PAL 4:3 to 16:9 but black L/R stripes instead of cropping

[jmwismer]

Hi there,
I switched to 16:9 taping and editing, and I am left with a short footage in 4:3 which I need to convert to 16:9 to include with the other 16:9 footage.
I have read a bit about converting by cropping.
Instead of chopping off heads, I would not mind converting by adding black stips on the sides.
What tool shall I use?
How wide should the stripes be?
thx,
jm

[manono]

Hi-

What tool shall I use?

You can use Virtual Dub and frameserve to your encoder I suppose, but better would be to use AviSynth for frameserving.

How wide should the stripes be?

You're a little short on the specifics of your source video - format and resolution. I'll make a guess and say you have a 4:3 720x576 PAL MPEG of some sort and you want to add black to the sides to make a 16:9 PAL DVD.

Use FitCD to get the values for this:

LanczosResize(528,576)
AddBorders(96,0,96,0)

You can resize from 720x576 to 528x576 and add 96 columns of black pixels to both the right and left. That will give you a 720x576 video ready to be encoded for 16:9.

[jagabo]

I switched to 16:9 taping and editing, and I am left with a short footage in 4:3 which I need to convert to 16:9 to include with the other 16:9 footage.
I have read a bit about converting by cropping.
Instead of chopping off heads, I would not mind converting by adding black stips on the sides.

Most video editing packages will do this automatically when importing 4:3 footage into a 16:9 project.

[jmwismer]

Thank you Manono,
yes, PAL 720x528, but AVI format. Would that mean another tool?
I looked into FitCD but must confess this is a bit of a stretch. The commands you describe to resize / add borders, are those to be given in somewhere in FitCD? I suppose since I am AVI it is no longer relevant, right?
thx,
jm

[jmwismer]

jagabo: Sudio 9 will not... (I am on the consumer side, not a pro at all)

[jagabo]

It's been a very long time since I used Pinnacle Studio. Try right clicking on the 4:3 video in the timeline and look for something like source video properties. Change the AR to 4:3.

[Midzuki]

@ jmwismer:

first, resize from 720x576 to 540x576;

second, add 90 pixels to the left and 90 pixels to the right;

[jmwismer]

@Jagabo:
I can indeed select 4:3 or 16:9 in Studio, but it just sets the flag. In my case, I can hence use my 4:3 footage in my 16:9 movie but the image is all stretched. Studio will not do the resizing exercise.

@all:
I tried a bit the image reformatting as you all suggested. I did the AVIsynch reformatting below:
-cropping:
Crop(0,68,0,-76)
LanczosResize(720,576)
SeparateFields()
Weave()
ConvertToYUY2(Interlaced=True)

-side stripes:
LanczosResize(528,576)
AddBorders(96,0,96,0)
SeparateFields()
Weave()
ConvertToYUY2(Interlaced=True)

, but somehow I must do something wrong in the re-encoding with VirtualDUB. The created new AVI does have the right 16:9 look (squeezed when I look at it on my PC), but then Studio will not let me use it (different format). I can't neither select 4:3 or 16:9, bith options are greyed out. Also the size of the file is much larger than the size of the original file.

How can I juts reencode from VirtualDUB so that it keeps the same initial encoding but with just the resizing changed? I did "Save as AVI".

[manono]

Hi-

I'm not sure why you added in the Separatefields/Weave commands, as the net effect is nothing. They're a waste of time. Nor am I quite sure why you added the ConvertToYUY2(Interlaced=True) line for VDub. Maybe you have your reasons. I sometimes do that when saving to Lagarith, but that's because it then goes to CCE for final encoding. Anyway, yes, you can reencode to AVI with just the crop and resize lines. Best, I think, would be to encode to a lossless codec. I use Lagarith. While the size will become larger than the source video, it'll be much smaller than saving as uncompressed.

I've never used Pinnacle Studio, so I can't help there.

[jagabo]

Crop(0,68,0,-76)
LanczosResize(720,576)
SeparateFields()
Weave()
ConvertToYUY2(Interlaced=True)

As manono pointed out, using SeparateFields() followed immediatly by Weave() is pointless. If your video is interlaced you need to SeparateFields() before resizing, then Weave() afterwards. Keep in mind, after SeparateFields() your video is half as tall as the original. So the sequence would look like:

Crop(0,68,-0,-76)
SeparateFields() #432 line frames are separated into two 216 line fields
LanczosResize(720,288)
Weave() #two 288 line fields are woven back into a 576 line frame
ConvertToYUY2(Interlaced=True)

Also be careful about cropping interlaced video. If you crop an odd number of scanlines off the top of the frame (possible with YUY2 and RGB) you will reverse the field order.

