How to get best 16:9 footage

[Rob Hester]

Hey all...I've mostly hovered around here...reading...learning but finally I have a problem that searching cannot solve for me.

I have a camcorder which has distinctly different shooting modes. Frank I really don't want to start an argument here, I have a question that I'd really like answered and getting it requires me to say certain things which you may object to....Okay, for the rest;
My camera shoots in 2 modes -
-Interlaced (60i) in:
"normal" (4:3)
"cinema" (16:9 using black bars)
"sqeeze" (anamorphic)
Non-Interlaced (30P) in:
-D.wide mode (4:3)

So, here is where my problems arise. I want to shoot in the best 16:9 that my camera has, but that would be sqeeze...so I have to give up 30 full fps shooting and take on interlace artifacts. However, d.wide mode on my camera uses the *full* ccd (takes away EIS) and therefore has a better overall end resolution and look to it. So...I am inclined to think that the best looking 16:9 can still be achieved by my d.wide mode...right? The problem is, how do I go about checking this? I have looking through both AE and Premiere Pro and the best I can find is AE's "Region of Intrest" but how much do I crop? What are the correct dimentions? I don't want to loose any more screen information that I already would have to...

Anyways, thanks to anyone who has suggestions...Frank if you read this...please; this is a real question I have and I came here looking for answers, not to start a debate over what in fact this camera's capabilities are. Thanks in advance for respecting that.

If anyone has any questions on the different modes of the camera or why I like d.wide so much, email me or give me some webspace...might be able to send some raw dv over a messaging program to anyone who could help or who is curious.

Thanks again,
Rob

[Rob Hester]

pleaseeee!! anyone! :(

[Boyd Ostroff]

OK, I'll take a stab at this but I don't really understand the distinction between the different modes. Obviously you want to use as much of the CCD surface as possible. Personally I don't think loss of 30p should be a big issue. Is it true 30p or is it like the Canon "frame mode"? You can come pretty close to approximating this in post using an adaptive deinterlacing program or plug-in.

As a general principle, you will need anamorphic video to get the best results. DV is always written to tape as a 720x480 frame (NTSC). Any mode where the camera crops this to acheive the 16:9 aspect ratio will degrade your resolution.

As for how to check, shoot a few seconds of footage in each mode, then export some still frames after capturing in your NLE. Import these frames into Photoshop. You can then stretch the 720x480 raw frame to 854x480 which should put it in the proper aspect ratio to view on a square pixel video monitor.

For your test you might want to shoot a resolution chart. You'll want to check the vertical resolution of the clips, since that's what suffers when the camera first crops then later stretches to make the results anamorphic. I did some tests to compare the 16:9 modes of my own 2 cameras. While these won't tell you anything about your own hardware they might provide a good model of how to do some A/B comparisons. You'll find my examples here. You can download and print up the resolution chart at John Beale's excellent website here. Again, the specific thing that I believe you'll want to compare is the vertical resolution of the different 16:9 modes, which will be visible by enlarging this section of the chart. In that example note how the PDX-10 produces superior 16:9 as compared to the VX-2000. The vertical resolution can be read by noting the point where the converging lines are no longer distinct.

Hope this gets you started in the right direction!

[Rob Hester]

The reason I wish to keep using the 30P mode is:
a) works as a free .7x wide angle adapter
-zoom starts at .7x and up as opposed to normal mode at 1x and up
b) the full ccd is being used
- more pixel information to reduce down to = better image (opposed to normal)
- think of the difference between taking 2 of the same still frame pictures of flowers; 1 at 1024x768 and the other at 800x600 and then reducing both to 640x480 - the larger one will look more detailed
- this is compared to the normal mode...but both normal and sqeeze use the same (inferior) starting image, i'd think that the image would be worse
- this frame mode on my camera...I am not knowledgeable enough myself to tell EXACTLY whether or not it is "30P", but it is definately 30 full frames per second, I can send people raw video if they wish to lend their opinions. it is not frame mode like on the canon because there is no loss in resolution; quite the opposite if anything


I think I have come to the conclusion that I would need an anamorphic adapter to get the same shot...this entire thing doesn't make sense to me...if someone could properly explain the .9dv/1square/1.2widescreen pixel sizes it would be appreciated.

