Convert 16:9 to 4:3 without Letterboxing?

[Kevin Janisch]

I'm doing a widescreen and fullScreen versions of a piece. I do not want letterbox for the fullscreen version, so I'm cropping the sides from 720 to 540 so I will have a 720x480 frame size with the .9 pixel aspect ratio (going from 1.2 pixel aspect ratio, XL2). So my procedure is I have a mask in my 16:9 composition so I can frame my shots. Once done I can then drop this 16:9 composition into a 4:3 .9 pixel aspect ratio composition, or just change the pixel aspect ratio in the 16:9 comp to .9 from 1.2.

Problem is, within After Effects, when I change my pixel aspect ratio (or drop the source comp into a .9 compostion) from 1.2 to .9, I'm losing sharpness (the interpretation of the 1.2 footage stays intact). So the sides get chopped to a correct 4:3 image, but the sharpness takes a hit, slightly noticeable upon zooming in. The reason for this is the change in pixel aspect ratio, I can see it when I toggle the change. It's going from displaying the image with the rectangular pixels to more square pixels (image is not changing itself, no stretching, squishing, etc) so there is softening.

My question is, has anybody else done a widescreen to fullscreen without letterboxing that didn't take a hit in sharpness? I can't seem to get around doing a pixel aspect ratio conversion. Tried going out 540x480 at 1.2 and of course it doesn't meet the DVD specification. Thanks.

Kevin

[Chris Barcellos]

Question: Is it that way in the actual render, or just in project preview monitor?

[Kevin Janisch]

The same loss of sharpness is there when Photoshop files are exported. Thanks.

[Chris Barcellos]

Kevin: My theory:


What may be happening is that you are indeed lowering resolution. With the 16:9, from what I understand about DV, you still have 720 by 480 pixel count. The pixel still contains the same amount of information, but is wider than tall. If that is the case, then when you cut off sides, your are lowering the pixel information. That would mean the editing program has to fill in information on the 720 wide side to actually make up for the lost information.

[Kevin Janisch]

Chris,

Thank you for your thoughts. Yes, I lose resolution when I crop the sides of the 16:9 image to 4:3. A 16:9 or 4:3 image is both 720 x 480, the only difference is that the 16:9 footage is anamorphic and must be stretched to 1.2 pixel aspect ratio vs the 4:3 .9 aspect ratio. So yes, 16:9 have rectangular pixels vs 4:3's more square, but the loss of sharpness is coming from the pixel aspect ratio conversion. More pixels have to be used when in .9 aspect ratio to fill the same frame that was 1.2. The image is not being stretched whatsoever, only the pixel ratio is being changed. My 16:9 image is merely having the sides chopped off, much like, but not exactly what you see in an XL2 viewfinder when you toggle 16:9 and 4:3 mode.

So what I'm doing is setting a mask on my 16:9 image in a AE composition with 1.2 pixel ratio that puts black bars over 90 pixels on each side for framing. Then I go and change the composition pixel aspect ratio to .9 which chops the sides off. I believe the conversion from 1.2 pixel aspect ratio to .9 is what is causing the slight softening. I'm just wondering if there is another way to do this that maintains the sharpness. Check out the photoshop file to see what I'm talking about. Zoom in and toggle the top layer on and off to see the difference.

http://www.surreelelements.com/conversion.zip

[David Jimerson]

You may find you have better results by cropping the top and bottom of the picture, rather than the sides. It's much better-suited to the way pixel aspect ratios work.

Yes, you will necessarily lose resolution. It's mathematically impossible not to!

[Kevin Janisch]

David,

Thank you for your response. Would you care to elaborate a little as I'm not following how cropping the top and bottom of a 16:9 frame to achieve a 4:3 frame would have better resolution than cropping the sides.

Taking a hit in resolution is acceptable if it is the only way to do it. I'm just making sure I'm not missing something. Thanks.

Kevin

[David Jimerson]

OK, here's where I embarrass myself.

Never mind; I had it totally backwards. Foot-in-mouth post at the end of a Friday of a long week. Sorry!

[Kevin Janisch]

No problem, it has been a long week. Thanks.

Kevin

[Kevin Janisch]

This same phenomenon exists within Premiere Pro as well in terms of a slight loss of sharpness. So if you create a 16:9 Premiere Pro project and later import it into a 4:3 project to get a full screen version you'll see very slight softening when compared to the same 16:9 frame. My guess is going from rectangular pixels (16:9) to square ones (.9 4:3), more square pixels are needed to interpret the image than the rectangular ones.

[Robert M Wright]

If the target audience will be viewing on an average living room, standard definition CRT television, there isn't likely to be any perceived loss in sharpness (an average standard def TV just doesn't display that many lines of resolution).

[Kevin Janisch]

Robert,

Thank you for putting things into perspective. There is less "perceived" sharpness loss when viewing the test DVD on a CRT 27" vs my 42" LCD, but still slightly noticeable when watched back to back with the original 16:9, otherwise negligeable for most. It just pains me that at every stage of production everything is done to ensure the highest quality product, and then only to take a slight hit in sharpness like that. Just wanted to be sure I wasn't missing something. Thank you all.

Kevin

[Dufu Syte]

Originally Posted by David Jimerson

You may find you have better results by cropping the top and bottom of the picture, rather than the sides.

Actually this might be the best solution depending on how the actual chip is made. Some chips are physically 4:3, and when you shoot in 16:9 it is actually just chopping off the top and bottom for you, and you actually get less pixels than if you shoot originally in 4:3. In this case, your best bet is to shoot originally in 4:3 to get all the pixels of the chip (and it provides your 4:3 final version) and then chop off the top and bottom to create your 16:9 final version. If the chip itself is 4:3, and you shoot in 16:9, you are losing pixels right off the bat, and then when you crop off the sides of your 16:9 image to produce a 4:3 version, your 4:3 version is very whittled down.

Sorry for the rambling. Bear in mind this only applies if the chip itself is 4:3 native. Some still camera chips are this way.