Doctor M's ÜberGuide for -Full- PAL to NTSC DVD Conversion v2.0

ÜberGuide for _Full_ DVD PAL to NTSC Conversion (v2)
by Doctor M

Disclaimer:
This guide is for the intermediate and advanced user.  It assumes you know the basics of audio/video and how to use some of the more common programs.
If you need to catch up, www.Doom9.org and www.VideoHelp.com are a wealth of knowledge and guides.

Software:
DVDRemake Pro (optional (pay software)) http://www.dimadsoft.com/dvdremakepro/index.php
DVD Shrink (optional)  http://www.videohelp.com/tools/DVD_Shrink
PGCDemux http://download.videohelp.com/jsoto/dvdtools.htm
Citizen’s Aspect Ratio Calculator Tool http://www.haku.co.uk/ARCT.html
DGMPGDec (DGIndex/DGDecode) http://neuron2.net/dgmpgdec/dgmpgdec.html
AVISynth http://sourceforge.net/project/showfiles.php?group_id=57023
VideoHelp's bitrate calculator http://www.videohelp.com/tools/VideoHelp_Bitrate_calculator
CCE (Cinema Craft Encoder)(pay software) http://sourceforge.net/project/showfiles.php?group_id=57023
DGPulldown http://neuron2.net/dgpulldown/dgpulldown.html
BeLight http://www.videohelp.com/tools/BeLight
Sonic Foundry Soft Encode (Deadware) http://www.digital-digest.com/dvd/downloads/showsoftware_softencode_67.html (Demo version)
BDSup2Sub http://www.videohelp.com/tools/BDSup2Sub
DVDSubEdit http://download.videohelp.com/DVDSubEdit/
NuMenu4u (Deadware) http://www.dvdr-digest.com/software/numenu4u.html (Free version, requires Scenarist)
Muxman (free Demo version 0.16.x does everything you'll need) http://www.mpucoder.com/Muxman/versions.shtml
VobBlanker http://download.videohelp.com/jsoto/vobblanker.htm

Feel free to substitute with software you feel does the job better.

Depending on the DVD in question, this can be a long complicated process, but the goal is to make a near perfect conversion of all menus, video, audio and subtitles to NTSC from a PAL DVD.

You may find it helpful to strip unnecessary parts.  This can include content in languages you don't speak, copyright warnings, splash screens, etc.  Just remember, if you don't delete a piece of video or blank it, you will need to convert it to NTSC.

Before beginning, it's important that you understand your source.  You will be asked to examine the frames.  I like to use VirtualDubMod since it can load/edit avisynth scripts and even index mpeg/vob files so you can view them 1 frame at a time.

PAL can come in several flavors and what type you have will determine your conversion process.

Case 1: Film source with speed up (no pitch shift/correction).  This is by far the most common.  Film is 24 frames/sec.  To achieve PAL's 25 fps they just crank the film faster and speed the audio up with it.  The run time of the movie will be shorter and the audio will be higher pitched.  Both the video and audio will need to be corrected.

To identify this type of video make sure your video is progressive (no hard interlacing) with no duplicate frames, just 25 unique progressive frames a second.  You should also check Imdb and confirm that the runtime of the film is (about) 96% of the original length.  (Imdb can have this wrong, so take their runtime with a grain of salt.)

Now the hard part: Is the audio pitched higher?  There's no database or place to look for this.  You'll have to do it by ear.  If you have an NTSC version (even a VHS tape), listen side-by-side with your PAL version.  If you have access to a soundtrack CD or music from the film, compare them.  If you're really stuck, another movie with the same actor will let you compare the normal quality of their voice.

Most people can hear the 4% difference if they have a reference.  If you can't find a reference to compare to, I find it best to assume this is the case since it's the most common.  If your guess turns out to be wrong, you'll probably be able to tell that things sound slow and low pitched.

Case 2: Film source with speed up and pitch shifted/correct.  There are many that claim this is common, but I've almost never seen it.  I've been told by a friend in the industry that the process is time consuming and expensive, so it is avoided whenever possible.

Similar to Case 1, the audio still runs faster, but has had the pitch lowered so people don't sound like they are huffing helium.  It is better to leave the speed up as it is rather than trying to re-pitch shift the audio.

Check the video just as in case 1.  The difference is, when you listen to the audio it should run faster but have the same pitch as the original.

Case 3: Film source, interlaced with blended fields.  This is one of the worst case scenarios and perfect restoration is NOT possible.  The video is interlaced with some fields being an untouched frame and some being a blend of two different frames.  There is no speed up, so the runtime and audio are unchanged.  You will need a  specialized avisynth filter that is mostly effective at best.  I currently recommend Srestore.  Expect some left over blurry frames and slightly upscaled looking video when you are finished (but to be fair the original DVD probably doesn't look much better).

Srestore can be found on the avisynth wiki (along with info on use and other filters that are required for use. http://avisynth.org/mediawiki/Srestore

To identify this type of video, you'll first notice your picture is 100% interlaced.  If it is you will need to separate the fields and examine them.  Avisynth’s separatefields(), bob() or whatever else you like.  VDub has "deinterlace filter" (set it to "Unfold fields side-by-side").  These will allow you to examine the fields as if they were individual frames.  You should now see for about every clean field, a blurry double exposure looking one.

The audio will be untouched.  You will also find the runtime to be unchanged.

Case 4: PAL Video 25i.  Usually shot for TV but may be found in DVD extra content.  This type of video is interlaced 25 fps.  Annoying, but not impossible to work with.

It can be identified by separating the fields like in Case 3, but you will find 50 clean unique frames with no blending.
You will be converting this to 29.97i.  Audio will not need to be altered.

Trying to convert this type of video to progressive is possible but not recommended.

Case 5: PAL Video 25p.  Like above, made for PAL countries but no interlacing this time.
Treat this just like Case 2 since the audio is at the right speed.

Case 6: NTSC Video.  These are a bit of a problem.  The source will have been previously 29.97 interlaced video but converted to 25i.  Shot for TV in NTSC countries, it is 100% interlaced.  I will admit little familiarity with this situation, but what I've seen involved them separating the video into 60 fields discarded 10 and then reinterlaced to 25fps.

