Is SD true 16:9?

M

matty mcphail

Member
Hi there,

We are currently planning a shoot on the HVX (PAL) and I am getting mixed feedback about the 16:9 format in mini DV. We may have to deliver in 2 aspect ratios, 16:9 and 4:3.

My understanding was that if we shot native 16:9 we could just drop off the edges and have a true 4:3 format for delivery. (assuming we shoot 4:3 safe). Some people are telling me that the 16:9 format when shooting onto DV tape will in fact be just 4:3 with a crop so if we wanted to deliver 4:3 we would instead have to blow the image up (obviously loosing quality).

Does this make any sense???

Any feedback would be greatly appreciated.

Matt
 
My understanding was that if we shot native 16:9 we could just drop off the edges and have a true 4:3 format for delivery
Click to expand...
Correct.

However, 16:9 is using anamorphic (non-square) pixels, it is recorded using the same number of pixels as 4:3. A 16:9 image can easily be cropped to 4:3 and re-encoded to square pixels.
 
matty mcphail said:
We may have to deliver in 2 aspect ratios, 16:9 and 4:3.
Click to expand...
There is no argument on this one - shoot 16:9 and then crop for the 4:3 version.

The people advising you otherwise are, I suspect, thinking of older cameras with native 4:3 chips, which may indeed give a cropped 16:9 picture, made even worse if you then derived 4:3 back from that. None of that is true for an HVX, or any HDV model, for that matter.

Incidentally, none of the variants of DV - 16:9, 4:3, and in NTSC or PAL - have square pixels.
 
Seems logical that even though the HVX has 16x9 native CCDs, there will be some apparent loss in image sharpness/quality by converting from 16:9 to 4:3. Same image shot in 4:3 will be sharper.
 
Segarza said:
Seems logical that even though the HVX has 16x9 native CCDs, there will be some apparent loss in image sharpness/quality by converting from 16:9 to 4:3. Same image shot in 4:3 will be sharper.
Click to expand...

Why would you say that? All you're doing by "converting" to 4:3 is cropping yourself rather than having the camera do it for you with black bars.
 
I would go one step further and offer this: Why shoot SD at all? Even if you have to deliver in an SD format you can downconvert in post, which will offer a better quality image.

Shoot HD16:9 (4:3 safe), then downconvert.
 
You're not only cropping, you're changing pixel aspect ratios... computers work with square pixels, so converting between pixel aspect ratios = scaling. You're changing from a 1.2 pixel aspect ratio to a 0.9. So you're scaling... which if one form of interpolation. If we were dealing with square pixels, I'd totally agree there was no loss extracting a 4:3 square pixel image from a 16:9 square pixel image.

There's also compression to deal with... 16x9 image has already been compressed once it's on tape or p2, so you'd have to convert to 4x3 and output uncompressed to prevent concatenation.
 
ncje said:
I would go one step further and offer this: Why shoot SD at all? Even if you have to deliver in an SD format you can downconvert in post, which will offer a better quality image.

Shoot HD16:9 (4:3 safe), then downconvert.
Click to expand...

That would be cool! Unfortunately we are not yey set up to deal with the workflow of HD. We are shooting a second series if 13 1/2 hour episodes which are VFX heavy and don't have the budget to fork out for the infrastructure. (hopefully on the next one!)
 
I agree with ncje...just mind the compression... you probably will benefit because there'd be some "oversampling" going on.
 
It's more of an aesthetic decision... I prefer letterboxing over pan & scan even if the image was protected for 4:3.
 
harddrive said:
There is no argument on this one - shoot 16:9 and then crop for the 4:3 version.
Click to expand...

If HD is not an option then this is your answer. You want to keep the highest resolution possible. Segarza does make a good point however.
 
If you shoot SD in the HVX and place the image in an SD 4:3 timeline (of the matching frame rate) your will crop the sides not the top or bottom. This allows for pan and scan to place the key area of the image in frame. Remember that SD 4:3 and SD 16:9 are both 720x480 pixels. If you shoot 16:9 and protect 4:3 you can output in both with out downrezing, uprezing, or adjusting PAR (pixel Aspect Ratio). Unlike the DVX the HVX uses native 16:9 chips not 4:3 which must be manipulated using the anamorphic adaptor or squeeze mode, or uprezsing in post to achieve 16:9.

Good shooting

S Cannon
 
Last edited:
How do you get by placing a 1.2 PAR image in a 0.9 PAR timeline with no scaling of any kind? (FCP, AE or any other NLE does it for you)

720x480 (1.2) = 720x480 (0.9) ... just by cropping?? No can do.
You'd have to convert to square pixels 720x480 (1.2) = 864x480 (1) then crop to 648x480 and after that scale to 720x480... there are other paths...
 
Last edited:
Segarza said:
How do you get by placing a 1.2 PAR image in a 0.9 PAR timeline with no scaling of any kind? (FCP, AE or any other NLE does it for you)
...
Click to expand...
True - scaling must be involved, whichever way you go, there's no question about that.

But going 16:9 to 4:3 involves cutting off pixels at either side (going from 720 overall to 12/16 that - 540) and straightforward one dimensional rescaling back up to 720.

But going 4:3 to 16:9 is not the same process turned through 90º because you have to contend with interlace. After vertically cropping the unwanted pixels, restoring the 576 lines is not a simple upscale as in the example above, but rather a process more akin to standard conversion.

If we were dealing with a progressive system it would be as broad as long - the derived version would be a certain amount worse than the original, whichever way round. But for PAL or NTSC that's not the case, interlace makes the image asymmetrical, and for a 16:9 chipped camera, originating 16:9 and deriving the 4:3 is by far the most satisfactory method. Lens angle issues also are more satisfactory doing it this way round.
 
You must log in or register to reply here.
Back
Top