Best 16:9 without anamorphic lens (YAT)

[Joshua Provost]

If this is the case, it might indeed change things. Do you know that this is so, or are you just assuming/hoping?

We'll never know, because Panasonic is not going to reveal how they do it. The end result would indicate that they are doing it properly. We can only reverse engineer and make educated guesses. Seriously, don't sweat it until you get the camera.

Someone wrote here that the squeezing is done by first throwing away the data outside the letterbox and then stretching it to fit the entire "frame". And the amount of vertical CCD lines wouldn't really make a difference because the data there can't be used when stretching the image, right? Or have I misunderstood something here?

True for PAL, but not true for NTSC. Both cams use the same chips. The chip has 576 lines of vertical resolution, regardless of whether it is PAL or NTSC. So, for PAL, yes, you won't get additional vertical resolution, but the color subsampling and DV compression still apply. With NTSC, the 576 vert lines are usually resized to 480 for NTSC to tape. However, in squeeze it would appear that crop-and-zoom is done, not on the 480 line ouput, but on the 576 lines right from the chip, for actual vertical resolutions of around 430 lines, which is what it appears to be.

I've put this theory out on a number of occassions, and so far no one has confirmed, denied, or otherwise commented. I'd like to see some debate on it, actually.

None of that matters to you, since you in PAL-land, but the color subsmapling issue and DV compression still apply.

 

[kurtmo]

I just ran some tests of my own with vegas uprez, squeeze and letterbox. Vegas handled squeeze the best (other options i had to tweek built in 16:9 settings to get things framed correctly) and I thought the crispness of squeeze beat the others out too. i also tried uprez w/ photozoom pro and on my monitor noticed slight increase in noise. I plan to render a few out to DVD and see how they look on a real monitor, but for my upcoming 16:9 I'm thinking sqeeze at this point.

 

[greenlighted]

Well, there you go. Thanks for sharing. I'm gonna do some testing as well once I get the camera. Will post my findings too.

 

[Joshua Provost]

Those who have tested or used SQUEEZE think its a perfect viable option for the right circumstance. If you haven't tried it, or don't own the camera yet, it's easy to get bogged down in the theory.

 

[greenlighted]

That is indeed exactly the case here. While waiting for the camera I checked out the forums for what would, theoretically, give the highest quality 16:9 images (without the use of an anamorphic lens). When the time comes, however, I'm sure I'll use several different ways of achieving anamorphic 16:9, based on the project's specific circumstances. Still, I must admit I like to anchor my usage and decisions in technical theory.

Right now, by the way, I'm really interested in knowing more technical details regarding the uncompressed output through the s-video. And, if it's not DV compressed, whether it will help me get a better key when doing blue- and greenscreen...

 

[Gordon JL]

This is a great thread, and should be stickied if it gets enough responses; maybe from people like Barry Green so that some of these theories could be confirmed.

Barry's always said that if your final product is for a 16:9 television, you should shoot squeeze (or do the Squeeze thing in POST, assuming the stretch/squeeze algorithm is good). So why should you should letterbox instead of Squeeze (or squeeze/stretch in post; therefore shooting 4:3 to begin with)?

Based on your summary of your findings, it seems that it may be an even better idea to shoot Letterbox in cam, than, in post, you stretch/squeeze (not sure which term to use) -- so that you get a similar effect to Squeeze mode, except you did it with Letterbox footage. Not 4:3 footage, which you have to crop.

Here's something I've read about shooting Letterbox mode. The Letterbox frames are actually not 16:9. If you measure your LCD screen, you will see that the black bars on the top and bottom do not crop out what you need for 16:9. I don't know if this is a result of the overscan inherent in the LCD screen or what...

Another thing to consider is; should you frame your footage for overscan, or for televisions that do not have the overscan problem? This is something to consider especially if you're shooting for 2.35:1 or 2.39:1, because every bit of info counts.

Here's a question: if I want to shoot a film at 2.35/2.39, should I frame for 2.35 or 2.39? Personally, I'd prefer for the final product to be 2.35:1 (even though it's technically 2.39 now) so as to not stretch the DV footage to it's limits. But I'm worried about how this would appear on DVD on a 16:9 television, or if I blow the footage up to film... Would my footage remain 2.35:1 (or would the footage get stretched or modified somehow)? Or should I consider the technically correct 2.39 to avoid these situations?

 

[greenlighted]

This is a great thread, and should be stickied if it gets enough responses; maybe from people like Barry Green so that some of these theories could be confirmed.

Barry's always said that if your final product is for a 16:9 television, you should shoot squeeze (or do the Squeeze thing in POST, assuming the stretch/squeeze algorithm is good). So why should you should letterbox instead of Squeeze (or squeeze/stretch in post; therefore shooting 4:3 to begin with)?

