Convert 30FPS MP4/FLV to DVD-compliant MPEG2

[abolibibelot]

I want to add a few YouTube videos as bonus to a customized DVD for my brother – the infamously “absoludicrous”, supposedly educational children video Be Somebody (Or Be Somebody's Fool by Mr. T (he's a fan of the actor), which I subtitled in french from scratch (in fact that was my first subtitling endeavour, I worked on it for weeks in 2011 – yeah you can pity me). Actually I already did that customized DVD years ago (also my first attempt at this, using DVDStyler, and I was quite satisfied with the result), but I don't remember how I converted those videos (possibly using Avidemux). I recently improved the subtitles and so I'd like to re-author the DVD with the best possible conversion process.
From what I could gather, also years ago, the highest recommended free MPEG2 encoder was HCEnc – is it still true ? Is it better than the MainConcept one found in Magix non-linear editing software ? Better than the one in Avidemux of ffmpeg ?
Then, I noticed that the quality of the source videos (for those which are still available on YouTube) had significantly improved in the mean time, as I kept those downloaded back then and re-downloaded one just now : this video is available now in 640x480 with a correct 4:3 aspect ratio, whereas the video files downloaded in 2011 from that same URL were a 540x360 FLV with measly 53kbps audio and a 480x320 MP4 with 125kbps audio (so the MP4 has a smaller picture frame and video bitrate but a much better audio bitrate), both with an incorrect 3:2 aspect ratio ; also the MP4 was in 29.97 FPS while the FLV was in 30FPS (what a mess !). I don't remember which one I used to create the MPG file for the DVD, probably the MP4 for it has a significantly better audio bitrate (I probably didn't know then that I could merge the video from the FLV with the audio from the MP4). I made screenshots, resized to 640x480 for those older videos, and the new one is definitely better. But it is in 30FPS, and HCEnc displays a “framerate is not DVD-compliant” warning. So what is the best way to do that conversion with all the proper settings ? Since I have to use an Avisynth script to load those videos into HCEnc, is it possible with Avisynth to change the framerate and keep the audio in sync ?

Also, based on the sample already linked above, would it be possible to improve the aspect of the subtitles, for better readability, or it this the standard display of DVD subtitles ?

[jagabo]

ChangeFPS(30000, 1001) will drop one frame out of every 1001 but will not change the overall running time so you don't need to modify the audio or subs. Depending on where those drops occur you may not even notice.

AssumeFPS(30000, 1001) will keep all the frames but the the running time will become 0.1 percent longer. For a 5 minute video that's 300 ms out-of-sync by the end. So you'll probably want to stretch the audio and subs too.

[abolibibelot]

Thanks for the quick reply.

In the mean time I tested another video, and I'm now not so sure about what I wrote earlier, regarding the encoding quality on Youtube having generally improved. For this one, the video files downloaded in 2011 are both in 352x288 resolution, the FLV has a 262kbps video bitrate and a 48kbps audio bitrate, the MP4 has a 245kbps video bitrate a 125kbps audio bitrate. The newly downloaded file, with youtube-dl and -f bestvideo+ bestaudio, is in 294x240 resolution with a 242kbps video bitrate and a 126kbps audio bitrate. What's interesting is that, comparing screenshots resized to the same 640x480 resolution, it appears that the newer file has less detail, but also much less concpicuous blocking artifacts than the former two. Is there any logic and consistency at all to all this ? In this case I wonder which one is “the lesser of two evils”, considering that the quality is pretty lousy for all three (and it seems a bit silly to convert 250kbps AVC to 4000kbps MPEG2 ! and even more so to do it all over again 8 years later expecting an ever-so-slightly improved outcome !...).

(2020-01-17 MP4)

(2011-07-10 MP4)

(2011-07-10 FLV)

[abolibibelot]

I searched about HCEnc (which I never used aside from some brief and tedious testing years ago, in a situation where I had a deadline and decided to use the more straightforward encoder from the NLE despite the reported superiority of the former) : so apparently it does only encode the video, not the audio, so something else has to be used for the audio, and for muxing the streams. But the discussions I can find on the subject are at least 10 years old. Are the methods they describe still relevant, or are there better workflows now ? What would be the most straightforward way of converting a video to MPEG2 using HCEnc and whatever else ?