What I don't get is...both the sqeeze and the dwide spit out 720x480 still frames; but when put in proper perspective (as Boyd points out to me -thank you so much for htat huge post!!!) the squeeze frame needs 854x480.

If I put my 4:3 cropped dwide footage into THAT, it is obviously smaller...and i am guessing here, that the only way i could get my footage to meet that res w/out stretching or blowing up the image would be to get an anamorphic adapter..and THEN I would have the best resolution AND keep my frame mode?

[Rob Hester]

AGH!!!! I just looked for prices for anamorphic adapters...how can century optics be the only one!? I cannot afford something that is over twice the price of my camera! :(

Anyone know of any cheaper ones? or maybe someone has agus35'd something? ;) ohhh thats right! it's lingo-ified now!

[Helen Bach]

An exasperated Rob wrote AGH!!!! I just looked for prices for anamorphic adapters...how can century optics be the only one!? I cannot afford something that is over twice the price of my camera! :(

Which Century anamorphic are you looking at?

Optex supply one as well, though the price is similar to the cheaper of the Optex ones. The machining on the one I handled was very like that on lenses from the Russian Lomo plant - this is not derogatory in any way because Lomo make some very high quality movie lenses, nor am I asserting that Lomo supply Optex.

The word is that the Optex permits a little more of the zoom range to be used, while the Century has the slight edge in definition. The BBC evaluated the two, and went for the Century for supply with the VX2000, but I think that it was a close-run thing.

Best,
Helen

[Rob Hester]

thanks for the link! unfortunately...that is still the price of my camera :(
I don't understand why anamorphic lenses are SO much more! The glass must be of higher grade...but come on...$45-$200 for a .5x wide angle'd lens, but $699-$1299 for an anamorphic?
frustrated :(
thanks for the post though!
Rob

***EDIT***
I just compared my "sqeeze" footage to my "regular" note: not "dwide"
Unfortunately...the sqeeze mode on my camera isn't using ANY more of the CCD than my regular 4:3 mode. This means that both modes are capturing the SAME data, it's just that the "squeeze" mode auto crops. I was hoping that it squeezed more image into the frame, if it had tapped into the extra CCD space ala dwide it COULD...but it isn't programmed to work that way.

The only plus I can see to squeeze then would be that not as many pixels need to be compressed, therefor less artifacts? I guess it is all irrelevant now; the "dwide" mode captures more data anyways.

**1 MORE question...I realize that no one is responding and I am prettymuch blabbing to myself...but this is my last one!**
If I put black bars on the screen on dwide mode ( I can using a title function) would I be saving compression data for the rest of the frame inside? I am theorizing yes, because total blackness must take less processing data than a ton of colours right? Think GIF format or JPEG, all that is encoded is hey, this is black throughout these pixels, not processing seperate colour info and optimizations for each one....a compressed image will take up less space if you had say..half of it black and half of it white than a full colour waterskiing picture.

Well, I semi found it out on my own, the discussion wasn't as much as I hoped...but maybe there are a few out there who didnt respond and it has helped them...because I know i've read many a reply and done the same (guilt!!) Anyways, thanks to all that read through all the blabbing!
Rob

[Boyd Ostroff]

Rob, with all due respect I think you're getting too hung up on the conceptual side of things. Did you try to photograph the resolution chart I recommended using your various modes? This will give you some hard data for comparison and should put your mind to rest as to the best approach. Time for some "due diligence" :-)

The real issue is that you want to use all 480 scan lines. Any approach to 16:9 which doesn't do this will suffer.

Anamorphic lenses are expensive because they're so specialized and compress the image in one dimension. Also there is not such a great demand for them as there would be for regular adaptors. Like other optics, the diameter is a big factor in price as well. Century sells an anamorphic adaptor to fit camcorders with 37mm filter threads for something like $250.

Somewhere I saw a post that described adapting an anamorphic projection lens to a camcorder. A quick search for "anamorphic" on ebay came up with several at prices like $30, $80, etc. If you're interested in a "science project" this might be an approach for you.

[Don Berube]

If you do not have the means to shoot 16:9 with an anamorphic lens adaptor - just shoot 4:3, frame for 16:9 while shooting and apply a letterbox mask in post.

