Meaning of: Shot in "True" 16:9 Widescreen?

[Dave Largent]

Does use of an anamorphic adapter with 4:3
native chips qualify for "true" status? Or is it
only "true" if it has been shot with native 16:9 chips?

[Simon Wyndham]

AFAIK it's either with an adaptor or a 16:9 CCD. The reason that both are genuine 16:9 is that both are capturing an actual 16:9 aspect ratio picture. Both methods horizontally squeeze the picture as the recording medium still only holds a 4:3 picture.

[Boyd Ostroff]

I think the real distinction is that the image is created in a way that uses all 480 vertical lines and is anamorphically squeezed in the horizontal direction. An anamorphic adaptor lens does exactly this, as would a camera with high enough resolution CCD's.

On the other hand, if the image is created by just cropping to 720x360 it would not be "true" 16:9, even if it has been stretched to make it anamorphic.

[Barry Green]

The historical distinction was that some 2/3" cameras have actual 16x9-shaped CCD's. So they capture the 16x9 image using the full surface of the CCD with no compromises whatsoever.

Anamorphic adapters have muddied that distinction -- a camera with an anamorphic adapter also uses the full surface of the chip, but the effect is accomplished by bending the light entering the lens, which may add its own unique image characteristics. As far as I'm concerned, this also counts as "true" 16x9 because it delivers full resolution.

Then there's the newest crop of megapixel CCD cameras that sample a smaller 16x9-shaped patch off the surface of the larger CCD -- examples include the PDX10, PV-DV953, and (apparently) the JVC HD1/HD10. Is this "true" 16x9? Well, why not? If a resolution chart shows that you're getting full resolution, without cropping and stretching and digitally manipulating the image, then sure, it should count as "native" or "true" 16x9. This method doesn't use the full surface area of the CCD, but it also doesn't involve digital stretching or manipulation, and it gives basically pixel-accurate imaging that shows true resolution on a res chart -- so it qualifies as "true" 16x9.

Finally, there's electronic crop 'n' stretch. This has also been called "fake" 16x9. This is where a 4x3 camera crops off a portion of the image and then electronically stretches the remainder to make 16x9 data. Examples include the Canon XL1/GL1/GL2, Sony PD170/PD150/VX2000/VX2100, DVX100A in "squeeze" mode, and -- well, most earlier camcorders that had a 16x9 mode. This method involves a significant loss in resolution. This is considered "fake" because the camera isn't recording a 16x9 image, it's stretching it digitally to accomplish the look. All the other methods involve recording a full-resolution 16x9 image with no digital manipulation applied.

[Boyd Ostroff]

<<<-- Originally posted by Barry Green : All the other methods involve recording a full-resolution 16x9 image with no digital manipulation applied. -->>>

Maybe I'm nitpicking, but this isn't really true. 16:9 on DV is actually a bit of a kludge since - regardless of the shape or resolution of the CCD's - the information has to be squeezed into the 720x480 frame used in regular 4:3 DV. This follows in the tradition of anamorphically squeezed widescreen movies where a special lens compressed the image onto standard 35mm film.

So unless you're using an anamorphic lens to provide optical manipulation of the image, your camera must digitally squeeze 854x480 pixels into a 720x480 frame.

[Barry Green]

Well, sort of, but not necessarily. That would only be true if the pixels were square.

NTSC pixels are not square, they're tall & skinny. Think about it: 720 x 480 is not a 4:3 ratio, it's actually a 3:2 ratio, yet the picture on the television is 4:3.

PAL pixels are not square either: 720 x 576 doesn't sort out to 4:3.

So 16:9 out of 720 x 480 is no big deal either: instead of the pixels being tall and skinny, they're short and wide. A pro camera like the DSR570WS, or the Panasonic SDX900, uses 16:9-shaped CCD's, yet still records at 720 x 480, and there's definitely no compromise going on there.

