squeeze or letterbox

Jack Daniel Stanley said:
FACT: Squeeze in camera and stretch in post use the same # of lines in the middle area. They travel an almost identical patch in the camera.
Click to expand...

Not a fact. The image content is the same, but the amount of image data is not.

Without a doubt, post crop-and-zoom uses the center 360 lines and vertically scales it to 480 lines.

In camera, the image sensor has a raw image of 720x576 even on the NTSC model(according to published the camera specs for both PAL and NTSC models), and it uses the center ~430 lines to vertically scale to 480 lines. That about 20% more image data to work with, thus a better uprez.

Think of it this way, even when creating a normal 4:3 image, the NTSC camera is actually downscaling 720x576 to 720x480 to create the image.

I didn't see Barry comment on this aspect, and would like to see his thoughts on this.

There was a question about progressive vs. interlaced. I'm talking exclusively about progressive and I use 24pA for that. Why is anyone shooting interlaced? (Just kidding)

HYPOTHESIS: The camera has less processing power and my not be able to do as good a job stretching as your computer.
Click to expand...

Yes, a hypothesis. The processing power and method is really an unknown, except that the camera has dedicated image-processing DSP to do this stuff. It's tough to compare it head to head with what a computer can do. It may in many instances surpass what a computer can do (except that the functionality is locked in and can't be improved/modfiied).

There are some really good algorithms out there for post-processing, but even these have trade-offs and certain techniques work better or worse with certain types of images and motion. The camera, of course, is locked into whatever it's doing.

Well, I should have some time Friday to shoot some tests. What I'd like to do, to keep the process open, is to provide the same shot in both 4:3 and SQUEEZE, uncompressed frame grabs from DV footage. Then, anyone can download the 4:3 image and do an uprez to 16:9 by any method of their choosing, and we can compare those results to the original pure SQUEEZE image.
 
Joshua Provost said:
In camera, the image sensor has a raw image of 720x576 even on the NTSC model(according to published the camera specs for both PAL and NTSC models
Click to expand...
Not true. The PAL version claims 470,000 pixels, with 440,000 active. The NTSC version claims 410,000 total pixels with 380,000 active.
 
I have not read every word of this thread, but most of it...and for what its worth..

I have shot about 8 short films in 4x3 and then simply letterboxed in post.

The other day, we decided to take a DVX100B out and shoot two scenes in squeeze mode. I brought them in on an anamorphic timeline and put them together.

I took a couple sections of my films shot in 4x3 ,brouoght them in an anamorphic timeline and scaled them up (whatever percent it is) to fill the 16X9 area.

Playing back on the bigscreen, the winner, BY FAR in quality was the scenes shot in squeeze mode. The images that were 4x3, that i had to "zoom in on" the timeline looked a bit blurry and you could definitely tell there was a loss. But the images shot in squeeze mode looked great on the big screen.

Now, not to confuse things, but these same movies shot in 4x3 look MUCH better when I simply keep them 4x3 and use the bigscreen tv remote to "zoom in " on the footage to fill the whole screen as opposed to bringing the 4x3 footage into an anamorphic timeline in FCP and zooming in on it in editing, so it is now a 16x9 image...In fact, these films look GREAT when i do it that way. Simply zoom in on them with the tv remote. They really hold up well.. much better than when i zoom in on the timeline in FCP......

I hope im making sense..

as a result, I am filming a movie this weekend and I too am having a hard time deciding which one i want to film in . But am CERTAIN what i will NOT do..i will NOT shoot 4x3 , then bring it in the timeline and zoom in on the footage. If i shoot 4x3 , its staying 4x3 (with a letterbox)and i will use the tv remote if i want to fill the screen. It looks WAY better, and i have tested multiple films this way.

Just my two cents...
 
