Cropping + resolution
 

I've read a few posts which mention that when you crop footage (eg cropped in FCP to make footage widescreen format) you lose resolution. Is this true and if so could someone please explain why this happens and is it noticeable when viewed on a Television. I use Photoshop a lot (with still images) and when you crop an image it doesn't affect the resolution...

Sorry if this has been covered before (I did a search but could not see any relevant info. Thanks.

Justin

A vector based graphics app will allow an image to be resized by redefining the pixel map...providing interpolated pixels where necessary, so that an enlarged image contains, proportionately, more pixels. NLE's are not vector based, so that when an image is cropped and subsequently stretched to fit the 740x480 format of the fixed DV frame size, the pixel count is reduced. Since pixel count relates directly to rez, the rez is also reduced. Consider what happens when you do a huge zoom in...the picture gets real blocky and unrecognizeable. That's not an instantaneous loss of picture quality, but, begins every time you zoom in.

If you're zooming/cropping a fixed image, it pays in rez to export the image as a png bmp or tif file to a vector based graphics program, crop and zoom, then re-import the image into your NLE. Don't use jpg for export/import because jpeg looses quality every time it's recompressed.

Sorry, maybe I'm being stupid but I still don't understand. Why is the image 'subsequently stretched'!!!? The image should be *exactly* the same size horizontally - the only difference should be vertically which would have black bars in place of the bits that were cropped off the top and bottom.

ahhh...I'm sorry, Justin, I missed that your question was IRT wide screen format. In fact, with Vegas Video, when I make a wide screen format video, I use a mask that simply blots out the letterbox...no cropping involved. I believe that part of the answer to your question lies with the fact that wide screen format actually uses a different pixel shape. That is to say the height to width ratio of each pixel is different.

I'm sure someone in here has the correct answer to your question.

Perhaps resolution loss might be a word that is a bit misplaced
here. You are the first to notice it though. People are talking
about resolution loss because you are throwing away part
of the images to replace it with black bars. Ofcourse your
resolution stays exactly the same, but you loose some portion
of your picture. Perhaps it should be called picture loss?

Why everyone is felling a "loss" is because DV has a fixed
bandwidth (3.6 mb/s -> 1:5 compression). Now if you throw
away a part of the picture you are in sense throwing away
bandwidth too which might have been better used for more
detail in the remaining piece of the picture. Because if there
is a lot of detail and the DV compression algorithm cannot
fit it into the 3.6 mb/s it will scale down on quality. This is
way a lot of people would want to have a true anamorphic
adapter on there lense. This would utilise the full available
bandwidth of the DV for the *actual image* that you are going
to see, not only a part of it.

I hope this has explained it a bit better. So to recoup, in theory,
yes, you are not loosing resolution. You are loosing a bit of
your picture though which in turn leaves you with less vertical
lines to show the picture you wanted. This can only be "increased"
with an anamorphic adapter (or true 16:9 progressive CCD
chips) when the footage is also shown on a 16:9 widescreen
television/projector. If not, you have the "picture loss" anyhow.

If you have some more questions about this or still don't
understand it lemme know and I'll try write it down clearer.

Thanks Rob - that's starting to make sense.

So other than the black bars (which are not a problem to me) is there any noticeable difference in the picture quality?

no you wont lose picture quality by placing black solids upon your picture.
like shooting 16:9 mode and scaling down to letterboxed look. This is the same thing.

though niether is true 16:9.

There is a big difference between 16x9 and Letterboxing.

If you screen a fake/cropped NTSC image on a true 16x9 TV you will lack 33% of line resolution compared to a real/true 16x9 image. A 4x3 image cropped to 16x9 holds about 322 lines of vertical resolution compared to the original 480 in a real 16x9 image.

ahhhh, thanx Martin...that's the answer.

Even after reading the various articles in the Watchdog I'm still sort of unclear if using the 16x9 mode on the XL1 is better than cropping the 4x3 image as described here? I know that a camera with native 16x9 chips is the best way to go of course, and after that an anamorphic lens, but otherwise which of those other methods is the preferred mode to use for eventual letterboxed presentation?

so cropping a 4x3 image is somewhat better than utilizing the 16x9
mode on an XL1? as in vertical resoultion.

though the 16x9 resultion on an XL1 is 640x360....360 pixels of vertical line resolution...and a cropped 4x3 is supposably 322? at 720 horizontal?

i see the resolution difference in aspect ratio, but not sure if theres any LOSS between the two.

From what I've read the 16:9 mode on the camera merely 'stretches' the 4:3 image horizontally to make it widescreen - yuk!!! Not an option!!!

So, is this correct - the cropped (letterboxed) footage would look fine on a regular TV but on an actual Widescreen TV the image would not look as good (by 33%)? Is it a visually noticeable difference from back on a sofa with the TV in the corner of a room or would you have to go right up to the screen and examine it?

The number of lines for your picture remains the same whether
you go with 16:9 in camera or letterboxing.... to make the top
3 in quality:

1. anamorphic attachment (adapter)
2. 16:9 in camera
3. 16:9 / letterboxing in post

Why is 16:9 in camera better? Because you use the full
DV bandwidth to compress the actual image you are going to
be seeing. If you do this in post you are throwing away
DV bandwidth, so your quality might be (depends on the
image you are taping) lower. There is however a benefit
to going with letterboxing in post: you can still move the image
up and down to get the "perfect" framing. If you shoot 16:9
with the camera you have no choice lateron.

Another thing to consider is, if you are shooting 16:9 how
are you going to display it on the web and on a normal
TV set? For the web you need to make an additional output
where you stretch the source back. For a normal TV you'd also
have to do this unless you are going down to DVD. With a DVD
you can indicate that the source is 16:9 widescreen. The DVD
player will then letterbox it on a normal TV if the player has
been setup correctly.

I prefer to letterbox since the quality gain isn't that much
according to my eye (I haven't seen the quality with an
anamorphic attachment yet). I prefer to have the option in
post to move the image up and down to get my perfect
framing.

Rob,

That was the same conclusion that the director of a project I shot a while back arrived at, after doing some tests. The short was shown on a 30 ft wide screen at Sony (I imagine it was a state of the art projector, don't know which one) and the image was sparkling. Those few dozen missing lines didn't seem to be a problem.

I also ran several tests using the XL1s in 16:9 mode and letterboxing in post in a 4:3 frame and shooting in 4:3 and letterboxing with a mask. I then looked at the footage on a 63" 16:9 RP TV in Cinema-Wide mode. The footage shot in 4:3 was much sharper and filled the screen perfectly w/o distortion. I then ran the test again on a 36" 4:3 TV. Again the masked 4:3 footage was the clear winner.