Resolution reduction in frame mode and 16 x 9 mode

[Hugh DiMauro]

So, I'm reading a DV INFO.NET article about progressive scan by Steve Mullen, and he describes how the XL1s frame mode delays the row two green elements by one time line and then added to the row three green elements etc etc yadda yadda yadda which results in reduced vertical resultion only giving 320 lines of resolution. Another article I read stated that using the 16 x 9 function further reduces vertical resolution.

Here's my question: If this is the case, why bother? Doesn't that just defeat the whole purpose of the Canon's superb lens optics and 3 ccd capability? So I should just always shoot interlaced, using the 16 x 9 guides and de-interlace in post?

[Bram Corstjens]

That's absolutely right... On the Canon you will lose resolution by the frame mode and by using 16:9 mode another 25% is lost.

Just shoot regular and de-interlace in post using a good de-interlacer like Dvfilm maker or Magic Bullet.

[Jon Fordham]

Frame mode is a type of de-interlaced imaging.

And you're forgetting about compression.

If you want 16/9 and you want a non-interlaced motion signature, then I would suggest doing it in camera.

First of all, yes, the XL1's Frame Mode does reduce vertical reoslution. However, so does de-interlacing. Even the best de-interlacer (regardless of whatever the advertising may say) will end up reducing your vertical resolution by as much as %50! So in my opinion, if you want a non-interlaced motion signature, you might as well do it in camera using the Frame Mode option and save yourself a bunch of rendering time.

IF, you are going to shoot interlaced and perform a de-interlacing in post, I would recommend you find a de-interlacer that performs an interpolation rather than just a straight de-interlacing.

Second, when it comes to achieving a 16/9 image, in almost all cases, I believe the in camera electronic mode is best. For two primary reasons. One being compression. If you want a letterboxed or widescreen image, why force the DV codec to compress information you don't want and don't need. By throwing out that part of the image you don't want, the compression codec can focus on compressing the image you do want with much less compression. Thus achieving a better picture. And two, the electronic 16/9 mode records a 16/9 ID Pulse. That ID pulse tells all widescreen capable television sets and monitors that the image is a 16/9 widescreen image. Thus hardware and software that is capable of reading that ID Pulse will be able to automatically switch to a 16/9 widescreen aspect ratio. And of course, let's not forget about pixel apsect ratio. Should your footage require or desire upconversion to a High Definition format, you need to have the pixel aspect ratio match the anamorphic ratio to fill the entire native 16/9 widescreen aspect ratio of HD standards.

If you choose to crop the 4/3 image in post, you'll be forced to either upconvert your footage to a "pillarboxed" HD image, or try to pull a 16/9 image out of the 4/3 by cropping and then re-sizing the remaining pixels.

In all cases a true 16/9 CCD or optical anamorphic lens will produce superior results.

And in all cases a true Progressive Scan capture will produce far superior results from a resolution standpoint.

[Dave Largent]

Hi,
There are resolution test charts on the net that show that,
with the VX2000, you get a higher res image by shooting 4:3 and then converting in post. I hear with the Canons, you're better off doing 16:9 "in camera".
Dave
P.S. I think someone on this board did the res tests I mentioned.

[Jon Fordham]

Dave,

Could you please provide the link that show these resolution tests?

I don't doubt these res tests show that the VX2000 and PD150 have a less than desirable electronic 16/9 mode. I can tell you from experience that I have not been able to get (what I feel are) acceptable 16/9 from either the VX2K and PD150's electronic 16/9 mode.

However, as opposed to cropping the VX2K's image in post, I would suggest that you use the camera's "memory mix" ability to actually crop the image in camera. As I stated previously, by throwing out the part of the image that you don't want, you allow the DV codec to focus on compressing only the image you need. The resulting lighter compression is noticably better. And since you're not asking the camera to vertically resize the image, it should achieve the same (if not better) resolution results as the "cropping in post" method described in these res tests as showing.

You can create these crop mattes to whatever ratio you desire in photoshop or even FCP. Create black bars at the width of whatever you want the ratio to be, then apply them on top of a blue background. Then export that image as a still frame that can be imported onto the memory stick using the supplied memory stick adapter. The use the camera's memory mix function to key the image onto the blue area of the matte. This is the same way you get SMPTE "split" color bars on the VX2K or PD150.


Also...

Dave, I don't mean to sound like a d*ck, but if we could all refrain from posting what we heard, I think these boards would be a better forum for real and factual information that we could all rely on. The whole, "my friend told me this", "my friend told me that", and "I heard" kind of thing only leads to misinformation. If you have a reliable source that you can name who has experience with such, then by all means please share. But just saying that you heard something doesn't help much.

Don't get me wrong, I'm not giving you a hard time. I too have heard it said that the Canon 3 CCD cameras achieve better electronic 16/9 images. And as an owner or the XL1 and VX2K I can tell you that the XL1 does appear to achieve a better electronic 16/9 image than the VX2K. To my eyes, my old TRV900 achieved a better electronic 16/9 image than both the XL1 and the VX2K. I have recently worked with the PDX10 and the TRV950 and can tell you from experience that both cameras produce the best looking 16/9 images of any of the 3 CCD MiniDV cameras that I've ever seen. I can also tell you from expierence that the electronic 16/9 mode on the 1 CCD Sony PC120 is quite beautiful.