Case 7: Film source, duplicate frames.  Uncommon, but still possible.

In this case, for every 4 frames, there will be a fifth that is a duplicate.  The video may appear jittery when played.  This is most visible during motion or camera pans.


Some of these types you may never see, there are even types I've missed, but this should cover 99% of what you are likely to find.

What IS important is that your DVD may have the movie representing one case, a trailer another case, and a studio logo a third.  You will have to check and keep track of which pgcs (program chains) need what treatment.

Appendix B has some additional info regarding issues that may come up with commentary tracks, or dubbed foreign language films.

Letterboxing – a judgment call
At this point I want to mention your friend the upscale.  While most people won't even consider upscaling video, when converting PAL to NTSC it's really a different story.

If your PAL source is anamorphic or full frame you'll be doing a straight up resize job.  If your disc is letterboxed, seriously consider making your NTSC conversion anamorphic.

A PAL letterboxed movie at a 1.77:1 aspect ratio has 432 lines of resolution.
Your option becomes making a letterboxed NTSC disc by shrinking the picture by 72 lines (17% less resolution), or upscaling to anamorphic by adding 48 lines (an 11% increase).

For further comparison, at 2.35:1 a PAL letterboxed movie has 324 lines, NTSC letterboxed is 270 lines (54 less lines).  NTSC anamorphic is 360 lines (36 more).

To me that's a no brainer.  PC upscaling is considered better than on-the-fly by hardware.  You'll have higher detail, better widescreen television compatibility and little chance of artifacts from such a minor resizing.  It's up to you, but I would never consider converting PAL letterbox to NTSC letterbox.

It's also not uncommon to find older PAL movies that were shot in 1.66:1.  Yup, halfway between 4x3 and 16x9.  When transferred to DVD they are commonly letterboxed to avoid black borders on 4 sides.  Even less upscaling is necessary when converting these to anamorphic.

Part I - Ripping

0) If you have already ripped your DVD to VIDEO_TS format, you will need to build an ISO disc image to mount.  ImgBurn is good for making the image.  Slysoft's Virtual CloneDrive is recommended for mounting the image.  It's also free (other virtual drivers may impact your real DVD drive's reading and writing speed/quality.) http://www.slysoft.com/en/virtual-clonedrive.html

Now is also a good time to wash your movie through DVD Shrink.
DO NOT(!) use any compression or blank any video.  DVD Shrink should only be used to remove any audio or subtitle tracks you don't want.

For example, if you have two audio tracks, the first is French 5.1 and the second English 2.0, and you don't strip the French track, when rebuilding you will need something for stream 1.  Otherwise the default track and/or language menu may not work correctly.  English would be associated with your French button and your English button may not work.  DVD Shrink removes and remaps audio and subpicture streams correctly without making a mess

DVDShrink also lets you output to an ISO image you can mount to work with.

Another important warning is that DVDShrink by default removes the layer break and P-UOPs.  Depending on your preferences you may want to disable those preferences, especially if you plan to make a DVD9 and want the layer break to be in the same location.  (These options are located under Edit (Menu)/Preferences/Output Files (Tab)).

Note: If your hard drive is format with an older filing system like FAT16 or FAT32, size limitations will prevent you from continuing.  Your working drive must be NTFS or higher.

1) Start PgcDemux.  Click the first 'Browse' button and select VTS_01_0.ifo on your source (or the first title set you will be converting).

Each title set can contain one or many PGC's (program chains).  Think of it as 2 video's just stuck together.  They need to be separated for conversion.

In my example, I previewed VTS1 in a media player and see that it's only a copyright warning.  While I could convert it, it's just easier not to, and will save a little space on the final disc.  You could convert it if you want.

VTS2 on the other hand has the movie in PGC #01 and the trailer in PGC #02.

You may also find trailers, and a myriad of other extras.  You will need to rip every individual PGC from every VTS that you want to keep.  

Menus are dealt with separately so don’t worry about text production notes, still galleries or actor’s filmographies right now (unless they are actually video in a title set/pgc and not part of the menu).

2) Once you have a good idea what the layout of your disc is, create folders in your working directory so you can keep everything sorted.

So in my example I need the following empty folders:

Conversion\VTS1
Conversion\VTS2\PGC1
Conversion\VTS2\PGC2



Make sure the 'Mode' is 'by PGC' and 'PGC Selection'/'Domain' is set to "Titles".

Select the first PGC in the title set from the drop down box.

Click the 'Browse' button and select the folder that corresponds to your input video.

In my example I've selected VTS2's PGC #01.  So my output folder is:
Conversion\VTS2\PGC1

3) Click the "Check A/V delay" button.
You will get a pop-up window that shows the Audio delay for each of the audio tracks in that program chain.
If any say something besides "None", get a pen and write down which track and what the delay is.  We will need those numbers later.

4) Under 'Options' check all the boxes except "Include end time" and "Create a PGC VOB".

5) Click "Process" and wait until you get confirmation that it's done.

6) At this point, if any of your audio had a delay value shown in PgcDemux, you should rename the file.

Rename the files adding Delay ?ms at the end.

So if your first audio stream has a -8ms delay, rename like this:
AudioFile_80.ac3 => AudioFile_80 Delay -8ms.ac3

Be aware delays can be positive or negative.  Always add the minus sign if it is negative.

7) Repeat for all VTSs and PGCs.

Note 1: If you notice any title sets that are 0 length (and you've checked and made sure they're blank), ignore them and blank them later.

Note 2: You may find DVDs with more complicated PGC structures that share content.  This can happen with trailers, deleted scenes (with a play all button), or similar type extras.

This sort of structure may have a large PGC1 that contains all the content with each clip also in a separate cell.  Consecutive PGCs (2, 3, etc.) would have one cell each (shared with PGC1).

In this case you don’t need to rip each PGC, only the one that contains all of the content (longest run time).

Sometimes you will find a titleset that has a PGC that calls a chapter from the movie (like for a karaoke extra).  This works the same as above, it should self-repair when you replace the movie, so you don't have to do anything with it.

Part II – Converting the Video

0) At this point you should know what 'case' type each video and audio file is you've just ripped.  If you don't, check them and make a list of what you'll need to do.