[jagabo]

You can draw few conclusions about youtube encoding quality from files uploaded by different users. Because you have no idea what those users uploaded.

You can use an old version of TMPGEnc to mux M2V (from HCEnc) and AC3 audio streams into MPG files. Or you can use ffmpeg.

[CODE]
ffmpeg -i video.m2v -i audio.ac3 -codec copy output.mpg
[CODE]

[abolibibelot]

You can draw few conclusions about youtube encoding quality from files uploaded by different users. Because you have no idea what those users uploaded.

In this case those videos were downloaded from the same link, so the original uploaded video is supposedly the same, only Youtube's processing to generate the available streams is different. I wonder, for instance, if those blocking artifacts were present in the source and well filtered on the currently available streams, or if they were introduced in the formerly available streams by a poor quality AVC encoder.

You can use an old version of TMPGEnc to mux M2V (from HCEnc) and AC3 audio streams into MPG files. Or you can use ffmpeg.

Alright, so not much has changed, the combination of HCEnc and Aften is still considered optimal to produce the best possible MPEG2 encoding, isn't it ?

Also, what specific settings do I have to verify to ensure DVD compliance ? I saw for instance that the extra videos I converted in 2011 were encoded in “bottom field first” mode (“ligne du haut d'abord” below), whereas the main video {*} is encoded in “top field first” mode ; does it matter ? As for the bitrate, for the former version, I had to set an average bitrate of 4000kbps for the six extra videos (the main video has an average bitrate of 7984kbps) to fit them all on a standard DVD. With that bitrate, and considering the resolution of the source files, is it better to do the encode in standard 720x480 resolution, or set the resolution to 352x480 (“Half-D1”) to optimize the compression quality ?

{*} The B.S.O.B.S.F. VOB, from a carefully crafted VHS to DVD transfer which someone had religiously shared back then for a small niche of 1980s wildly weird vivid vidz devotees — and the crazy thing is that among the handful of commentators was a dude living in the same town as myself in south France, who complimented me on my subtitles and gave me some tips about DVD authoring when we had a chat later on.

MediaInfo report for the main video (which was encoded with HCEnc by the way, I just noticed it) :

Format                                   : MPEG-PS
Taille du fichier                        : 3,03 Gio
Durée                                    : 51 min 33s
Type de débit global                     : Variable
Débit global moyen                       : 8 407 kb/s
Bibliothèque utilisée                    : |HCenc 0.23.0.0 - (c) 2004/2008|

Vidéo
ID                                       : 224 (0xE0)
Format                                   : MPEG Video
Version du format                        : Version 2
Profil du format                         : Main@Main
Paramètres du format                     : CustomMatrix / BVOP
Paramètres du format, BVOP               : Oui
Paramètres du format, Matrice            : Personnalisée
Paramètres du format, GOP                : M=3, N=12
Paramètres du format, structure de l'ima : Frame
Durée                                    : 51 min 33s
Type de débit                            : Variable
Débit                                    : 7 984 kb/s
Débit maximum                            : 9 000 kb/s
Largeur                                  : 720 pixels
Hauteur                                  : 480 pixels
Format à l'écran                         : 4/3
Images par seconde                       : 29,970 (30000/1001) Im/s
Norme                                    : NTSC
Espace de couleurs                       : YUV
Sous-échantillonnage de la chrominance   : 4:2:0
Profondeur des couleurs                  : 8 bits
Type de balayage                         : Entrelacé
Ordre de balayage                        : Ligne du haut d'abord
Mode de compression                      : Avec perte
Bits/(Pixel*Image)                       : 0.771
Time code de la première image           : 00:00:00:00
Source du time code                      : Group of pictures header
GOP, ouvert/fermé                        : Ouvert
GOP, ouvert/fermé de la première imageGr : Fermé
Taille du flux                           : 2,88 Gio (95%)
Bibliothèque utilisée                    : |HCenc 0.23.0.0 - (c) 2004/2008|
Coordonnées de chromaticité              : BT.601 NTSC
Caractéristiques du transfert            : BT.601
Coefficients de la matrice               : BT.601