[Dave Largent]

Here it is Jon. I think you'll also find interesting the comparison between the wide modes of the PDX10 and the VX2000.
http://www.greenmist.com/pdx10/

[Dave Largent]

By the way Jon, hows the low light reach on the PC120. Think the
cam would be okay indoors?

[Jon Fordham]

Dave,

Interesting charts indeed.

My experience with the PC120 is pretty casual. Two filmmakers that I have worked with numerous times both own one. One of them was using his PC120 to shoot a documentary of the music video that I had been hired to DP. He was shooting everything with the in camera electronic 16/9 mode. After we wrapped one day he showed me some of his BTS footage from the PC120. The camera was connected via Y/C to a Sony PVM-14m2u 16/9 color production monitor. The footage was excellent. No unusaul aliasing or rescaling artifacts that I could see. The footage was shot exterior in average daylight. So the light was pretty good to begin with on that stuff. After viewing the stuff I was so impressed that I asked to play around with it a bit. I did some very unscientific shooting in his home. This was interior during the day with moderate to low daylight coming in. The camera held color rather well and got a resonably nice level of exposure. I took the camera into the no windowed bathroom to check out its low light. Again, the color and exposure came out pretty nice. It started to show a bit of loss under the three or four 60 watt tungsten bulbs in the bathroom. But nothing unexpected from a 1 CCD MEGA pixel camera. I found I could easily push it to 6db's gain without noticing ANY noise. 9db's introduced some noise, but not unacceptable levels.

This of course was not the best of shooting conditions or any sort of careful evaluation of the camera. Just some playing around. But all in all, I was very impressed. I wouldn't say surprised by the level of performance. Beacuse I've done a few small projects with the Sony PC100. But the newer PC120 definitely took things up a notch. After that the director of the music video went out and bought one!

[Barry Green]

<<<-- Originally posted by Dave Largent : I hear with the Canons, you're better off doing 16:9 "in camera".
-->>>

Here's a link that shows how high-quality a job a Canon GL1 does with in-camera 16:9

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

Adam Wilt's rankings of which ways are best to get 16:9 are at:
http://adamwilt.com/DV-FAQ-etc.html#widescreen

Those are the two references normally given when saying that Canon's in-camera is pretty good. For showing that Sony's in-camera is pretty bad, the references are usually Adam Wilt's site and Boyd Ostroff's site.

[Jon Fordham]

Barry,

Your statement, "For showing that Sony's in-camera is pretty bad" is not an accurate representation of the findings respresented on Ostroff's site. Ostroff's res test shows that the Sony VX2000 does not achieve the same level of image quality and resolution as the newer Sony PDX10. That is all. It is not a representation of all Sony cameras. Nor is it a representation of Sony's inferior 16/9 compared to any other cameras.

In my experience, the PDX10 is capable of much higher image quality and resolution in 16/9 mode than the Canon XL1/XL1S and GL1/GL2. This is only to my eye viewing the image on production monitors while working. I have not compared any of those cameras in a controlled and scientific testing enviroment.

If I recall correctly, Wilt's site states that Sonys don't do as good a job as the Canons. Not that the Sonys "are pretty bad". And if I recall correctly, that particular statement was written prior to the introduction of the PDX10 and does not compare any of the Sony 1 CCD MEGA pixel cameras.

[Barry Green]

The PDX10 is excluded from this discussion because it doesn't do stretching. It uses an entirely different method, and therefore is only tangentially related to the discussion at hand.

The VX2000 delivers about 230 lines of resolution in 16:9 mode. That is, by my standards at least, "pretty bad." If you want another source for comparison, check out:

http://216.239.57.104/search?q=cache:HgKJwbcCrA4J:www4.big.or.jp/~a_haru/0208_3CCD.html+mx5000+trv950+vx2000&hl=en&ie=UTF-8

At the bottom of the page they have resolution charts shot in 16:9 mode on a TRV950, MX5000, VX2000 and a Canon XV2. The TRV950 and MX5000 are substantially, dramatically, incredibly better than the others, which is to be expected because they use the megapixel method. But of the VX2000 vs. Canon, the Canon beats the VX2000 easily. Whether you agree with me or not that the VX2000 looks "pretty bad", I will stand by the statement: as far as electronic 16:9 stretching goes, the Sony's look "pretty bad," whereas the Canons look better. And neither holds a candle to the megapixel system as employed by Panasonic and Sony.

[Alex Dunn]

If you want to shoot 16:9 with 4:3 CCDs without losing resolution there's one true way to do it without losing resolution. The same way the SOME of the film guys do it, with an anamorphic lens. You can get a great 16:9 Anamorphic lens from Century Optics for around $395. This optically squeezes the image horizontally, so you get an image that is truly wider, not cropped when you smoosh it back down in post (or with a 16:9 projector or monitor that has anamorphic capability).

[Ignacio Rodriguez]

Yesterday I connected the PDX10 to a big Sony 4:3 TV with 16:9 mode. I was very impressed by the image at 1/30. Sony cameras are famous for the line doubling that happens when going below 1/60. Now it would appear that the PDX10's line doubling is not as bad as the other Sony cameras I have tried. In 16:9, I could not see the line doubling fenomenon at 30fps. Then again, perhaps the TV was not so good a monitor... I did use S-Video. I don't have access right now to my PC3 so as to run some comparisons, but I am sure line doubling is pretty pad on the PC3 when going below 1/60 (or using some of the toyish digital effects), and I remeber noticing something simlar with a friend's PD150. Don´t know why it should be better with this cam at 16:9. Any ideas?