Last edited:
wow. this is very interesting. I always though shooting letterboxed was the same as cropping in post without the option to adjust the framing, but when I plug my camera into a regular SD tv and switch between normal and letterboxed mode, the center of the screen does get slightly clearer in letterboxed mode. Its hard to tell with the squeeze mode cause it is stretched. Maybe I will start filming in letterboxed mode and when playing on a plasma screen, do the zoom crop on the tv. Another consideration is that any color correction or effects in post will apply to the black bars as well, and this can be distracting, so you still might want to put a mask over the footage in post. Funny after all these years this is still kinda unresolved.
 
It seems as though the difference is so minor, that it comes down to personal preference. My experience thru post production letterboxing and reframing turned out uglyer footage than just sqeeze, and proper framing from the begining. So no matter what is said in terms of micro numbers and what not, I use squeeze. Its easier to me from start to finish, and to me it looks cleaner.
 
Please help me clear up some confusion here.

Is the letterbox mode built into the DVX, in fact, the squeeze mode I keep hearing about?

I shot a squeezed image on a GL2 the other day and it stretched back to 16:9 just fine in FCP, but when I apply the letterbox bars on the DVX, I don't see any stretching of the image taking place. Is the squeeze function buried somewhere else in the menus and I'm just missing it?

:dankk2:
 
my gosh i just read this thread from beginning to end and my head is spinning...

Im more confused than ever now as to whether or not to shoot squeeze or letterbox this weekend and i better decide quick!!!!:crybaby: but at least i have it narrowed down to those two!!!

one thing that is making me lean toward squeeze mode though is the pain in the but of getting the software and then hoping i have enough power and time to uprez with photozoom...

Thanks to all for your input here...
 
Last edited:
Barry_Green said:
Don't you have an HVX? You can do the same test, just compare letterbox against 4:3 and examine the central area. DV is smart enough to reallocate bits from understressed blocks to those blocks who need the bits.
Click to expand...

Good to hear that it does take advantage (though in most of my work, I prefer the reframing). I'll search down the pics! As to the HVX test - does it really letterbox the "stupid" way? I wasn't sure... but hoped it letterboxed by downrezzing the full sensor (which would invalidate the comparison).
 
ullanta said:
I wasn't sure... but hoped it letterboxed by downrezzing the full sensor (which would invalidate the comparison).
Click to expand...
All standard-def is done by downrezzing the full sensor, but 4:3 is done by downrezzing a center extraction. So it takes the 4:3 chunk out of the 1080p frame and downrezzes that to a 4:3 SD frame, then adds black bars over it.
 
Barry_Green said:
All standard-def is done by downrezzing the full sensor, but 4:3 is done by downrezzing a center extraction. So it takes the 4:3 chunk out of the 1080p frame and downrezzes that to a 4:3 SD frame, then adds black bars over it.
Click to expand...

Yes, that's what I meant by the "stupid way" - blech!
 
Smart way would be downrez the full 16x9 sensor to get a 16x9 image, then add black bars around it - that is, get improved sharpness/noise reduction by starting with the largest image possible, rather than using a subset of the already smaller 4:3 area.

Optimistically, 4:3 uses 75% of the imaging area
Letterboxing that uses about 75% of THAT image area, or about 56% of the full sensor.

SQUEEZE uses the whole sensor (on the HVX) downrezzed to SD anamorphic; I had hoped that HD letterboxing would downrezz the full sensor to SD as well, rather than using only half (56%) of the sensor.

(I say "optimisitically" because I think I read somewhere that 4:3 may use a 480-pixel-tall subset of the sensor, rather than downrezz the 540 full sensor image. If that's the case, the percentage used in 4:3 and letterbox modes will be that much less).


(Also, I guess there are SOME benefits to the small image area... once, when I was very far away in a concert hall and needed more than the 13x lens magnification, I switched from Squeeze to 4:3 to get an effectively closer-zoomed image, then changed it to 16:9 in post... )
 
Last edited:
You must log in or register to reply here.
Back
Top