I don't know if Studio will accept it, but rather than opening the AVS script in VirtualDub and reencoding, try opening the AVS script directly in Studio.

Manono, one possible reason to use ConvertToYUY2(interlaced=true) for VirtualDub: VirtualDub doesn't handle interlaced YV12 from AviSynth properly. It converts the YV12 frames as if they are non-interlaced (the chroma components aren't handled properly):



That's a 4x enlargement of a grey bar moving over a reddish background. On the left is the proper handling, on the right is VirtualDub's incorrect handling of interlaced YV12.

[manono]

If your video is interlaced you need to SeparateFields() before resizing

Good catch. I kind of skipped over the first script as he said specifically earlier he wanted to convert to 16:9 by adding black bars to the left and right rather than by cropping from the top and bottom. As you say, once he begins cropping from the top and bottom followed by a resize, he has to be very careful how and in what order he does it when interlaced video is involved.

[jmwismer]

I picked this from another post where it sounded like just cropping and resizing made the image look jerky. According to that post, the proposed sequence solved it. I took it at par value, but would be most happy not to loose time on unnecessary computing which is always very time consuming.

I tried to open the AVS with Studio directly but it won't take it.

I guess I am close to solving my problem: re-encoding the re-sized image so that the final AVI format and size is the same as the initial one. Would VirtulDUB do so?

thx again,
jm

[Midzuki]

I think that a screenshot from your 4:3 footage would be helpful.

[manono]

Yes, VDub can encode to AVI for you. Open the script. Set Video to Fast Recompress. Go Video->Compression and choose and set up a codec. As I said before, you'll probably want to install and use a lossless codec if the AVI is going to be further edited and/or reencoded again afterwards. Otherwise VDub will save it uncompressed at a huge file size.

Manono, one possible reason to use ConvertToYUY2(interlaced=true) for VirtualDub: VirtualDub doesn't handle interlaced YV12 from AviSynth properly. It converts the YV12 frames as if they are non-interlaced (the chroma components aren't handled properly):

Like I said, when using Lagarith prior to sending to CCE for final encoding, I convert to YUY2, but only because I use CCE afterwards. I guess I accidently got lucky, as I wasn't aware of any VDub problem with interlaced YV12 sources. As perhaps you've figured out by now, there's a huge gap in my video encoding knowledge of anything to do with colorspaces and interlacing (and DV, and MJPEG). And in spite of your expert guidance, I don't seem to be getting any better. I must have some kind of a mental block about the subject.

[jmwismer]

Thank you guys, got it to work

You mentioned lossles codecs: any suggestion? I have a list of half a dozen, but no idea about how they behave. I picked one which save a file about the same size as the original size, by lack of more intelligence...

[guns1inger]

I like Lagarith, but Huffyuv is also good. Both will give you files much larger than your source. If you really only have a 20 MB HDD (and I'll assume you meant to type 20 GB when you filled in you specs) then it had better be a short video.

[jagabo]

Here's a 1 minute 720x480 29.97 fps video clip compressed (or not) in several different formats:

codec:               AVI file size (bytes)
------------------------------------------
DV:                  227,502,080
MSU Lossless:        356,954,776
Lagarith:            403,314,520
HuffYUV:             526,933,128
uncompressed YUY2: 1,255,344,164
uncompressed RGB:  1,877,078,564

[ronnylov]

Do you really need the vertical cropping? The vertical cropping would shop off the heads which you did not want to do.

Here's what I would do in avisynth:
Crop (8,0,704,576) # Crop the sides a little bit before horizontal resizing
Spline36Resize(528,576) # Resize width to 528 (using mod 16 for MPEG)
AddBorders(96,0,96,0) # 96 pixels black borders on both sides

And then encode everything as anamorphic 16:9 widescreen. You will loose some horisontal resolution compared to a 4:3 project but you keep the full vertical resolution on a 16:9 TV. But if you watch this on a 4:3 TV with letterboxed 16:9 setting on a DVD player you'll get both vertical and horisontal black borders (the 4:3 will be in a small window centered on the screen). This may be annoying for 4:3 TV owners but they may change the setting to pan/scan or fullscreen but then they loose the sides of the 16:9 parts...

Another alternative would be making everything in 4:3 format. Crop off the sides of the 16:9 parts to 540x576 and enlarge to 720x576 and keep the 4:3 video untouched. You'll loose something whatever method you choose...