Based on your summary of your findings, it seems that it may be an even better idea to shoot Letterbox in cam, than, in post, you stretch/squeeze (not sure which term to use) -- so that you get a similar effect to Squeeze mode, except you did it with Letterbox footage. Not 4:3 footage, which you have to crop.

Yeah, that is indeed what the theories (in the posts here) indicates. When it comes to confirmation, Barry said:

I agree with pretty much everything here, but for two things: [anamorphic adapter is better and HVX is a lot better at 16:9 than DVX]

Joshua disagreed:

If implemented correctly, the in-camera uprez has the potential to be superior to post uprez, and it does seem to be implemented correctly, as the results are great.
[...]
If properly implemented, the SQUEEZE happens before sizing to NTSC/PAL size, before the color is subsampled, before DV compression is applied. It appears that this is the case. If you do it in post, you are uprezing after all of this destructive activity has occured. The damage is done.

There's a logic to both, I'd say, but I've got no idea which one is actually correct.

However, "everybody" seems to think that SQUEEZE is, at the very least, good enough.

Here's something I've read about shooting Letterbox mode. The Letterbox frames are actually not 16:9. If you measure your LCD screen, you will see that the black bars on the top and bottom do not crop out what you need for 16:9. I don't know if this is a result of the overscan inherent in the LCD screen or what...

Where did you read this? In any case, isn't that, as you say, probably just the LCD? I assume you don't need to do additional cropping on letterboxed material to get 16:9, which probably means it is proper 16:9 to start with...?

Another thing to consider is; should you frame your footage for overscan, or for televisions that do not have the overscan problem? This is something to consider especially if you're shooting for 2.35:1 or 2.39:1, because every bit of info counts.

I can never keep overscan and underscan apart but in any case I always make sure I've got a "clear frame" for the bigger of those two and all my relevant stuff within the smaller one, keeping it within action safe margins.


About the 2.35/2.39 stuff; I've got no idea. There's probably some useful information in the forums here, though... :cheesy:

 

[Joshua Provost]

There's a logic to both, I'd say, but I've got no idea which one is actually correct.

Barry's response and my response do not conflict. Barry said that anamorphic adaptor or HVX 16:9 would yeidl better results than any of the methods being discussed. I wholeheartedly agree with that.

I only disagree that post uprez of 4:3 or 4:3 LETTERBOX is better than SQUEEZE. SQUEEZE does the upres in-camera before many destructive image processes occur, so it is superior. You'll find out soon enough when you get your camera.

 

[kryptics]

Oh yes, it's Yet Another Thread about 16:9... And this post is gonna be long, but please bear with me...

As preparation for my purchase of a DVX (will someone *please* sell me a DVX100BE) I've been looking into this entire 16:9 thingy and after reading literally hundreds of posts here about 16:9, 4:3, letterboxing, squeeze, anamorphic, etc etc, I thought I'd do a summary on how to achieve the highest quality anamorphic 16:9 image with a DVX. *I*'m actually not telling anyone, I'm just summing things up, so please, all you really technical guys and gals out there, correct me if I've drawn the wrong conclusions or listened to the wrong people... .

The funny thing is that everyone has a different way of doing this; the one solution that has worked like a charm for me is Magic Bullets Instant HD - since i can't afford a 16:9 - $99 buck is not bad and if you don't mind the long rendering time on after effects it works like magic - then again that's just me...

 

[-zach-]

Doesn't "squeeze" mode on the DVX do something like this?

 

[greenlighted]

Barry's response and my response do not conflict. Barry said that anamorphic adaptor or HVX 16:9 would yeidl better results than any of the methods being discussed. I wholeheartedly agree with that.

I only disagree that post uprez of 4:3 or 4:3 LETTERBOX is better than SQUEEZE. SQUEEZE does the upres in-camera before many destructive image processes occur, so it is superior. You'll find out soon enough when you get your camera.

The way I interpreted Barry's post was that he was agreeing with the summary I had made with two additions, the anamorphic adaptor and the HVX. The summary I had made concluded that shooting letterboxed and using a good resizing algorithm in post would yield the best possible 16:9 image quality (without adapter). You did disagree with that, right? I'm sorry if my choice of words indicated anything else.

 

[Joshua Provost]

The way I interpreted Barry's post was that he was agreeing with the summary I had made with two additions, the anamorphic adaptor and the HVX. The summary I had made concluded that shooting letterboxed and using a good resizing algorithm in post would yield the best possible 16:9 image quality (without adapter). You did disagree with that, right? I'm sorry if my choice of words indicated anything else.

Yes, that makes it more clear. I haven't seen anything in post that does as well as SQUEEZE mode. I've done the shoot 4:3 and uprez to 16:9 in post method a number of methods. It always requires a fair bit of sharpening to not look like mush (whereas SQUEEZE is sharp, even if the resolution isn't optimum).

 

[Joshua Provost]

Doesn't "squeeze" mode on the DVX do something like this?

No. SQUEEZE gives you 16:9 footage, plain and simple. SQUEEZE was a bad choice of a name, IMO.