1) Load your video file to be converted in DGIndex.  Make sure Video/Field Operation is set to "Ignore Pulldown Flags".  (If you need more info on this program read the manual.)  Click File/Save Project and output the .d2v file to the same directory.

2) Create an avisynth script making the tweaks you need for that particular video.

If you are converting from letterbox to anamorphic you will need to Crop(0,Top#,0,-Bottom#) the black borders before resizing.  You should see 72 lines (unless the AR is less than 1.77:1) of black above and below, but usually they’ll put more on the bottom.  Reposition the movie vertically to taste.  Just make sure the crop equals 144 lines.

Use this webpage if you need help:
Aspect Ratio Calculation Tool: http://www.haku.co.uk/ARCT.html.

Suggested scripts (same video types/case numbers as above).  Feel free to riff on these however you like (deblocking, removing noise/grain, tweaking colors, brightness, levels, etc.)

Filtering can of course improve your picture, or make it worse.
At minimum check your video for blocking compression artifacts.  DGDecode can fix this with the 'CPU=' option: http://neuron2.net/dgmpgdec/DGDecodeManual.html#MPEG2Source.
Stronger the post-processing can mean a softer picture, but complex scenes or panning shots may have macroblocking artifacts you want to clean.  CPU=1 or 2 can usually improve even good transfers with imperceptible negative effects.

The ConvertToYUY2() is added to the scripts assuming you will be using CCE.  If you are using an alternate encoder, change the colorspace as needed.

Case 1:

mpeg2source("movie.d2v")
spline36resize(720,480)
assumefps(23.976)
converttoyuy2()

Case 2:

mpeg2source("movie.d2v")
spline36resize(720,480)
assumefps(25)
converttoyuy2()

Case 3:

mpeg2source("movie.d2v",iPP=true)
loadCPlugin("yadif.dll")
yadif(mode=1) # Bob mode - generates double the framerate
Srestore()
spline36resize(720,480)
converttoyuy2()

Check http://avisynth.org/mediawiki/Srestore for other options and suggested scripts for Srestore.  Tweaks may improve your output.

There are numerous Bob-ers out there, but for speed/quality I prefer Yadif. http://avisynth.org/mediawiki/Yadif

Case 4: (Your final video WILL be 29.97i.)

mpeg2source("source.d2v")
loadCPlugin("yadif.dll")
yadif(mode=1) # Bob mode doubles framerate
Spline36Resize(720,480)
ChangeFPS(59.94) # Creates duplicate fields. Use ConvertFPS if you prefer blended fields.
AssumeTFF()
SeparateFields()
SelectEvery(4,0,3)
Weave()
converttoyuy2(interlaced=true)

ConvertFPS is frequently recommended.  It creates blended fields in order to attain the extra frames.

ChangeFPS is a good alternative since it will just make duplicates of existing frames.  Unfortunately this may effect the smoothness of motion.

I've been told using MFlowFPS (http://avisynth.org.ru/mvtools/mvtools2.html) is even better.  It is suppose to extrapolate new frames.
I have mixed emotions about its output, but I'm passing this information on anyway.

super = MSuper(pel=2)
backward_vec = MAnalyse(super, isb = true)
forward_vec = MAnalyse(super, isb = false)
MFlowFps(super, backward_vec, forward_vec, num=60000, den=1001, ml=100)

Note: If you get a really good quality deinterlacer (yadif in Mode=0 is considered pretty good) you can create 25p content from this type of video and then treat it like Case 5.  The process is more complicated and the better your desired quality, the better the deinterlacer and the more powerful of a computer you will need.
You're probably better off keeping things interlaced like the source.

Case 5: (same as Case 2)

mpeg2source("movie.d2v")
spline36resize(720,480)
assumefps(25)
converttoyuy2()

Case 6:
To be fair, I've never actually converted this type of video.
My best guess is to use the same script from Case 4.

Case 7: (The script while technically imperfect is vastly easier than being right.)

mpeg2source("movie.d2v")
spline36resize(720,480)
fdecimate(23.976)
converttoyuy2()

3) Repeat for all of your video assets.  Index with DGIndex and prepare a specific script for each piece.

4) You need to figure out the bitrates for each asset.
The best way to do this is calculate how much space the original asset took up (video+audio+subtitles) as a percentage of the disc. (Get the numbers from your post-DVD Shrink disc, not the original.)

So for example if your original movie took up 3.88gb out of 4.03gb (discs aren't always full) you want it to use 3.88/4.03 = 96.3% of your final DVD (4.37gb/4482mb for a DVD5, 7.95gb/8144mb for a DVD9).  (1gb = 1024mb)

That's .963 * 4.37 = 4.20gb or 4308mb.
This math also works if your original is a DVD9 and you're going to a DVD5 or vice versa.
Grab VideoHelp's bitrate calculator: http://www.videohelp.com/tools/VideoHelp_Bitrate_calculator and click the "Advanced" button.

Plug in the length for your new video (if you're changing the playback speed, get this number from your AviSynth script by playing it in a media player or loading it in VirtualDub).

Set the “Audio Bitrate” to account for all audio tracks you are keeping.  If you have multiple streams of different bitrates, you may have to be creative. (192kbps + 448kbps = 1 Track of 640 kbps or 2 Tracks of 320 kbps)

Set the ISO __ MB/DVD to 0 (you will have to account for those 25MB of overhead, just not more than once.)

Set "Custom Size" for the DVD to reflect the number you just calculated.  Still 4308mb in my example.

Now this won't be perfect.  If you've removed audio streams, it will account for that and you'll use that extra space.

You will be re-encoding the menus, and if you stripped other clips or content from the disc, you'll have more wiggle room and can go a little higher.



Your encoding software will also need a maximum bitrate.  Max bitrate is shown on the calculator, and in theory 9800kbit/s is the max total allowable for a DVD.  I prefer to give extra headroom.  Subtitles or variable bitrate spikes can take you too high.  I usually subtract about 300-800kbps from the maximum calculated here and never use higher than 8800kbps.

5) I prefer CCE for video encoding, but use what you like as long as it can handle AVS files.


Make sure you keep straight which of your videos are widescreen, fullscreen, interlaced, progressive, 23.976, 25 or 29.97 and set your encoder accordingly.