You are not quite losing resolution this way. Try the following as an example... turn on your television and play the best quality DVD you can through it. Now, temporarily apply a strip of gaff tape to the top and bottom of the screen to emulate a 16:9 frame. You have, in effect, just applied a letterbox mask. Now, look at the image quality of the DVD material with this "letterbox mask" which you have just applied. Did the image quality of the image inside the frame deteriorate once you applied your letterbox mask? No.

Sure, it would be nice to "squeeze" the entire 4:3 image from top to bottom into the narrower 16:9 frame - the idea being that you would have more resolution inside that 16:9 image frame... but your camera just doesn't do this without softening the image, as you have stated.

So, move forward and shoot in 4:3 mode while framing for 16:9. Get a wide-angle adaptor such as a .7x so that your wide shots can be even more majestic looking. Just make sure that you do not pull out too wide that you induce vignetting in the corners of the frame or induce too much barrelling on the left/ right sides of the frame - unless you like that look of course.

I shot a 24P DV short recently called "Full Service" on a DVX100 with the above technique using the camera's built-in 16:9 letterbox mask mode and a Century Optics .7x anamorphic adaptor. It premiered last evening at the 2003 DV Film Festival sponsored by Panasonic at the DV Expo West. They had a h-u-g-e screen at the Los Angeles Civic Center. It looked great, everyone applauded and we received many compliments afterwards. Nobody complained about any resolution loss or even noticed that it was shot with built-in 16:9 and not anamorphically. Everyone thought it looked great.

You can do this too!

- don

[Rob Hester]

Boyd, I did that last summer. If you go to my website in my signature you can look at the res chart too and tell me what you think. Normal mode uses the same shot as squeeze as I have found out later.

I have pretty much figured that d.wide 4:3 is my best bet...I am only now wondering about letterboxing in post, or in camera. Do you have thoughts on whether I could get a better, less compressed image if I letterboxed before or is this just a little bit over the top?

I'll shoot another res chart if that would show me the difference, but I am not expecting a HUGE one anyways....it's just a fine detail that would make me feel better knowing that I am getting the best quality frames I can...isn't that what we all want? Don, hearing that makes me feel a little bit better about this situation, thanks for your input...it's nice to hear of real-world examples...specially ones that were viewed on screens MUCH bigger than monitors or TV's. Thanks for your time guys,
btw....being a philosophy major may lend some reason to why I am so attentive to detail :S
Rob

[Boyd Ostroff]

<<<-- Originally posted by Rob Hester : Do you have thoughts on whether I could get a better, less compressed image if I letterboxed before or is this just a little bit over the top? I'll shoot another res chart if that would show me the difference -->>>

Like I said before, this is the only way to really answer the question. It seems that not many people around here are familiar with the camcorder you have, so you will have to blaze your own trail.

I tried this on the VX-2000 and was surprised to find that cropping in post gave much cleaner results. http://www.greenmist.com/pdx10/mode/03.jpeg Evidently the DV compression mangles the image further (look at the converging lines at the top of this example).

[David Warrilow]

Dude,

you're working way to hard with all this. If you've decided to shoot in the D.Wide mode - because you like the 'Framey' look or because you think it provides a better picture or whatever, then you really have three choices; an anamorphic adapter, (which you have pointed out are ridiculously expensive for most of us and therfore not really an option - unless you can find a good second hand one on ebay), in-camera letterboxing (using the JVC title function detailed on another forum which you frequent - normal 'cinema' letterboxing don't work in D.Wide mode) and letterboxing in post in say, you're editing program.

These are really six of one, half a dozen of the other. If you letterbox in camera - then you have no problems framing everything correctly as you can see where your boundries are. However, if you decide you want different framing after you've shot - too bad. The image is already masked and there's not much more you can do except re-shoot - or resize and reframe and that will definitely degrade your image.

If you shoot without the title-letterboxing, 16:9 framing is more tricky cos you're looking at a 4:3 image - but after the fact you have the whole 4:3 frame to play with. If you want to re-position your framing in post - you at least have a little room to play with. Re-frame slightly in your editing program, then letterbox. Hey presto.