Audio
ID                                       : 189 (0xBD)-128 (0x80)
Format                                   : AC-3
Format/Info                              : Audio Coding 3
Nom commercial                           : Dolby Digital
Type de muxing                           : DVD-Video
Durée                                    : 51 min 33s
Type de débit                            : Constant
Débit                                    : 256 kb/s
Canaux                                   : 2 canaux
Channel layout                           : L R
Echantillonnage                          : 48,0 kHz
Images par seconde                       : 31,250 Im/s (1536 SPF)
Profondeur des couleurs                  : 16 bits
Mode de compression                      : Avec perte
Taille du flux                           : 94,4 Mio (3%)
ServiceKind/String                       : Complete Main

Menu

[davexnet]

I would just use AVStoDVD, popular on this forum, the author still makes an appearance from time to time.
Uses HCenc and Aften by default

What is the running time for all the videos you want to put on one DVD ?

AVStoDVD will calculate the optimum bitrate. You can do some test encodes to see if full or half D1 is better.
Field order is irrelevant if the source is progressive

[abolibibelot]

I would just use AVStoDVD, popular on this forum, the author still makes an appearance from time to time.
Uses HCenc and Aften by default

I'll try that then, thanks.

What is the running time for all the videos you want to put on one DVD ?

The main video has a size of 3.03GB (see above) and I don't want to re-compress it, to preserve its original quality. For the extra videos from YT, the total running time is 34m36s, plus a 1min / 42MB menu video, and also a native bonus with a size of 95MB (a rap album released the same year, in audio only, each track being a VOB with a single video frame, also transferred as-is for the customized DVD), for the remaining ~1.2GB capacity (full capacity of a DVD-R is 4.37GB but it's good to keep a safety margin and not fill them beyond 4.20-4.25GB, as the outer edge tends to degrade quicker, I learned it first hand the hard way, I also learned that DVDs from reputed brands like Verbatim could degrade much quicker than DVDs from lousy brands — anyway, I no longer use DVDs for archival purposes).

Field order is irrelevant if the source is progressive

Well the original source may have been interlaced (since those were mostly from TV footage from mid-to-late-2000s — I don't know how it was broadcast then in the USA or UK, or when the broadcasting standard switched from analog to digital, all I can see for sure is that apart from the Snickers commercials all of them were in 4:3 aspect ratio), but in this case, the only source I have access to is those YouTube videos, which are indeed progressive.
So there's no standard field order for the DVD format ?

[davexnet]

If the source is DVD compliant, it will not be recompressed. AVStoDVD uses mediainfo to find the information it needs,
field order is taken care of automatically.

[abolibibelot]

Oh, and I just read this – in my case, do I have to convert the progressive footage to interlaced MPEG2 with a similar script, or not at all ? (The first time around I had done so, and it would seem that I used the MainConcept MPEG2 encoder in Magix Video Deluxe for those conversions.) The main video as shown above is interlaced (VHS transfer), can this be problematic to mix interlaced and progressive content on the same DVD ? (The main video and the bonus videos are not in the same “title” if that's relevant.)

[manono]

...can this be problematic to mix interlaced and progressive content on the same DVD ?

No. Same title or different titles, makes no difference.

[abolibibelot]

I'm not sure I understand as on this highly detailed article I can read :
“MPEG-2 progressive_sequence is not allowed, but interlaced sequences can contain progressive pictures and progressive macroblocks.”
Does the author mean something else by “progressive sequence” ?

Tried AVStoDVD, so far it's more confusing than helpful... For instance I can't find explicit video bitrate settings, only minimum and maximum thresholds. Is this on purpose ?