 

[-zach-]

SQUEEZE gives you 16:9 footage, plain and simple.

That sounds good, so do you see a squeezed image in your LCD or something?

 

[Joshua Provost]

That sounds good, so do you see a squeezed image in your LCD or something?

On the original and A model, yes. I suppose that's where the name came from. Of course, that's just how it appeared on the LCD. The B model has a 16:9 mode for the LCD itself, so it looks correct.

 

[Gordon JL]

You say that Squeeze seems to give you the best 16:9 footage, but what about in conjunction with uprezzing? People who use Photo Zoom Pro seem to think that uprezzing 4:3 looks better than uprezzing from Squuezed (whether squeezed in cam or in post). So, assuming you will uprez, isn't 4:3 the best for 16:9?

What I really want to know is what's the best procedure for getting 2.35/2.39:1 assuming I will uprez eventually... I'd prefer 2.35 even if it's not technically correct. But I'm also unsure about 2.35 being acceptable for film out or for viewing on a widescreen tv --will the image appear slightly stretched or modified somehow because I chose to shoot 2.35:1 instead of the real standard, 2.39?

I wish someone like Barry could chime in about stretch mode in cam v.s. doing it in post. At first he agreed; but it could be that the camera stretches (some people use the term "uprez") before compression takes place, making it better than doing it all in post.

I have a DVX100a; I'd be happy to shoot some test footage on my free time to see which is better, and post them for you guys to see. But I'm unsure about the whole stretching in post thing, and don't know which program has the best stretching algorithm -- so somehow has to do that for me.

 

[William_Robinette]

I believe the issue can be viewed as such.

If you have the time/programs/patience, uprezzing 4:3 footage in post can give you a sharper picture,

but....

for almost every circumstance I can think of, squeeze mode does such a good job and saves so much time, it is almost a non issue. I don't think you will be seeing a difference on a TV screen anyhow....

 

[agcohn]

You say that Squeeze seems to give you the best 16:9 footage, but what about in conjunction with uprezzing? People who use Photo Zoom Pro seem to think that uprezzing 4:3 looks better than uprezzing from Squuezed (whether squeezed in cam or in post). So, assuming you will uprez, isn't 4:3 the best for 16:9?

What I really want to know is what's the best procedure for getting 2.35/2.39:1 assuming I will uprez eventually... I'd prefer 2.35 even if it's not technically correct. But I'm also unsure about 2.35 being acceptable for film out or for viewing on a widescreen tv --will the image appear slightly stretched or modified somehow because I chose to shoot 2.35:1 instead of the real standard, 2.39?

I wish someone like Barry could chime in about stretch mode in cam v.s. doing it in post. At first he agreed; but it could be that the camera stretches (some people use the term "uprez") before compression takes place, making it better than doing it all in post.

I have a DVX100a; I'd be happy to shoot some test footage on my free time to see which is better, and post them for you guys to see. But I'm unsure about the whole stretching in post thing, and don't know which program has the best stretching algorithm -- so somehow has to do that for me.

Whether you go for 2.35:1 or 2.39:1, the actual shape of the final image needs to be either 4:3 or 16:9 with letterbox bars on top to give you the wider aspect ratio.

You should shoot squeeze or use the anamorphic adapter, and then add letterbox later on. You could use the anamorphic adapter in conjunction w/ letterbox, but don't use it in conjunction w/ squeeze.

 

[iain.bason]

Right now, by the way, I'm really interested in knowing more technical details regarding the uncompressed output through the s-video. And, if it's not DV compressed, whether it will help me get a better key when doing blue- and greenscreen...

Alas, no. (Or anyway, I'm pretty sure not.) S-video is still sending NTSC or PAL, which have the same kind of color sub-sampling problem as DV.* I recently made a disc to help tune my projector and DVD player. Part of the image I put on the disc was a red-green-blue checkerboard pattern. When I was using the s-video connection to the projector, the pattern appeared gray. When I was using the component video connection I could see the colors.

As for the issue of letterbox vs. squeeze, I think the best we can do is conduct some double blind tests to see whether anyone can tell the difference. I'm not volunteering to do those tests, and in the absence of any better evidence all I can do is offer my best guess as a software engineer. I don't think that the image compression is going to be noticeably affected by the black bars. I don't think that software resampling is going to be noticeably better than hardware resampling. I think that it's possible that squeeze mode could do a better job because it's doing the resampling before compression, but I doubt it would make a big difference in most cases.

These are just educated guesses.

I know that squeeze mode is good enough for me, and the convenience factor is important.

----
*They're analog, so the details are of course completely different. However they do suffer from lower bandwidth for color information than for luminance, which is the essential problem.

 

[Rubix³]

Correct me if I'm wrong, but if you're outputting to anamorphic DVD and uprezzing in post, why shoot letterbox at all? The DVD player automatically adds the bars and corrects the image, no?