You shouldn't need more than 3-5 passes (according to Doom9), or you can try OPV as long as you don’t exceed your target bitrate.

6) With DGPulldown, add pulldown flags as needed.
Case 1 & 7: “23.976-->29.97”
Case 2, 3 & 5: “25-->29.97”
Case 4 & 6: No pulldown needed.

Browse to your encoded video file.  A destination file name will be automatically chosen.
Click the button that corresponds to the pulldown you need and click the Convert button.

7) Repeat for all your video assets.

Part III – Converting the Audio

There are plenty of pieces of software for converting audio.  BeLight isn't perfect, but it does cover all the bases for what we need and while giving the most control.  It is buggy, but you don't have to worry about your output as long as you double check the command line before encoding.

0) If your movie is case #1, it needs to be slowed down and re-encoded.  For all other cases, just go to Part IV.

1) Find out all you can about your audio.

If your source is PCM (not common.  Mostly concert DVDs.) and you want to keep it that way you can (just skip step 3), or you can re-encode to Dolby Digital (don’t skip any steps).

If your source is DTS you can either re-encode to high bitrate AC3 (Dolby Digital @ 640kbps is considered comparable to DTS @ 768kbps) or convert DTS to DTS.  Either way, you’ll still need to slow the audio as shown in steps 2 and 3.
See Appendix A for more info on working with DTS.

For Dolby Digital you can get information on the file in a few ways.  One is to play it in a directshow player that lets you view codec properties.  Open up details for AC3Filter and click to the SPDIF tab.  There you can view decoder information.  Scrolling down and you’ll see the bitrate, number of channels, clev (center mix), slev (surround mix) and dialnorm (dialog normalization).  Write these numbers down.

You can get the same info in Sonic Foundry Soft Encode by opening the AC3 file in 'frame view'.  Here you may see additional information.  Make a note of whatever you find.

I prefer Sonic Foundry Soft Encode for Dolby encoding which means using an intermediate wave file.  If you are a fan of BeLight/BeSweet/Aften, you can do this all in one step and go from AC3 to AC3.

2) Decode to a wave file with BeLight.

File (menu)/Add Files to Load the PAL audio, click the WAV/PCM tab and click the "WAV" button as well as "16 Bits (Six) Mono Waves".
If your source is stereo, use "16 Bit Stereo" instead.
If the source is 1-channel mono, you can keep it as 1 channel, but I feel it’s more professional to use “16 Bit Stereo” to make it a 2-channel mono track

Don't select any SSRC, Boost, resampling etc.  You want this as untouched as possible.  Check the "(Azid) Dynamic Compression" and "Mode" buttons, but don't worry about the specific settings, we'll deal with that later.

You should notice that if you added a Delay to the file name, BeLight will have automatically checked the Delay option and filled in the number.  That's good.  If it didn't added it now.

Click the Advanced Settings button.  Under SoundTouch, check the "Covert Frame Rate from ... to ... fps" button and enter: 25000 and 23976 in the boxes.  Uncheck everything else (except perhaps the Output Log File if you want it).  Click Ok.

Click “View or Edit BeSweet Command Line” and verify your settings, which should look like this (but may be wrong since BeLight is quirky):

"C:\BeLight\BeSweet.exe" -core( -input "c:\work folder\VTS_03_1 - 0x80 - Audio - AC3 - 2ch - 48kHz - DRC - DELAY 0ms.AC3" -output "c:\work folder\VTS_03_1 - 0x80 - Audio - AC3 - 2ch - 48kHz - DRC - DELAY 0ms.wav"  -6ch -azid( -c normal ) -ota( -hybridgain -r 25000 23976 )

If you have a delay, you should also see it listed in the -ota section.
-ota( -d -8 -hybridgain -r 25000 23976 ) would be for a -8ms delay.

[u]Now make a these changes:[/u]
Where it says: "-azid (-c normal)", change it to read: "azid (-c none)".
Where it says: "ota( -hybridgain -r 25000 23976 )", change it to read: ota( -r 25000 23976 )
The gain type may vary, just remove it.  (Also leave any -d switches, that's your delay.)

This prevents the volume from being maximized or dynamically altered.

You should now have:

"C:\BeLight\BeSweet.exe" -core( -input "c:\work folder\VTS_03_1 - 0x80 - Audio - AC3 - 2ch - 48kHz - DRC - DELAY 0ms.AC3" -output "c:\work folder\VTS_03_1 - 0x80 - Audio - AC3 - 2ch - 48kHz - DRC - DELAY 0ms.wav"  -6ch -azid( -c none ) -ota( -r 25000 23976 )

Click Run and wait a long while.

Note: If you prefer to encode directly to AC3 instead of using 16 Bit (Six) Mono Waves, click the AC3 tab, select Stereo or 5.1, use Aften (a pretty good quality encoder), click Bitrate, and move the slider to the bitrate you want (you’ll probably want to match the original).
When you edit the Command Line (as detailed above), make sure you give your output file a different name than your input file or nothing will happen.

If the when previewing the command line, things seem completely wrong, close BeLight and re-open it.  It will usually get things working again.

3) Encode back to Dolby Digital with Soft Encode.
File (menu)/Open your 6 Mono Wave files (or single Stereo Wave file).

Make sure your channels match up.  If your source was stereo or mono it should automatically be mapped correctly.

The six mono files each have had their names appended with C, FL, FR, LFE, SL, or SR representing Center, Front Left, Front Right, Low Frequency Effects, Surround Left and Surround Right respectively.

On the left side for each channel you’ll see a mini representation of speaker positions.  Click the speaker position that corresponds to each file’s extension.



Go to Options (menu)/Encode Settings.
On the left is the “Audio service configuration”.
The Data Rate is your bitrate.  It’s best to use the same or better than the original PAL file to prevent further degradation (Dolby Digital is a lossy format).
Audio coding mode is the speaker configuration.  It’s in the format of Front/Rear.  So 2/0 is 2 front speakers, no rear speakers (stereo) and 3/2 is 3 front speakers and 2 rear speakers (5.0).  Check the LFE enable if you are using 5.1 (or 4.1, etc).
Set the Dialog normalization to the number you found Step 1 of Part III.