[davexnet]

It's set here when the project is loaded. Greyed out by default,
deselect "auto video setup" and "auto bitrate calculation" to change.

Be aware the program always chooses an auto bitrate as high as possible so that the title(s) will fit
on the disk, or, so the setting does not exceed your Max rate in the preferences.

[abolibibelot]

Alright, then I'm not sure if it's the best tool for that particular job. As already said, I already re-authored this DVD years ago using DVDStyler, I kept the DVDStyler .dvds project file which (I hope) should still work with the current version (if not I can go back to the version I was using then), and there's probably nothing to change there, I just want to properly convert those six extra videos and re-create the DVD with improved subtitles. AVStoDVD seems more suited to starting from scratch. Or perhaps I should set a “custom disc size” to the remaining size once I remove the main video and the menu, to do the conversion with optimal settings ? (Or the total size of the six MPG videos created for the former version, which was 1124MB and resulted in an ideal total size of 4.24GB.)

[davexnet]

You could do that - set DVD size to your custom size, then set Output to Muxed mpeg2

[manono]

“MPEG-2 progressive_sequence is not allowed..."

It's a flag that can be turned on or off and it's not allowed for NTSC DVDs. Well, you could use it if you want your DVD to turn out all screwed up. But it has nothing to do with the DVD content which can be progressive, interlaced, or a mix of the two.

If you don't want to use AvsToDVD, you could always reencode the extras yourself so they and the already created main feature add up to 4.37GB or less. Not so hard. But I think AvsToDVD can easily do what you want - keep the main feature without reencoding, encode the extras, and put them all together into a single DVD-R.

As for the subs, you'll have to fix them yourself (maybe you already have) and either create DVD subtitles yourself (I make SST subs for use in Muxman) or let AvsToDVD do it. I don't use AvsToDVD myself but I think it can do all those things and davexnet can advise if I've said anything incorrect.

[abolibibelot]

Well, the thing I still don't quite get (I have little experience in DVD authoring) : I understand that progressive content can be encoded as interlaced MPEG2 (in which case two consecutive fields will belong to the same original progressive frame), or encoded as progressive MPEG2 (no change), but do the video DVD specifications allow progressive encoding, or is it required to encode everything as interlaced ? Or am I completely confused about the whole thing ?
Also : when progressive content is encoded as interlaced MPEG2, how come software players display the “combing” effect typical of (truly) interlaced footage ?

As for the subtitles, up to now, I simply loaded the SRT files into DVDStyler. Can AVStoDVD do something that DVDStyler can't ?

[davexnet]

Well, the thing I still don't quite get (I have little experience in DVD authoring) : I understand that progressive content can be encoded as interlaced MPEG2 (in which case two consecutive fields will belong to the same original progressive frame), or encoded as progressive MPEG2 (no change), but do the video DVD specifications allow progressive encoding, or is it required to encode everything as interlaced ? Or am I completely confused about the whole thing ?
Also : when progressive content is encoded as interlaced MPEG2, how come software players display the “combing” effect typical of (truly) interlaced footage ?

As for the subtitles, up to now, I simply loaded the SRT files into DVDStyler. Can AVStoDVD do something that DVDStyler can't ?

See the PDF for the HC encoder all the details are there. AVStoDVD sets the progressive flag for
progressive source and this is passed to the encoder. The author of A2D spoke to the author of the HC encoder
while A2D was being developed; all these things were accounted for.

[abolibibelot]

So, I created an Avisynth script for each of those 6 videos, encoded them with AVStoDVD, then generated the DVD with DVDStyler, after a few hiccups it seemed to have completed as intended... But now I realize that all those videos are in 1.5 aspect ratio ! I already resized them to 720x480 within the script, but AVStoDVD insisted on adding 30 pixels black bars and re-resizing, and I can't change that damn setting, whether I type "4:3" or "4/3" or "1.33" I get : "Warning! Inserted value is invalid." If I remove the cropping / resizing in the Avisynth tab, everything becomes red, supposedly because it considers the output invalid / non compliant...
How is this supposed to be done ?

