Going Anamorphic? at DVinfo.net

Bennis Hahn · August 9th, 2005, 08:19 AM

I have been thinking for a while that I would do me good to pick up the 37mm anamorphic adapter from Century Optics. The only hesitation is that I have never seen footage from it. I am wondering, how much of a difference will I see in picture quality? Right now I just use the slim mode on my camcorder to make 16:9 DVD's and it looks OK on my 57" TV, but not great. How much of a difference will those extra pixels make? I am filming a longer short this fall and I plan on outputting to DVD, with the possibility of projecting in (digitally) to a decent size. Therefore, I am thinking all the pixels I can get will be beneficial.

Jason J. Gullickson · August 9th, 2005, 11:08 AM

This site goes into all sorts of detail about what happens when using "stretch" mode vs. optical widescreen adapters. You may have seen it already but I thought it was worth mentioning here:

http://members.macconnect.com/users/b/ben/widescreen/

I didn't see any still example shots but I'm sure you could email the guy for more info.

Boyd Ostroff · August 9th, 2005, 01:57 PM

I haven't used the 37mm anamorphic - and the topic has come up here before, but nobody had personal experience with that particular model century lens. However, the following tests might give you an idea of native 16:9 compared to cropped/stretched 16:9. See the comparison images from the PDX-10 which shoots real 16:9 with megapixel CCD's vs the VX-2000 which just crops the 4:3 image and stretches it back to make it anamorphic:

http://www.greenmist.com/dv/16x9/

Bennis Hahn · August 9th, 2005, 02:42 PM

Those two links were great, thank you.

Now another question.

If I shoot with the anamorphic adapter and slim mode (for 2.35:1) how to I "un animorph" the footage in Vegas? I know regularly one would render out to 16:9 but then the footage is still squeezed vertically.

Marco Leavitt · August 9th, 2005, 08:33 PM

I haven't used that version either, but I have used the bayonet mount version for the GL1. Compared to digital 16:9, the difference is dramatic.

Boyd Ostroff · August 10th, 2005, 04:19 AM

I can't tell you specifically how to unsqueeze this in Vegas since I use FCP, but there are a few principles to consider. Are you planning to work in standard definition DV? If so then there's probably very little advantage to the "double anamorphic" technique you describe, and it may even degrade the image more.

There should be some way to verically squash the stretched image to bring it into the correct proportions, but you will still by limited by DV's 720 pixel width. If you wanted to up-rez it somehow to HD you could stretch it horizontally to 1280 or 1920 pixels but I think it would look very soft at that resolution.

If you really want the 2.35:1 aspect ratio I think your life will be much easier to shoot in 16:9 with the anamorphic lens, then just crop to 2.35:1 in post by putting a black bar above and below the active image area. If your goal is to put this footage on a DVD then that's basically what you're going to end up with regardless of how you accomplish it - just look at any commercial DVD...

Bill Porter · August 14th, 2005, 12:13 AM

Quote:

Originally Posted by Boyd Ostroff

There should be some way to verically squash the stretched image to bring it into the correct proportions, but you will still by limited by DV's 720 pixel width. If you wanted to up-rez it somehow to HD you could stretch it horizontally to 1280 or 1920 pixels but I think it would look very soft at that resolution.

1) Is the 720 width limit found on the data put onto the tape, or is it found on NTSC screens, or in non-HD editing programs, or the raw footage itself no matter where it's located, or all of the above?

2) What would be the differences between uprezzed 1280 or 1920 pixels wide footage, and double-anamorphic (adapter plus in-camera anamorph) footage? Field of vision? Quality?

Thanks

Boyd Ostroff · August 14th, 2005, 04:23 AM

NTSC DV is 720x480, period. If you're working with DV then that's what you've got. Anamorphic 16:9 stretches the pixels wider on playback, but there are still only 720 of them on each line. If you wanted to stretch them wider to acheive 2.35:1 you're still only working with 720 pixels.