If your source was not originally AC3, you’ll want to read this guide http://forum.doom9.org/showthread.php?p=332259#poststop section on “Referencing Volume to a Known Level - Dialogue Normalization”.

On the right side under the Bit Stream tab, fill in all the settings to match the original mix.  I’ve found no point to the copyright bit and original bit stream options so feel free to ignore them.

Click the Preprocessing tab.  Uncheck everything here.  You can set the Surround channel processing options if you know you need to, but otherwise forget it. 
If your source was DTS or PCM you’ll probably want to use Soft Encode’s help file suggested settings.



Dynamic range compression should be Film:Standard for pretty much all movies.  You can use Film:Light if you have a generally quite movie with little variation in volume throughout.  (I have yet to find any software that can read the original file’s Compression characteristic so that’s our scientific method.)



Click OK to close the Encode settings then File (menu)/Encode and pick a file name, click save and wait while Soft Encode does its thing.

4) Repeat for all audio assets.

Part IV - Subtitles

0) If you find yourself with more .SUP files than there are listed in the language menu of your DVD, it's likely there are forced subtitles or separate titles for widescreen and letterboxed display.

If that is the case, load DVDSubEdit. Select File (menu)/Open full domain.

Navigate to your PAL DVD’s directory and select the first VOB of the titleset you want to examine. In my example it’s VTS1 so I open VS_01_1.VOB (don’t select any that end in “0”, those are menus). Click Open.

Once the program has scanned the entire titleset, go to the 'Subpic selection' drop down box (where it says "All Streams"), and select the individual streams one at a time and take a look at the preview.

(You may need to move the slider under the preview window back and forth (the left/right arrow keys work too).)

The preview window will resize in response to the subtitle's corresponding video format.  That is, if the window is square with the image letterboxed, that sub stream is for letterboxed.  If it shows a 16x9 screen, those subs are for wide.

You'll remux the subpictures back in order later, so the only thing you need to worry about is wide versus letterboxed.

The stream name in the 'Subpic selection' box shows you the track's StreamID and WS or LB.

It's a good idea to rename your ripped subs to keep this straight.  For example:

0x20(en)WS is English Wide
0x21(en)LB is English Letterboxed
0x22(en)WS is English (SDH) Wide
0x23(en)LB is English (SDH) Letterboxed

Rename as follows:

Subpictures_20.SUP => Subpictures_20 WS.SUP
Subpictures_21.SUP => Subpictures_21 LB.SUP
Subpictures_22.SUP => Subpictures_22 WS.SUP
Subpictures_23.SUP => Subpictures_23 LB.SUP

Close DVDSubEdit and DON'T SAVE ANYTHING.

1) For each of your subtitle tracks, you will need to make a copy of the original .IFO to that titleset and rename it to match your .SUP file.

For example, if VTS1 has 4 subtitle tracks, you will need to copy VTS_01_0.IFO four times.

So continuing with the above example.

Subpictures_20 WS.SUP
Subpictures_21 LB.SUP
Subpictures_22 WS.SUP
Subpictures_23 LB.SUP

Will need four copies of VTS_01_0.IFO named:

Subpictures_20 WS.IFO
Subpictures_21 LB.IFO
Subpictures_22 WS.IFO
Subpictures_23 LB.IFO

WARNING: A PROBLEM WITH BDSUP2SUB CAUSING JITTERING HAS BEEN FOUND.  THIS PORTION OF THE GUIDE NEEDS REWORKING.  IN THE MEAN TIME IF YOU TRY THIS METHOD, TEST PLAYBACK ON THE DEVICE YOU WILL BE WATCHING YOUR DVD ON.

2) Open BDSup2Sub.  Select File/Load and navigate to your working directory.

Open your first IFO file that you need to edit.  (Do not open the SUP file or you will receive an error.)

'Conversion Options' will appear after the titles are loaded.

Check the box for 'Convert Resolution' and select 'NTSC (720x48)'.

If this video is Case #1 you will also need to check 'Change frame rate' and select 'FPS Source' as '25' and 'FPS Target' as '23.976'.

Click 'Ok'.

3) The top window will show you the original subtitles.  The bottom window the resized version.

If you don't like your preview, you can change the resizing filter.  If you're not sure what looks best, I recommend 'Bilinear'.  It produces titles about as sharp as Lanczos3, but without the inherent ringing that can remove detail and make the top/bottoms look squashed.

Just open the drop down box at the top and select the one you like.

Make sure that at the top 'Palette' is set to 'keep existing'.

4) When you're ready, make sure the 'Output Format' at the top is set to 'SUP/IFO', and select File/Save.

BDSup2Sub will automatically use the same director and add 'exp' to the file name.

Just click the 'Save' button.

5) Repeat for all your subtitles for all your VTSs and PGCs.

BDSup2Sub will remember your last settings between files.  After getting the first titles how you like it, all you have to do is 'Load', click 'Ok' to the 'Conversion Options', and then 'Save/Export'.

When you are done you can delete all the Subpictures...IFO files.  They serve no purpose from here on out.

Part V - Chapters

1) Open your first PGC's Celltimes.txt file in Notepad.

2) All those numbers now need to be corrected.

Case 1 needs to be multiplied by 1.25. This can be done manually or using this XL spreadsheet: http://www.mediafire.com/?jogaymnnywe

All other cases need to be multiplied by 1.1988.  Here is a spreadsheet for that: http://www.mediafire.com/?hqnnm1nenzy

Copy and paste the list of cell times into the left column. Copy the new cell times from the right column.

Paste these into a new text file and save it.  (I like to call them 'Celltimes_NTSC.txt' to avoid confusion.)

Part VI – Menus

NuMenu4u was developed for re-encoding menus of DVDs before good quality single-program options were available for DVD9 to DVD5 conversion.  It was used in conjunction with other software to semi-manually re-encode and rebuild DVDs.
Unfortunately it is unsupported and disappearing into oblivion.  The basic free edition requires commercial authoring package Scenarist.  The full registered VIP version supports the free/cheap Muxman.

This guide assumes you have the registered version.

0) NuMenu4u can be a bit overwhelming at first.  Here is a good guide for configuring it properly: http://www.doom9.org/mpg/big3-preperation.htm#NuMenu4u

You’ll need to point NuMenu4u to your video encoder, audio encoder, and authoring software.  For the rest you can probably get away with the default settings, and let the wizard configure everything for you.