Also, if an Avisynth script is loaded in 30 FPS, it adds a ChangeFPS framerate conversion, but if the framerate conversion is aldeady done in the script, it adds "Video = Video.MCJMFPS(29.97)" — why ?

Also, in DVDStyler, I can't find any setting to determine whether the subtitles should be active by default or not.
And apparently the current version is not happy with accents in subtitles' file names (accents in the videos' file names don't seem to be a problem). It doesn't warn about it right away though, only halfway through the generation process. With a former version I generated that same DVD with SRT files which had the same base name, some of them with accents, and there was no such issue.


EDIT : Found here : "MCJMFPS uses motion interpolation via mvtools. It will create lots of distortion with some material. DgPulldown is far better." Why, why does it add such an overkill processing to convert from 29.97 FPS to 29.97 FPS ?...
And why does it convert to YV12 footage which is already in YV12 ?...

EDIT : This is really stupid... If I change the AR value from "1.5" to "1" it works (but it's not what I want since now it insists on adding 120px black bars), but if then I try to set it back to "1.5" it stubbornly warns me that this value is invalid !...

[abolibibelot]

Another issue : I found out that DVDStyler does not apply formatting on SRT subtitles, italics in particular. Apparently this is a well known issue, and not a trivial one to fix. And as already mentioned, the aspect of the subtitles generated by DVDStyler is not exactly pretty (rough edges which make them quite difficult to read). Another problem is the fact that some subtitles can be too long and cut into 3 lines, depending on the font size and the width of the margins, and it's quite unpredictable (I've seen this with a 42 characters line, but elsewhere a 43 characters line can be displayed correctly), so it would be necessary to re-watch the whole thing without blinking in order to be sure that there's no such mistake.
So, is there a tool which can reliably convert SRT files to DVD subtitles, allowing to preview the output, to control the size, to check if no line is too wide, to set the font type, size and whatnot, and which can preserve italic / bold formatting ?

[davexnet]

I don't think the motion compensation is a good default, you can modify the preferences (hard frame rate conversion) to use
ChangeFPS (insert/drop frames) or ConvertFPS (blend frames) instead.

Once you change the preference, you'll need to rebuild the project.
See these images to see how to change the A/R of the source, since A2d assumes 1.5, the SAR of the source.
30 FPS is not DVD compliant
Any other issues provide the external script and A2d log.

Somebody else can help with DVDStyler and subtitles, they're not my thing

[babygdav]

http://subtitlecreator.sourceforge.net/AddingSubtitles2aDVD.htm
http://www.dvdflick.net/
https://sourceforge.net/p/dvdstyler/discussion/318795/thread/c67ce125/

No idea about making a dvd using only freeware tools - paid programs do everything fast and without a hitch. But, maybe pages like this can help?

[manono]

So, is there a tool which can reliably convert SRT files to DVD subtitles, allowing to preview the output, to control the size, to check if no line is too wide, to set the font type, size and whatnot, and which can preserve italic / bold formatting ?

MaestroSBT does all that for SSA files (easily converted from SRT) and I use it all the time. Of course, you can set a lot of the styles in the original SSA subtitles.

[babygdav]

Hcenc. Of the free one, yes, generally the best choice..

That's said, procoder, cinema craft, ateme, etc commercial products are better.

Home users, Adobe media encoder is a very nice choice.

All encoders can do Better or worse depending on the settings and bitrate, so you'll want to test your clips

http://forum.doom9.org/archive/index.php/t-176669.html

[abolibibelot]

See these images to see how to change the A/R of the source, since A2d assumes 1.5, the SAR of the source.

I did exactly what is shown on those pictures, but then any value I type is considered "invalid", except "1", which is of course not correct. I tried 1.33, 1.333, 1.333333333333333, 4:3, 4/3... and 1.78 is not accepted either. I also tried to modify the DAR values in the .a2d project file, then reload it : doesn't work with "4:3" (AR displayed in the GUI as "4") but seems to work with "1.33" (AR displayed as "4:3").
The only workaround I could find was to remove all unwanted conversion in the Avisynth tab. Then it seems to proceed as intended, and the output files have a correct AR.
(Using version 2.8.8 if that matters.)