1280x720 and 1920x1080 are high definition standards. You could blow up DV to these sizes, but garbage in = garbage out :-)

What kind of camera are you using? Anamorphic adaptors have limitations of focus and zoom range, and since you're adding additional glass in front of the existing lens there will be some image degradation. If you really want to get higher quality for your widescreen TV you might want to upgrade to a camera with higher res CCD's that can do real 16:9. Canon, Panasonic and Sony have inexpensive models that can do this now. Then there are medium priced models like the Sony HC-1000 and Panasonic GS-400. The Sony PDX-10 does nice 16:9 although it may be discontinued.

Beyond that there are the new Sony HDR-HC1 and HVR-A1 HDV cameras and then the HDR-Fx1 and HVR-Z1, but these are more expensive. They do shoot in the 1080i format however.

Bennis Hahn · August 14th, 2005, 08:02 AM

Right now I have a GS200. Laugh if you want, but really the camera produces beautiful pictures when there is enough light. Since I am shooting a movie, the lighting will be controlled and that will not be a problem. I really would like a new camera, but I simply do not have the money for it. Right now what little money I have is going toward getting good audio.

Mathieu Ghekiere · August 14th, 2005, 08:10 AM

Quote:

Originally Posted by Bennis Hahn

Right now I have a GS200. Laugh if you want, but really the camera produces beautiful pictures when there is enough light. Since I am shooting a movie, the lighting will be controlled and that will not be a problem. I really would like a new camera, but I simply do not have the money for it. Right now what little money I have is going toward getting good audio.

Don't be ashamed of that, that's a great decision!
The ears are much less forgiving then the eye!

Bill Porter · August 14th, 2005, 09:30 AM

Thanks, Boyd. I hadn't "gotten it" that all we are doing is stretching pixels.

What, then, is the difference between squeeze mode and shooting in 16:9, in real resolution? If NTSC DV is 720 pixels no matter what, how is squeeze mode not every bit as good as an anamorphic adaptor? The way I understand it (which I know is incorrect, hehe), in both instances an image is getting projected onto the same number of pixels across the camera's CCD's.

BTW to answer your question, I'm not trying to shoot 2.35:1 widescreen, actually. Just trying to learn so I don't feel like such a dipsh*t.

You're right that Panasonic offers some inexpensive models that shoot in true 16:9. Trouble is, the DVX doesn't (I now have one), and the inexpensive models lack the good stuff the DVX has.

Bennis, the GS200's do shoot nice stuff with enough light. I had a GS120, its forerunner, and was impressed as heck. Current 3-CCD consumer cams are very powerful, especially considering how small they are! It is only a matter of time before cameras that size - and smaller - are shooting 24p, onto large-capacity, affordable media cards.

Boyd Ostroff · August 14th, 2005, 12:54 PM

OK... in talking about 2.35 the issue stretching the width to something greater because I assumed you were already getting the maximum vertical resolution.

But putting the anamorphic adaptor on a 4:3 camera is a little different matter. In this case, you're increasing your vertical resolution. If you shoot in "squeeze" mode there are a couple things happening. First, the camera crops its native 4:3 image to 720x360 in order to acheive the 16:9 proportion. In doing so you're throwing away 120 pixels in the area above and below the image, and 120/480 = 0.25, so you have just discarded 25% of your vertical resolution. The "squeeze" is really more like a "stretch" actually. The resulting 720x360 image is stretched vertically back to 720x480 which makes it a proper anamorphic image.

There is no difference in the format of this image than what you get from a "real" 16:9 camera; what's different is the way it was created. On a camera with higher resolution CCD's there would be a minimum of 960x480 pixels in the initial 16:9 image. Then that would be horizontally squashed to 720x480, but you would have all 480 vertical lines so there's no resolution loss. Chris has a nice graphic which shows how this works on the XL2:

http://www.dvinfo.net/canonxl2/articles/article06.php

I did something similar showing the CCD setup on the PDX10"

http://www.greenmist.com/dv/16x9/10.JPG

But the DVX-100a is a slightly different beast because of its progressive modes. In progressive mode you can actually use all 480 lines of the video because they're captured at the same time. On interlaced cameras like the PDX-10, there is some blending of scan lines applied to the image. Picture a very thin horizontal line that would be resolved as one pixel wide. On an interlaced camera this line would appear in the odd video field, but not in the even field. The result would be a flickering effect. To prevent this, all interlaced cameras do some averaging of scan lines.