As to the 'quality' issue - any time you apply an effect in a program, even letterboxing, you are deconstructing and then reconstructing your image - therefore there is some (albeit negligable) loss in quality even in programs that operate in yuv colourspace (avid/PremPro/DVStorm) - so in that respect, technically it would be better to letterbox in-camera.

But please remember, although we all strive for the best possible images from our equipment - the 3K is a single chip, consumer camera - not a cinealta. So you are working within at least some limitations - so don't drive yourself nuts worrying about infinitesimal details that, if you're telling a good story, no one will notice anyway.

Best Luck.

DW.

[Rob Hester]

Sorry guys...here's why it is so important...
I have gone away to university and my family is going on a cruise together over the xmas break. This will be our last trip "as a family" since my sister and I have grown older. I want to get the best footage of the sights possible...I KNOW I am being anal about it, but this is something I'd really like to give to my parents as a special...pro as possible gift after the trip.

I know the 3k is a 1ccd consumer cam, but it isn't HORRIBLE, and it can take some pretty darn good images in the right settings...I am just trying to get the best out of my equipment - if this was another dormatory prank I wouldn't care too much, but this is something special to me. A 16:9 framing would accentuate the beautiful panoramic views that I am sure we will see.

All I have left to decide (I tried res charts last night but I don't have a tripod at school, results aren't reliable):
letterboxing in post, or via title on camera?

David (thanks for the time btw!), I know about the reconstructing, but that really isn't an issue; it's going on to the computer once, not back to tape. The idea that I thought was that if you use the title, you are already blocking out some useless data which would otherwise degrade picture quality for the middle - what you would be using. I know the results will be quite small, but...if they are better, I can hopefully have a better end picture to present to my parents (DVD).

Thanks for the input once again!
Rob

[David Warrilow]

Hi Rob,

all I can recommend is what I would do - ( I've had a 3K for about 18 months and know it's a good little camera). I would shoot d.wide with the title letterboxing. It just allows you to compose your shots well on the fly and not worry about 'will this look good 16:9? Am I allowing enough head room?' etc. etc.

I think that d.wide produces better, less harsh looking video and is more sensitive when the light starts to fail. Then in post I'd look at boosting the saturation about 15% to really make those vistas shine. If you letterbox after the fact it will lose a little quality because of the com/decomp process of adding the effect. I don't really understand what you mean by "...if you use the title, you are already blocking out some useless data which would otherwise degrade picture quality for the middle..." I think with either method you are going to end up using 432 lines out of 576 - (full vertical height minus 25% letterboxing - erm, different numbers if you are in NTSC) there ain't really any way around that - but remember with d.wide - you're starting off with a 1280 x 960 pixel sample that is squeezed down into 720 x 576 - so this fact (and from your testing it seems to pan out) should ensure a better quality of video detail than starting with only 720 x 576 as in normal video mode. I guess the only other question is are you going to leave the image as 4:3 letterboxed - or are you going in and resizing and changing the aspect ratio so that when the dvd plays on a widescreen tv it fills the frame with no letterboxing? If you're doing that then that's where the potential for loss comes in bigtime.

To get an idea of how the images will translate; shoot something in title-letterbox d.wide - then play it back on your tv and then using the " T - W" buttons on the little remote that comes with the camera, zoom in the picture until the letterboxing disapears, This is in effect the quality of the image you will get by converting the 4:3 letterbox to true 16:9 aspect before encoding to dvd. If you have a look - I think the results are pretty damn good.

It's obvious you care about this a great deal and that is going to be the mittigating factor in your footage. It's going to look alot better if you care about what you're shooting than if you don't give a damn. I hope, however you decide to shoot it, that it turns out stunning.

Best of luck.



David.

[Don Berube]

>>>>>>>>I would shoot d.wide with the title letterboxing.

- I would agree with David Warrilow that, all things considered, shooting in "D.Wide" Mode would be the best approach for your particular project. I'm assuming that you intend the DVD video to be resolved on a 4:3 screen with a letterbox mask.

Just keep in mind that if you affecct the contrast and brightness of your images during color correction - that you may need to re-apply the black 16:9 letterbox mask so that it is the same black level consistently throughout the project. This is no biggie.

- don