Furthermore, there's not necessarily a 1:1 relationship between CCD pixels and DV frame pixels (although it seems like pro cameras like to have a close to 1:1 ratio). The frame samples a patch off the CCD pixels, and it can be a widescreen-shaped frame or a 4:3-shaped frame. If there are enough CCD pixels, there should be enough data -- but, the Canon XL1 only has about 250,000 pixels on its CCD's, nowhere near enough to account for 1:1 on the 350,000-pixel DV frame, yet it still manages to make great-looking video.

[Alex Dunn]

EXCELLENT point Barry on the tall pixels. I was hoping someone would mention that.

I'd just like to add that if you have native 16:9 chips and a 16:9 Adapter lens, you can get a wicked-sharp 2.33:1 image. Whoa!

[Ram Nagarajan]

Alex: Hmm, 16:9 chip AND an anamorphic adapter? That's be too much of a good thing, wouldn't it? :-) As I understand it, the widescreen adapter would be used to squeeze the 16:9 image into a 4:3 area. How does the adapter work with a 16:9 chip without just using a smaller (4:3) pixel area on the chip?
Best,
Ram

[Boyd Ostroff]

It should work, and is not tied to a 4:3 chip. But your point is somewhat valid. Even with a camera that shoots native 16:9 the image ends up getting compressed to 720x480. This would also be the case with the adaptor, you'd still have a 720x480 image but the pixel aspect ratio would get more extreme.

Now you would need to consider your options for using this footage since monitors don't display at 2.35:1. You could letterbox it as 16:9 DV, which sort of defeats the whole idea, or upconvert to HD and letterbox it as 1280x720. But DV is limited to around 530 horizontal lines; you can stretch it as wide as you like but you won't get any more detail out of the image.

I've thought about doing this since a 37mm anamorphic lens for my PDX-10 would be relatively inexpensive. But I'm not sure what I would do with this sort of footage, so I think if I want a 2.35:1 aspect I might as well shoot 16:9 and crop.

[Alex Dunn]

All true, but in reference to the goal being a wide(2.33:1)screen without losing resolution, the best option is to optically anamorphically stretch the image across a natice 16:9 chip. Then, when you anamorphically smoosh it down to 2.33 on a widescreen HD monitor or projector, the image quality is significantly better than cropping (even for DV).

It's not unlike the anamorphic enhanced DVD's. The lines of resolution possible on a DVD is a given, so you look for other variables. Have you seen a SD that has 16:9 mode (Sony) or an HD that will compress a anamorphic enhanced DVD internally? It almost double the resolution by narrowing the gaps between.

But like you guys said, it all depends on your output plans.

[Ram Nagarajan]

Point taken: I wasn't thinking in terms of the final output being seen on a 16:9 HD monitor at all. Sounds interesting: Guess I'll just go cast around for a good HD monitor and check it out!

[Bill Pryor]

I guess my understanding of the term is a little different. If it's "true" 16:9, to me, that means it's anamorphic--if you play it on a 16:9 monitor it looks good, but if you play it on a 4:3 monitor, everybody will be squished tall and thin.
I'm on the board of a regional festival, and though most films we get are 35mm, we take video too, and lots of the documentaries are video. Almost all the tapes we get will be labeled 16:9 when in reality they are 4:3 that have been cropped to 16:9. So If we set the video projectors to 16:9, the image will be distorted. That means I have to check every tape and re-label each one correctly. Occasionally there will be a "true" 16:9 one.

[Ram Nagarajan]

Yup, that's the way I've got 16:9 taped in my head too. Plays back correctly on a 16:9 monitor, looks squished ona regular 4:3 screen.

[Boyd Ostroff]

Fine, but you have to separate the issue of quality from that of format. You can take any 4:3 video, crop it, then squash it anamorphically. It will then play back correctly on a widescreen TV and look squished on a 4:3 screen. But the quality will be noticeably worse than one shot with a camera that uses all 480 vertical lines.

Think of it this way: you can print a 72 dpi photo at 8"x10", and you can also print a 300 dpi photo at 8"x10". Which one will look better?...

[Tommy Haupfear]

A good read on 16:9 and DV.

http://www.maxent.org/video/16x9.html