Converting .VOB or MPEG2 from 16:9 to 4:3 Letterbox

[g1275]

I am trying to put together a bunch of DVD clips onto one DVD but I'm having trouble keeping everything 4:3. I am using Womble MPEG Video Wizard to edit everything and it was working fine, but it does not maintain aspect ratio with the 16:9 files. It maintains aspect ratio in the monitor, but when it encodes it stretches the image to 4:3. I'm trying to figure out how to convert a 16:9 VOB or MPEG2 file into 4:3 while maintaining aspect ratio and without any loss of quality. I am aware I have to reencode the file but how do I reencode a MPEG2 file without losing quality. I don't want to have to convert to AVI and then back to MPEG2 to fix the aspect ratio. Also is there a good program to add text to MPEG files without converting it to DV? Wombles text editor is not very good.

Thank you ahead of time

[guns1inger]

When you export the video, click on the Video tab, then on the Expert button. Toward the bottom of the dialogue you should be able to set the PAR back to 9/16 (their terminology).

You cannot convert a 16:9 video to 4:3 without quality loss. To begin with you have to resize the image down but up to 30% to make it fit within a 4:3 frame, then re-encode everything. Womble cannot do this for you anyway. It can encode to 4:3, but all it is really doing is adding 4:3 flags so you get a stretched image.

If you must convert perfectly good 16:9 source to lower quality letterboxed 4:3 (and it will be lower quality, it simply cannot be avoided with so much less data), I would use DGIndex and Avisynth to resize and add the borders, then serve this to HCEnc for encoding. You can also add text in Avisynth.

[g1275]

Thank you for your post, but how exactly would I go about using the 3 programs to convert to 4:3? How do you add text with AviSynth?

Thank you again

[guns1inger]

DGIndex creates a .d2v file which can be opened by avisynth using Mpeg2Source(). If you haven't used Avisynth then sit down and read through the basics at the Avisynth Wiki pages. It is a text based frameserver (no graphical front-end) that can be daunting to begin with. However it is extremely powerful, has filters to do pretty much anything you need in video, and is fast for basic processing such as cropping and resizing.

This is a typical Avisynth Script to convert a 16:9 NTSC DVD to a 4:3 Letterbox NTSC DVD

Mpeg2Source("full path to your video")
LanczosResize(720,368,0,0,720,480)
AddBorders(0,56,0,56)
ConvertToYUY2()

You would replace the path with the full path to your .d2v file, and save this as a text file with the extension .avs

If you then load this into virtualdubmpeg2 you will be able to see the effect the script has. If you are happy you can then load the .avs into your encoder and encode it.

A couple of things to note. This script doesn't handle audio at all. In your case that should not be an issue as you really want to demux the audio and use it as is when you re-author. Also, I have deliberately left out any attempt to put text on the image as I don't know what you are really after there. If it is a few lines of simple text at a couple of points in the video then you can use the Subtitle statement in avisynth. It is well covered in the on-line manual at the Wiki.

[g1275]

It did work, but there is a really bad ghosting problem. I had profile set to best. Any way to fix this in HCEnc or a script that would help in Avisynth? BTW I had to change "ConvertToYUY2()" to "ConvertToYV12()" and add "LoadPlugin("DGDecode.dll")" to get it to work. Could that be the problem?

Thanks for your help

[manono]

Hi-

It did work, but there is a really bad ghosting problem.

Are these clips interlaced? Or did you make the D2V project file in DGIndex using the (default) Honor Pulldown Flags? If it's film, make the project file using Forced Film. Then (assuming no other complications), there won't be any ghosting as a result of resizing. Read the included QuickStart guide for more information, as well as the other 2 docs. If the source is interlaced, or if you want to use Honor Pulldown Flags in DGIndex for some reason, you'll have to resize differently, as you can 't just resize interlaced material like that. This is not meant as any criticism to guns1inger, as he knows full well how to resize interlaced material, and his script was just meant as an example.

BTW I had to change "ConvertToYUY2()" to "ConvertToYV12()" and add "LoadPlugin("DGDecode.dll")" to get it to work. Could that be the problem?

No, that has nothing to do with the problem. And this is the first mention of HCEnc. It requires YV12, while most of the others accept YUY2 (except for TMPGEnc which accepts RGB24). And good you figured out about loading the DGDecode.dll.

If possible, maybe upload a 10 second sample of the untouched source for more definitive advice.

[g1275]

I didn't force film, and it's progressive, I'm sure that was the problem.

Thanks