1) Open NuMenu4u. Click Wizard (menu).

Check “Convert Format (PAL -> NTSC / NTSC -> PAL)”, and “Use Muxman for authoring instead of Scenarist”.

Select the drive of your mounted image and click Next. In theory NuMenu4u can work with a ripped folder, but I’ve never been able to successfully accomplish that.

Click the disc icon and select an output directory for the converted menus. Click Next.

Select a preferred Language and click Next.

The Minimum Menu Size to Process option is unimportant since all menus will be re-encoded.  Just click Next.

Select your preferred video encoder.  You have lots of options.  Cinema Craft Encoder 2.70 SP works fine. Click Next.

If you want you can click the “Show Advanced Encoder Settings” and tweak some options.  I like to use One Pass VBR with the Q set to 10, but you have plenty of options to choose from.

Click Next, then “Press to Start Processing”.
It will beep, flash, open and close loads of windows and programs and generally go nuts.  Go get a drink or something because you won’t be able to do much else with your PC while this is going on.

When it’s finished it will close itself and you’ll hopefully (among other things) have a VIDEO_TS folder in the directory you indicated.  Feel free to delete everything except the VIDEO_TS folder.

Part VII – Rebuild (finally)

0) The next step is to mux the audio, video and subs for each pgc.  Create some folders.  You will need one for each asset.  I like to keep things neat like this:

Conversion\Mux\VTS1
Conversion\Mux\VTS2_PGC1
Conversion\Mux\VTS2_PGC2

1) Load MuxMan.
Under Video, click ‘browse’ and locate your recoded video file.
The “Video Multi-File Manager” will open.  Just click the “Close” button at the bottom right.

Under Audio, click ‘browse’ and locate your recoded audio file.
The “Video Multi-File Manager” will open.  Just click the “Close” button at the bottom right.  You can set the Language and Extension, but these will be replaced later when you rebuild the disc.
If you re-encoded your audio or there was no delay, leave the 'Delay' box blank at 0.  Otherwise, enter the number in ms.

If you have more audio streams to add, click the up arrow by “Stream:”.  You can then browse for your next file.  Repeat as needed, adding your tracks in numerical order.


Under SubPictures, click ‘browse’ and locate your recoded SUP file (those are the ones that end in 'exp').
The “Video Multi-File Manager” will open.  Just click the “Close” button at the bottom right.  Again you can set the Language and Extension, but it’s not necessary.

If you have separate widescreen and letterboxed subpictures, make sure to click the "LB" or "Wide" that corresponds to each track (which you should have renamed with LB or WS to keep straight).

Otherwise, if you video is widescreen, click the “LB” and “Wide” buttons.

If you have more subtitle streams to add, click the up arrow by “Stream:”.  You can then browse for your next file.  Repeat as needed, adding your tracks in numerical order.

For example:

If your subtitles are...
Subpictures_20 WS_exp.SUP
Subpictures_21 LB_exp.SUP
Subpictures_22 WS_exp.SUP
Subpictures_23 LB_exp.SUP

Stream 1 will be 'Subpictures_20 WS_exp.SUP' and the 'Wide' button selected.
Stream 2 will be 'Subpictures_21 LB_exp.SUP' and the 'LB' button selected.
Stream 3 will be 'Subpictures_22 WS_exp.SUP' and the 'Wide' button selected.
Stream 4 will be 'Subpictures_23 LB_exp.SUP' and the 'LB' button selected.

MuxMan's track list will show...
Track 1: W 20 LB 21
Track 2: W 22 LB 23

If you got it wrong, you'll probably see a red warning.

For the Audio and SubPictures don't worry about the 'Language' or 'Extension' boxes.  That information will be copied from the original DVD when you rebuild.

Click File (menu)/Import Chapter.  Browse and select your retimed cell time list.

Browse and select the correct output folder, and click “Start”.

Note: If your last cell is right at the end of your movie (used for chapter skipping to the end), you will receive an error saying "reference to non-existent scene".  That’s because the final chapter is too small.  This is why we do not let PgcDexmux read the last chapter during ripping.

If you still get this error, open your cell times file and subtract 30 from the final number.  This will make the last chapter a little more than a second and fix the error.



2) Repeat muxing with MuxMan for all the PGCs you’ve re-encoded.

3) Open VobBlanker.
Under “Input Video Manager (VIDEO_TS.IFO)” click Browse and locate your NuMenu4u authored files.  Click down to VIDEO_TS and select the VIDEO_TS.IFO file.  Click Open.

You will get an error.  That’s fine because there aren’t any video files yet. Just click OK.
Under “Output Folder” click Browse.  The folder you select will be the final DVD unless you make further optional changes.  You'll want it to be a VIDEO_TS folder:

NTSC_Movie\VIDEO_TS

Under the TitleSet list, you will see that each of the VTS’s list the Video Format as PAL 720x576.
For each of these you will Right-Click and select “Video Attrib”.
This will open the Video Attributes settings.  Where it says “Standard”, click the NTSC button. Click OK to close the attributes window.  It will ask if you’re sure, click Yes.
Repeat for all VTS's.



Now click the first VTS (VTS_01_*.VOB) in the TitleSet list.  You will see a list of PGCs in the bottom window.

Click the first PGC in the list “PGCs in Selected TitleSet”.  Click the Replace button to the right.
Navigate to the corresponding MuxMan’ed folder, and select the VTS_01_0.IFO in that folder.  Click Open.

If you’ve converted the video from letterboxed to anamorphic wide, or changed the number of audio channels or type, you will receive a dialog asking if you want to correct the IFO.  Click Yes.

Repeat this process for each PGC in each titleset.
For any PGC’s you’ve decided to get rid of, instead of clicking "Replace", click the “Blank” button.

You will notice at the top a box that says Size(MB).  This will give you your final DVD size. 
A DVD5 can hold up to 4482mb and a DVD9 can hold about 8144mb.
If the number seems too small or too large, you'll want to go back and resize some assets.

It's a good idea to be about 20mb below the maximum if possible.
At about 5% undersized you may want to consider re-encoding the main movie with a higher bitrate.