The end result is that the DVX-100a in progressive mode can do pretty decent 16:9 when compared to interlaced cameras, even though it only uses about 360 lines. I think most interlaced cameras only have a vertical resolution of around 400 lines, so it isn't a huge difference. Of course you'll get even better results with the anamorphic lens on your DVX-100a.

Bill Porter · August 14th, 2005, 03:18 PM

Good stuff! This part explained it better than I've seen elsewhere:

Quote:

Originally Posted by Boyd Ostroff

The "squeeze" is really more like a "stretch" actually. The resulting 720x360 image is stretched vertically back to 720x480 which makes it a proper anamorphic image.

To add upon what you wrote, the "squeeze" is really more like a "crop and then stretch".

This part I don't get though:

Quote:

Originally Posted by Boyd Ostroff

But putting the anamorphic adaptor on a 4:3 camera is a little different matter. In this case, you're increasing your vertical resolution.

I think you mean, "In this case, you're not decreasing your vertical resolution like you would if you were shooting in 'squeeze' mode." Right?

Also,

Quote:

Originally Posted by Boyd Ostroff

On a camera with higher resolution CCD's there would be a minimum of 960x480 pixels in the initial 16:9 image.

A minimum? Why, do some cameras pack more pixels into their 16:9 images?

And also, even if they do pick up 960 pixels wide, or more, in the end isn't that pointless because it gets downrezzed to 720 wide?

Thanks, I'm finally feeling some relief about grasping this stuff!

Boyd Ostroff · August 14th, 2005, 04:59 PM

Quote:

Originally Posted by Bill Porter

I think you mean, "In this case, you're not decreasing your vertical resolution

Sure, "not decreasing" = "increasing" :-)

Quote:

Originally Posted by Bill Porter

Why, do some cameras pack more pixels into their 16:9 images?

Well in the case of the lower cost cameras, I think the real motivation was providing CCD's capable of taking better still photos (in "photo mode"), but as a fringe benefit they are able to sample them in such a way as to give your "real" 16:9. The goal is to have a minium of 480 scan lines, since that's what the final image will have.

We know the aspect ratio of the unstretched widescreen image must end up as 16/9 = 1.78. To conform that to DV's standard of 720x480 we'll have to squeeze it horizontally - that can't be helped, but the important thing is to have 480 vertical lines to start with.

When you take into account the fact that pixels aren't square on 4:3 video CCD's, 960x480 would be the minimum area needed to shoot full resolution 16:9 - see Chris' explanation of the XL2.

If pixels are square (like a computer monitor, plasma TV, LCD screen) then your width is (480 x 16 / 9) = 854. But pixels on 4:3 CCD's are not square, so in order to end up with 854 on playback you need 960 of them. But that's really a camera design decision, and a way to get 16:9 out of native 4:3 CCD's. It would be easier to understand if the CCD's had 854x480 square pixels and were shaped in 16:9 proportions. Then you could just use the center 720x480 area to shoot 4:3. I believe this is how some of the more expensive cameras (like the DSR-570) do it.

So it really doesn't matter how many dots there are on the camera's CCD's, the principle is always the same:

1. Focus an image on a 16:9 proportioned area of the camera's CCD's
2. Scale it so the height is 480 pixels without changing the proportions
3. Squash it horizontally to 720 pixels while keeping the height at 480
4. Apply DV compression
5. Record to tape

Everything that happens after that depends on your TV set!