30 FPS is not DVD compliant

I thought that pulldown flags could take care of that, but apparently this works only with native resolutions inferior to the target one, right ?
But then, why does AVStoDVD insist on converting the framerate even after I already converted it (with ChangeFPS(30000,1001)) in my Avisynth script loaded as source ? (Under "Source titles" it displays the framerate as "29 fps" for those clips, which is also confusing.)

For the 25 FPS clips, according to what I could read in the Avisynth tab, "#PullDown: using DGPulldown/HCenc to upsize FPS", the original framerate should have been preserved, with a pulldown flag applied to mimic the valid value of 29.97 FPS on playback, yet according to MediaInfo the resulting files seem to be encoded as 29.97 FPS with "hard pulldown". VirtualDub2 also identifies them as 29.97 FPS, and "MPEG-2, 4:2:0 interlaced".

Hcenc. Of the free one, yes, generally the best choice..
That's said, procoder, cinema craft, ateme, etc commercial products are better.

As it's been already mentioned in this thread, AVStoDVD uses HCEnc by default. It's supposed to make the process simpler and to ensure that the output is DVD compliant, but so far it first and foremost ensures a lot of screaming in the dead of the night !

That's said, procoder, cinema craft, ateme, etc commercial products are better.

Well, I don't have millions to spend, surely not on this revamped DVD which at best will be watched by a handfu’ of crazy foo’s...
The alternatives I have available right now are ffmpeg, Avidemux (don't know which encoder it uses under the hood), and the MainConcept encoder included in Magix Video Deluxe.

All encoders can do Better or worse depending on the settings and bitrate, so you'll want to test your clips

In this case the bitrate is quite constrained as there's room for about 1125MB worth of extra content, and the total running time for those six clips is 34m36s as already stated, which gives an average video bitrate of about 4300kbps.

[davexnet]

Update A2d to the latest version, 289 alpha dated June 2019.
When I drop an Avisynth script with changefps(29.97), It shows up in A2d as 29.97.
I suspect there is something wrong with your script
https://www.videohelp.com/software/AVStoDVD

It uses DGpulldown when the source is a standard framerate, here the source is 25 fps

[abolibibelot]

Update A2d to the latest version, 289 alpha dated June 2019.

I downloaded that one on VideoHelp about two weeks ago, I figured that it was the most recent stable version. Is this one stable, despite being “alpha” ?

It uses DGpulldown when the source is a standard framerate, here the source is 25 fps

That line was there for the two 25 FPS source clips, but the output video were still 29.97 FPS, unless I'm mistaken.
Also (probably unrelated) (?) the timings are wrong in VLC Media Player, for instance a 3m18s video is shown as 1h18m16s.
And in VirtualDub2 I can see the pulldown pattern (for the MPG videos rendered from 25 FPS sources), but not in VLC Media Player. Is this normal ? Does this mean that it's indeed 29.97 FPS, and VLC is somehow deinterlacing on-the-fly, or that it's 25 FPS with pulldown flag, but then why does VD2 show those frames as interlaced ? How do media players and video editors usually deal with pulldown flags ?

I suspect there is something wrong with your script

Apart from the first one (which appends several source clips of various resolutions) the scripts are very simple, like this one :

OBrienV = LWLibavVideoSource("H:\Mr. T YouTube\YouTube - Late Night - Mr. T Interview [mEyz3IUWFWg -f 43+251].webm").Lanczos4Resize(720,480)
OBrienA = LWLibavAudioSource("H:\Mr. T YouTube\YouTube - Late Night - Mr. T Interview [mEyz3IUWFWg -f 43+251].webm")
AudioDub(OBrienV,OBrienA).ChangeFPS(30000,1001)

There is one script for which I added a blank clip with a subtitle at the end, and as the source file was identified as "2997/100" FPS, it didn't work if I set the framerate of that blank clip with "fps = 30000, fps_denominator = 1001". Are the DVD specs so sensitive that 2997/100 would be considered invalid ? (It's only that one file, the others are either in 25 FPS and I let the framerate unchanged, or 30 FPS in which case I added ChangeFPS(30000,1001).)