If your DVD is less than 5% oversized, using DVD Shrink on it as a last step will have a negligible impact on quality. Any more and I’d strongly recommend stripping something or re-encoding some assets at a lower bitrate.

Once you’ve either replaced or blanked every PGC, click “PROCESS!!” and wait.



Close and quit VobBlanker.

Part VIII - Final Tweaks (Optional)

DVDRemake Pro http://www.dimadsoft.com/dvdremakepro/index.php

This is a really nice program that lets you change just about anything about a DVD.

If you have a menu that’s not in English (or the language you want), you can save a still image of it, change the text then reimport it.  (If the menu was originally animated, it won't be any more, but at least you'll be able to read it.)

You can also completely hide buttons that are no longer useful so people don’t try to select content that you stripped.

Other considerations:
Your original DVD may contain PC, DVD_ROM or JACKET_P content.
Feel free to copy any or all of this to your new disc.
JACKET_P is cover art used by some players.  There will be three files in the folder, a small, medium and large resolution video file that contain a still picture.  These need to be converted to NTSC as well.

This guide should help you get started: http://forum.videohelp.com/topic232001.html

Appendix A: DTS to DTS Conversion

While I won’t go into the full details, KpeX at the Doom9 forums wrote a nice FAQ on DTS audio. http://forum.doom9.org/showthread.php?p=424069#post424069

    3. How can I transcode a DTS file?

    Azidts http://www.doom9.org/software2.htm#audio can be used to transcode DTS with BeSweet (Note that you'll need WinDVD Platinum installed) or with one of the BeSweet GUIs (eg. OggMachine). The Hypercube Transcoder http://hypercube.is.dreaming.org/ supports DTS transcoding as well. One could also use DTSdec (assuming a large amount of HD space) to decode to wav and then encode to any desired format.

    Recent versions of foobar2000 http://www.foobar2000.org/support DTS playback and transcoding to any of fb2k's many output formats. Thanks to Kurtnoise there is an easy guide on transcoding DTS with fb2k http://forum.doom9.org/showthread.php?s=&postid=566664#post566664 to a variety of formats.

    4. Is it possible to encode DTS with BeSweet?

    Yes, but you'll also need Surcode CD/DVD Pro, which encodes DTS. Please see Q32 of the Besweet FAQ. http://forum.doom9.org/showthread.php?s=&threadid=15738

I've found Hypercube to be a great program for converting DTS to WAV.  The website has been down for awhile, but you can still find the program here: http://www.videohelp.com/tools/Hypercube_Transcoder

An important step in transcoding DTS is that the channels tend to be in a different order and you need to remap like this (if I recall correctly):
FL  -> FL
FR  -> C
C   -> FR
LFE -> SL
SL  -> SR
SR  -> LFE

If you want to re-encode back to DTS after applying slow down in BeSweet, SurCode for DTS is the only program I know of that encodes DTS and it's $100 for a single purpose encoder.  You may be better off using Dolby Digital at 640kbps.

Appendix B: Audio Commentaries and Dubbed Films

Sometimes audio dubs (for non-English films) and commentary tracks are recorded while watching the film with speed-up.

As a result the film should be slowed, but the audio track in question needs to be time stretched without slowing it.

Example 1:
Movie with a PAL region created commentary track:
Track 1 = Main movie soundtrack with speedup.
Track 2 = audio commentary without speedup.

Track 1 should be slowed as described in the full guide.
Track 2 time expansion with pitch correction is needed (see details below).

Example 2:
Foreign language film. (This can be found with some Japanese anime): Track 1 = Original non-English language track with speedup.
Track 2 = PAL country's English dub.

Track 1 should be slowed as in the main guide.
Track 2 should be time expanded.

If you aren't particularly worried about the foreign language soundtrack being pitch perfect, or you're going to strip the non-English track, you can treat the movie as a Case #2 by encoding the video at 25 fps and adding 25 to 29.97 pulldown flags.  The speed up in the non-English track will still be present, but the dub track is copied untouched.

To time expand:
While decoding from AC3 to WAV in BeLight, click on the "Advanced Settings".
Before we used the 'Convert Frame Rate' SoundTouch settings.  Make sure it is UNchecked.

Check the box that says "Tempo" and enter: -4.096
Click Ok.
Your command line should now look like:

BeSweet.exe -core( -input CommentaryPAL.wav -output CommentaryNTSC.wav -2ch -logfile Commentary.log ) -azid( -c none ) -soundtouch( -tempo -4.096 )

If you find any errors (technical, grammatical...), have any suggestions or run into any problems, please let me know. I'd like this guide to be as complete and accurate as possible.

PLEASE DO NOT CROSS POST THIS ANYWHERE ELSE. IT IS AN OT FORUMS EXCLUSIVE.

Amazing work, Dr. M.

@Possessed: I think you missed the point of this thread.

I've been playing around with programs that rip the celltimes.  There seems to be no accurate software for doing this and I'm not sure I understand why.

I tried ChapterXtractor, DVDDecrypter, PGCDemux and IFOEdit.

IFOEdit and PGCDemux produce the same results.  ChapterXtractor will generate matching results if you force it to display all cells and set it to 30fps (instead of the 29.97 default), but it may skip the last celltime. (This is when ripping from an NTSC disc not PAL).

The movie I was testing this with was only a frame or so off for chapters using the PGCDemux cells, which were much closer than the DVDDecrypter list which was just plain wrong.

I dunno what the situation is with DVDDecrypter, I'll play a bit more, BUT I'm going to update the guide in the next few days removing DVDDecrypter as the ripper of choice.

Edit: Or not.  It would seem the error creeps in ONLY in NTSC celltimes.

On PAL DVDs, DVDDecrypter and PGCDemux produce the same results (with PGCDemux on the default settings).

The guide is fine as it is, but I recommend using PGCDemux if you're working with NTSC celltimes.

3/25/2009: Expanded and changed recommended settings in the section for DVDSubEdit.  Better placement for 16x9 vs 4x3 displays explained.

3/26/2009: Corrected and clarified my clarifications from yesterday.

Edit: And a quick edit.  Mistakenly referred to 5% being safe for subtitles.  I meant 10%.  That's fixed now too.