August 9th, 2005, 08:19 AM	#1
Bennis Hahn Major Player Join Date: Mar 2005 Location: Richmond, VA Posts: 221	Going Anamorphic? I have been thinking for a while that I would do me good to pick up the 37mm anamorphic adapter from Century Optics. The only hesitation is that I have never seen footage from it. I am wondering, how much of a difference will I see in picture quality? Right now I just use the slim mode on my camcorder to make 16:9 DVD's and it looks OK on my 57" TV, but not great. How much of a difference will those extra pixels make? I am filming a longer short this fall and I plan on outputting to DVD, with the possibility of projecting in (digitally) to a decent size. Therefore, I am thinking all the pixels I can get will be beneficial.

August 9th, 2005, 11:08 AM	#2
Jason J. Gullickson Regular Crew Join Date: May 2005 Location: Madison, WI Posts: 170	This site goes into all sorts of detail about what happens when using "stretch" mode vs. optical widescreen adapters. You may have seen it already but I thought it was worth mentioning here: http://members.macconnect.com/users/b/ben/widescreen/ I didn't see any still example shots but I'm sure you could email the guy for more info. __________________ Jason J. Gullickson Producer the second society http://2soc.net

August 9th, 2005, 01:57 PM	#3
Boyd Ostroff Wrangler Join Date: Dec 2002 Location: Mays Landing, NJ Posts: 11,802	I haven't used the 37mm anamorphic - and the topic has come up here before, but nobody had personal experience with that particular model century lens. However, the following tests might give you an idea of native 16:9 compared to cropped/stretched 16:9. See the comparison images from the PDX-10 which shoots real 16:9 with megapixel CCD's vs the VX-2000 which just crops the 4:3 image and stretches it back to make it anamorphic: http://www.greenmist.com/dv/16x9/

August 9th, 2005, 02:42 PM	#4
Bennis Hahn Major Player Join Date: Mar 2005 Location: Richmond, VA Posts: 221	Those two links were great, thank you. Now another question. If I shoot with the anamorphic adapter and slim mode (for 2.35:1) how to I "un animorph" the footage in Vegas? I know regularly one would render out to 16:9 but then the footage is still squeezed vertically.

August 9th, 2005, 08:33 PM	#5
Marco Leavitt Inner Circle Join Date: Jul 2002 Location: Albany, NY 12210 Posts: 2,652	I haven't used that version either, but I have used the bayonet mount version for the GL1. Compared to digital 16:9, the difference is dramatic.

August 10th, 2005, 04:19 AM	#6
Boyd Ostroff Wrangler Join Date: Dec 2002 Location: Mays Landing, NJ Posts: 11,802	I can't tell you specifically how to unsqueeze this in Vegas since I use FCP, but there are a few principles to consider. Are you planning to work in standard definition DV? If so then there's probably very little advantage to the "double anamorphic" technique you describe, and it may even degrade the image more. There should be some way to verically squash the stretched image to bring it into the correct proportions, but you will still by limited by DV's 720 pixel width. If you wanted to up-rez it somehow to HD you could stretch it horizontally to 1280 or 1920 pixels but I think it would look very soft at that resolution. If you really want the 2.35:1 aspect ratio I think your life will be much easier to shoot in 16:9 with the anamorphic lens, then just crop to 2.35:1 in post by putting a black bar above and below the active image area. If your goal is to put this footage on a DVD then that's basically what you're going to end up with regardless of how you accomplish it - just look at any commercial DVD...

August 14th, 2005, 04:23 AM	#8
Boyd Ostroff Wrangler Join Date: Dec 2002 Location: Mays Landing, NJ Posts: 11,802	NTSC DV is 720x480, period. If you're working with DV then that's what you've got. Anamorphic 16:9 stretches the pixels wider on playback, but there are still only 720 of them on each line. If you wanted to stretch them wider to acheive 2.35:1 you're still only working with 720 pixels. 1280x720 and 1920x1080 are high definition standards. You could blow up DV to these sizes, but garbage in = garbage out :-) What kind of camera are you using? Anamorphic adaptors have limitations of focus and zoom range, and since you're adding additional glass in front of the existing lens there will be some image degradation. If you really want to get higher quality for your widescreen TV you might want to upgrade to a camera with higher res CCD's that can do real 16:9. Canon, Panasonic and Sony have inexpensive models that can do this now. Then there are medium priced models like the Sony HC-1000 and Panasonic GS-400. The Sony PDX-10 does nice 16:9 although it may be discontinued. Beyond that there are the new Sony HDR-HC1 and HVR-A1 HDV cameras and then the HDR-Fx1 and HVR-Z1, but these are more expensive. They do shoot in the 1080i format however.