BackstageV = LWLibavVideoSource("H:\Mr. T YouTube\YouTube - Mr. T - Talks 'Rocky III', 'A-Team' and his Life and Career [qckK3x0httg -f 43+251].webm").Lanczos4Resize(720,480)
BackstageA = LWLibavAudioSource("H:\Mr. T YouTube\YouTube - Mr. T - Talks 'Rocky III', 'A-Team' and his Life and Career [qckK3x0httg -f 43+251].webm")
logo = BlankClip(length=50, width=720, height=480, pixel_type="YV12", fps=2997, fps_denominator=100, audio_rate=48000, channels=2, sample_type="16bit", color=$000000).Subtitle("Comcast CN8, 2006", y=420, size=26, text_color=$FFFFFF, align=2).FadeIn(10).FadeOut(15)
Backstage = AudioDub(BackstageV,BackstageA).Trim(36,0).FadeOut(10)
Backstage ++ logo

Meanwhile, I tried MaestroSBT, it does what I want indeed, but now in what format should I export the result ? There are quite a few options, but some of them seem to be proprietary formats used by specific utilities. What format(s) among those available can either be loaded directly into DVDStyler, or be converted to the standard SUB/IDX format ?

[abolibibelot]

Regarding the subtitles, I switched to using SubtitleCreator, which can directly convert SRT to SUB/IDX format, with a better previewing interface (although the menus are a bit confusing and I'm already getting tired of that semi-kinky Ashley Judd picture displayed as default background). DVDStyler does not accept SUB/IDX subtitles, but a workaround — found here — is to first mux the MPG or VOB with the SUB/IDX subtitles with MKVToolNix, then load the MKV into DVDStyler.
On the same page someone states that : “The subtitle format used on a DVD (idx/vobsub) is a low resolution image based format and tends to look pixelated.” So this means that it couldn't possibly look as good as what I see in the preview, with the smooth anti-aliasing, correct ? Is this about the best that can be done with DVD subtitles ?

It's already better than before (and the italics are respected).

But, new problem : the subtitles thus created get progressively out of sync, they appear sooner than they should. (Same if the SUB/IDX subtitles are loaded separately, into VLC Media Player or MPC-HC.) Why ? How to fix this ?
In the IDX file I can see that the last subtitle is set to appear at 00:51:27:021, yet in the SRT file it's set to appear at 00:51:30,108. The timecodes for the first one are almost identical : 00:00:02:267 / 00:00:02,269.

EDIT : And strangely, if I re-load the SUB/IDX into SubtitleCreator, the timecodes displayed are exactly the same as in the SRT...


EDIT : It gets sort-of synchronized by converting the framerate in SubtitleCreator from 30 FPS to 29.97 : it's no longer progressively out-of-sync, only slightly in advance throughout the whole video. The same effect is seemingly obtained by selecting “PAL” instead of “NTSC” under the “Format” menu, but then the placement of the subtitles on the frame is all wrong, being based on a 576px height. According to the author, this is “not a bug, but feature” : “Not a bug but feature. SC applies drop-frame when NTSC in Display menu is selected. If you want to get non-drop NTSC subtitles choose PAL before saving.” What is this supposed to mean ?

EDIT : So apparently the workaround described here works to get a perfect synchronization :
– creating a set of SUB/IDX files in NTSC mode
– creating a set of SUB/IDX files in PAL mode
– then associating the PAL IDX with the NTSC SUB...
And the position of the subtitles on the frame is corrected by editing the IDX to change the frame size from “720x576” to “720x480”...
I only spend half a night trying to solve this mess. é_è


--------------------------

So, what about that other mess regarding AVStoDVD and the framerates with hard or soft pulldown ?

