used.

[Marty Hudzik]

What if you used an anamorphic adapter in addition to the 16x9 mode? Would you effectively get 2.35:1 aspect ratio as with the DVX100 when using anamorphic lens and squeeze mode?

just asking.........

[Chris Hurd]

Yes, you'd get a "super widescreen" CinemaScope-type of image. It's still written to tape as 720 x 480, but it would be as high-quality as you can get with standard-definition DV.