3/29/2009: Discovered bug in more recent versions of DVDSubEdit.  Guide now recommends only v1.41 and older until it's fixed.

Lol. Good timing.

I've been working on a conversion today and making some tweaks to the guide while I was at it.

So far I've commented on DVDShrink stripping layerbreaks and P-UOPs.  The 25i script has been updated (suggesting some alternate filters, methods and it previously reversed the field order (oops)).

Things I've noticed that still need to be updated:

A comment on letting BeLight adjust for audio delays (if you ripped with DVD Decrypter, BeLight will read the delay and automatically set itself to correct it).

Possibly a recalculation for stream delays to reflect the slowdown (though the difference would be microscopic for anything less than 1000ms).

Further info on what to do if there are subs at the top AND bottom of the screen (I described it to you in a PM.  Appendix B still covers any problems there anyway.)

On the whole the changes are minor, wouldn't effect all discs and even if you didn't do any of it probably would have no noticeable effect on your finished DVD.

I suspect this guide will never be 100% done, but it's in very good condition right now.  Give it a go.  If you have any problems contact me.

Hmm.  I had some problems with NuMenu4U tonight... but then again I always do.  It's a pretty screwy program, but I've found nothing else that does the same thing.

First, I had to reauthor the disc with DVDRemakePro before it would successfully recode the menus... and even then I found that the audio was sped up to a comical pitch after conversion.

I don't know if it was a default setting I changed or not.  It converted the menus from 25 to 29.97 by using AssumeFPS(29.97) and speeding up the audio to match.

DJ, if you get a chance could you tell me the settings you see when you select AviSynth (Menu)/Show Format Conversion Script Editor?

Just the duration buttons at the bottom.  I had it set to "Keep all frames - Video length will be different and Audio will be adjusted".

The other option "Drop/Add Frames to keep Video/Audio length the same" worked better.  I never noticed a problem before, but I definitely did tonight.

Ok, time to eat a big hunk o' crow.

I'll be the first to admit that the celltime recalculations always made my brains hurt, but working with them today I realized that my math was completely wrong.

For Case 1 video (the most common) you multiply by a factor of 1.25.  I had wrongly assumed that this was true for most of the other cases.  After playing around with some files today it turns out Case 1 is the exception.  Cases 2-6 need a factor of 1.2 (I think).

I've never noticed my error before now because by and large cases 2-6 occur in trailers and stuff where there are few if any chapters and the content is short enough that it isn't a large enough discrepancy.

Now I may still be completely wrong and if anyone has a firmer grasp on this please chime in.  For now I recommend you check that the chapters are in the right locations after conversion.

Edit: Or not.  I'm really up against it here and I'm hoping the Doom9 guys can thrash it out.

The problem is I'm seeing a several second discrepancy now in the chapter locations and runtime (even though the audio syncs). 
Something very odd is going on that I mean to find out about.

THIS ONLY applies to cases where you are using 25 to 29.97 pulldown.

Edit 2: Ok, or NOT.  Here's the deal, case 1 which you are essentially using 23.976 to 29.97 pulldown is a multiplier of 1.25.  ALL OTHER cases require a multiplier of 1.1988. 
The guide and spreadsheets have been updated accordingly.

Interestingly I discovered another type of PAL video today.

A DVD extra I was converting was originally 24p, but converted to 25i by adding 1 duplicate frame per second and then interlacing.

I won't be adding it to the guide since it was a bit of a mess and I don't know how often they actually pop up like this.

My script was:

telecide(guide=2).fdecimate(rate=23.976)

It's not perfect, there were some interlace artifacts left behind, but I can't think of a better way to do it.

Something else that will probably need to be added to this guide.  Sometimes audio dubs (for non-English films) and commentary tracks are recorded while watching the film with speed-up.

As a result the film should be slowed, but the audio time stretched without slowing it

Example 1:

PAL movie:  Track 1 = Movie soundtrack with speedup.  Track 2 = audio commentary w/o speedup.

Track one should be slowed as described in the full guide.  Track 2 can be slowed but it will probably sound funny.  Instead, time expansion with pitch correction is needed (details below).

Example 2:

Japanese anime film: Track 1 = Original Japanese language track with speedup.  Track 2 = PAL country's English dub.  (Ghost in the Shell 2: Innocence UK release is like this).

Track 1 should be slowed, track 2 should be time expanded.

If you aren't particularly worried about the Japanese soundtrack being perfect (or you're going to strip non-English tracks) you can also encode the video as 25p and add 29.97 pulldown flags leaving the new dub track untouched.

For now, here is how to time expand:

(Edited to replace SoundForge with BeLight)

While decoding from AC3 to WAV in BeLight, click on the "Advanced Settings".

Before we used the 'Convert Frame Rate' SoundTouch settings.  Make sure it is UNchecked.

Check the box that says "Tempo" and enter: -4.096

Click Ok.

Your command line in theory should look like:

BeSweet.exe -core( -input CommentaryPAL.wav -output CommentaryNTSC.wav -2ch -logfile Commentary.log ) -azid( -c none ) -soundtouch( -tempo -4.096 )

Okay... the guide got a major overhaul.  (Hence it is now v2.)

Several programs are out, better programs are in.

More details are filled in to cover more eventualities.

Subtitle conversion is now easier and accurate for all cases thanks to a new tool available.

Lots and lots of changes, I don't anticipate any problems, but I always welcome contributions and corrections.

Edit: I also have a downloadable version of the guide available in Word format.

http://www.mediafire.com/?46vy3uond75v5u3

Edit 2: Um, ok made a quick fix to make sure the proper color palette is used during subtitle conversion.

I've removed the links to what you call "deadware" - I suspect there are still legal issues with the downloads of these programs.

More info:
http://en.wikipedia.org/wiki/Abandonware

Moth3r said:

I've removed the links to what you call "deadware" - I suspect there are still legal issues with the downloads of these programs.

More info:
http://en.wikipedia.org/wiki/Abandonware

Sure thing.  I'll update it with links to demo & free versions instead.

Couldn't have said it better.  If you ever get too confused, Citizen's aspect ratio calculator is something I really couldn't live without.

Not that you really need it in this case, but it would have answered the question as well.