[davexnet]

AVStoDVD[/url] and the framerates with hard or soft pulldown ?

recreate the problem with only one title then post the a2d log and the Avisynth script

[abolibibelot]

I already explained what the problems were, to the best of my knowledge, but here :
AVStoDVD test.a2d.txt
For this test I imported one of the Avisynth scripts with the ChangeFPS(30000,1001) command (below), and the corresponding source video (FLV 30 FPS).

WendyV = LWLibavVideoSource("E:\TEMP\MrT BSOBSF\YouTube - Mr. T on The Wendy Williams Show 10-13-2010.flv").Lanczos4Resize(720,480)
WendyA = LWLibavAudioSource("E:\TEMP\MrT BSOBSF\YouTube - Mr. T on The Wendy Williams Show 10-13-2010.flv")
AudioDub(WendyV,WendyA).ChangeFPS(30000,1001)

The Avisynth source is correctly identified as 29.97FPS in the "FPS" and "FPSAVS" fields, but appears as "29" in the "VideoFPS" field and in the GUI. Then AVStoDVD insists on adding those unnecessary conversions : ConvertToYV12, AddBorders, Spline16Resize, and MCJMFPS. And the aspect ratio is identified as 1.5 based on the 720x480 conversion in my script (actually I'm realizing just now that this one was already in 720x480 despite being downloaded from YouTube — but with a wrong 3/2 AR — so that conversion wasn't even necessary {*}), but I can't change this to the correct 4:3 value no matter what I type in the "Video Display Aspect Ratio" sub-menu. As I mentioned earlier, it seems to work if I edit the .a2d file and type "1.33" wherever there's "1.5" then reload it.
Strangely, the 30FPS FLV is identified as 29.97FPS in the "FPS" and "FPSAVS" fields. And AVStoDVD still wants to convert the framerate, but with "ChangeFPS(29.97)" instead of "MCJMFPS(29.97). By the way "29.97" doesn't seem as accurate as "30000,1001".
As already mentioned, I managed to get (I think) the expected result by removing those unwanted conversions in the Avisynth tab for each one of the "titles" — seems to work, but quite tedious, and not exactly foo’ proof.

Then I tested an encode of a 25FPS clip with HCEnc followed by a conversion with DGPulldown, and the result seems to be exactly the same : every media player or analyser or editor identifies the output as 29.97FPS, either displaying the partially interlaced pattern (VirtualDub2) or not (VLC Media Player, SMPlayer, MPC-HC, Avidemux). So is this indeed to be expected ? How can I verify that a 29.97FPS file is really 25FPS with pulldown flags if it's reported everywhere as 29.97FPS ?

Now, for those subtitles issues, I managed to remux the VOB or MPG videos with the SUB/IDX files exported with SubtitleCreator (using the trick described above to get the proper synchronization despite that pesky nonsensical "drop-frame" feature) with MKVToolNix, then replaced the VOB or MPG videos with the MKV ones in DVDStyler, then generated the DVD... And it seemed to work fine, but I got this :

While the .vob files in the "dvd-cache" folder display the subtitles normally :

How could this be explained, and fixed ?
I tried to find an explanation in the log file but couldn't find any.
dvdstyler 6 (aspect aberrant des sous-titres, et non activés par défaut).log
Also, the subtitles are not active by default (I want them to be).

By the way, on the subject of font choice for subtitles, I came across this excellent article, well structured, well written, well worth sharing :
https://www.md-subs.com/saa-subtitle-font
(All this research to choose Arial Bold in the end... é_è)



{*} I'm wondering what Avisynth does in such a situation, when a source video is in 720x480 and a script contains a XXXResize(720,480) command : does it do nothing, considering that the source is already in the target resolution, or does it proceed anyway, with some quality loss ? If it's the latter, what would be the net effect with Lanczos4Resize, a slight blurring, a slight sharpening, some added artifacting, or nothing noticeable ?