August 14th, 2005, 08:02 AM	#9
Bennis Hahn Major Player Join Date: Mar 2005 Location: Richmond, VA Posts: 221	Right now I have a GS200. Laugh if you want, but really the camera produces beautiful pictures when there is enough light. Since I am shooting a movie, the lighting will be controlled and that will not be a problem. I really would like a new camera, but I simply do not have the money for it. Right now what little money I have is going toward getting good audio.

August 14th, 2005, 09:30 AM	#11
Bill Porter Major Player Join Date: Mar 2005 Location: Ventura, California, USA Posts: 751	Thanks, Boyd. I hadn't "gotten it" that all we are doing is stretching pixels. What, then, is the difference between squeeze mode and shooting in 16:9, in real resolution? If NTSC DV is 720 pixels no matter what, how is squeeze mode not every bit as good as an anamorphic adaptor? The way I understand it (which I know is incorrect, hehe), in both instances an image is getting projected onto the same number of pixels across the camera's CCD's. BTW to answer your question, I'm not trying to shoot 2.35:1 widescreen, actually. Just trying to learn so I don't feel like such a dipsht. You're right that Panasonic offers some inexpensive models that shoot in true 16:9. Trouble is, the DVX doesn't (I now have one), and the inexpensive models lack the good stuff the DVX has. Bennis, the GS200's do shoot nice stuff with enough light. I had a GS120, its forerunner, and was impressed as heck. Current 3-CCD consumer cams are very* powerful, especially considering how small they are! It is only a matter of time before cameras that size - and smaller - are shooting 24p, onto large-capacity, affordable media cards.

August 14th, 2005, 12:54 PM	#12
Boyd Ostroff Wrangler Join Date: Dec 2002 Location: Mays Landing, NJ Posts: 11,802	OK... in talking about 2.35 the issue stretching the width to something greater because I assumed you were already getting the maximum vertical resolution. But putting the anamorphic adaptor on a 4:3 camera is a little different matter. In this case, you're increasing your vertical resolution. If you shoot in "squeeze" mode there are a couple things happening. First, the camera crops its native 4:3 image to 720x360 in order to acheive the 16:9 proportion. In doing so you're throwing away 120 pixels in the area above and below the image, and 120/480 = 0.25, so you have just discarded 25% of your vertical resolution. The "squeeze" is really more like a "stretch" actually. The resulting 720x360 image is stretched vertically back to 720x480 which makes it a proper anamorphic image. There is no difference in the format of this image than what you get from a "real" 16:9 camera; what's different is the way it was created. On a camera with higher resolution CCD's there would be a minimum of 960x480 pixels in the initial 16:9 image. Then that would be horizontally squashed to 720x480, but you would have all 480 vertical lines so there's no resolution loss. Chris has a nice graphic which shows how this works on the XL2: http://www.dvinfo.net/canonxl2/articles/article06.php I did something similar showing the CCD setup on the PDX10" http://www.greenmist.com/dv/16x9/10.JPG But the DVX-100a is a slightly different beast because of its progressive modes. In progressive mode you can actually use all 480 lines of the video because they're captured at the same time. On interlaced cameras like the PDX-10, there is some blending of scan lines applied to the image. Picture a very thin horizontal line that would be resolved as one pixel wide. On an interlaced camera this line would appear in the odd video field, but not in the even field. The result would be a flickering effect. To prevent this, all interlaced cameras do some averaging of scan lines. The end result is that the DVX-100a in progressive mode can do pretty decent 16:9 when compared to interlaced cameras, even though it only uses about 360 lines. I think most interlaced cameras only have a vertical resolution of around 400 lines, so it isn't a huge difference. Of course you'll get even better results with the anamorphic lens on your DVX-100a.