DivX playback on DVD Player not working

[Jawor]

OK, they were always saying that they're certifying hardware for “DivX playback”, not “MPEG-4 playback”. But when DXN nailed down the specs for their “Certified” profiles, capabilities of already existing hardware were the real limitation (some features of their then-current codec, like Qpel and 1wp-GMC, had to be omitted). Besides, by not introducing any proprietary extensions to MPEG-4 SP/ASP (not counting Packed Bitstream), DXN took advantage of the “network effect” (already then, Xvid was more popular than DivX and that popularity must have given a major boost in sales of MPEG-4-capable hardware). The hardware certification program generated substantial income for DXN while their software products (“DivX Pro”, old “Dr. DivX” and that spectacular failure called “MasterMind” or sth like that) were quickly falling into obscurity ;)

Even some encodes generated with DivX may prove incompatible with their “Certified” hardware due to simple bugs in (some older versions of) DivX's VfW GUI (for example, I remember a DivX version that allowed the usage of MPEG quantization in profiles that don't support it).

[Ghitulescu]

... DXN took advantage of the “network effect” (already then, Xvid was more popular than DivX and that popularity must have given a major boost in sales of MPEG-4-capable hardware) ...

Exactly my point - people converting DVDs to Divx (and especially those that didn't do this themselves :p) helped DXN earn a lot of money. As a corollary, they also made people not converting pay more for features they don't need, but are trendy.

[kurkosdr]

I can't contest your assumptions as I do not have Divx videos. However, once a company produces and licences both the hardware specs and the encoder/codec, and this company is well known monopolist [supporter] I wouldn't be surprised to see a success rate of 100% for their videos vs. less than 100% for competitors.

Divx Inc never intended their Divx-certified players to only play videos made with Divx Inc software (unlike Microsoft with WMV, which does want WMV files to only be generated by MS software) because their format would fade into obscurity.

Instead Divx Inc keeps the basic specs open, but hides extensions like the DVD-like menus and the rest. That way, they make money from device certifications (since only they know all the parts of the format) and professional authoring software they sell, but this doesn't stop people from encoding videos for their certified devices.

Divx Home Theater specs have been known for quite a while (and been implemented by xvid home profile), and Divx Inc also have instructions in their site on how to use x264 to produce a Divx Plus HD compatible file.

The problem is caused by lack of interest by the AutoGK and the avidemux folks, otherwise they would have a "force xvid home" button in their programs.

Can anyone post the settings i need to make to AutoGK to achieve a 100% compatible Divx Home Theater/Xvid Home clip?

[Jawor]

Divx Inc never intended their Divx-certified players to only play videos made with Divx Inc software because their format would fade into obscurity.
Just like the aforementioned WMA/WMV and Nero Digital did :p

DivX's DMF container isn't really “closed” (the DMF SDK was released to the public) and yet its “extended” features are not supported by anything except certified hardware and the DivX Plus Player (DXN's DirectShow filter for XSubs was discontinued... and AFAIR it worked only on XP). Sure, we can create XSub subtitles (the most successful part of DMF) with a number of third-party freeware tools, but playback is a different matter (did anyone manage to do it on Linux? ;) ). Freeware/OSS programmers were not willing to promote something proprietary (even if it's relatively “open”), especially for DXN (keeping in mind the OpenDivX/encore2 affair). The multi-audio part of DMF works everywhere, 'cause it was already supported by some AVI muxers (like VDM) and splitters.

Look at Nero Digital... Using MP4 instead of AVI and AAC instead of MP3 (with MPEG-4 SP/ASP video) simply killed their whole certification program (besides, “DivX” was already a household name and almost nobody heard of Nero Digital). A great example of a missed opportunity for using the network effect :p It was like picking a fight with the neighborhood's strongest bully...

[kurkosdr]

(did anyone manage to do it on Linux? ;).
Use VLC (latest version). It supports XSUBs. In fact, support for XSUBs existed in VLC for quite a while using an old demuxer. But then Videolan moved to a new demuxer and forgot to port the XSUB decoding feature, and was only ported very recently. Just don't ask for DVD-like menus.

Now, can we get back on topic of how to produce divx compatible encodes using non -Divx Inc programs?

[Jawor]

How do I enable two pass encoding in AutoGK, plus any other things I may need to chance.
According to the manual (http://www.autogk.me.uk/modules.php?name=TutorialEN), it is sufficient to check Enable standalone support and ESS under “Hidden Options” and then select a target size (instead of target quality) to force two-pass encoding (that makes perfect sense, since Xvid uses VBV only in two-pass mode and two passes are required to hit a target size).

Target quality basically means “single-pass constant quantizer encoding”, so no wonder that the bitrate can spike above the “DivX Home Theater” limit.

[hello_hello]

I can't contest your assumptions as I do not have Divx videos.

I don't own any aeroplanes but I don't let that stop me logging into aviation forums and offering my opinion either.

However, once a company produces and licences both the hardware specs and the encoder/codec, and this company is well known monopolist [supporter] I wouldn't be surprised to see a success rate of 100% for their videos vs. less than 100% for competitors.

It doesn't surprise me you'd think that without offering anything but mere speculation. It's mpeg4 video. Either the decoder can decode it or it can't.
All DivX capable DVD players aren't created equal, they've just got to comply with the DivX specification (probably the home theatre profile). If they can handle higher bitrates, great. If they're not bothered by global motion compensation, yippeee. If they can handle 27 consecutive BVOP's, terrific. If they laugh in the face of custom matrices, fine. If they'll decode anamorphic video with the correct aspect ratio, it's party time.
But if a DVD player is DivX certified and if it's guaranteed to play any mpeg4 video which is encoded within certain limitations, usually a DivX profile such as home theatre, then as long as you restrict the encoder to those home theatre settings, the decoder should be able to decode it. If you don't know exactly what your player can decode then you're 100% safe sticking to the basic home theatre profile. If you don't stick to the home theatre profile restrictions then it doesn't matter which encoder you use in respect to whether your player will decode it.

It's like MS-DOS vs. DR-DOS, or Netscape vs. IExplorer. Both of them fulfilled the public specs laid out by MS, yet some applications couldn't work on DR-DOS or Netscape.

It's nothing like that. Microsoft deliberately went out of their way to sabatage other versions of DOS and if Netscape didn't render properly it's probably because the page used proprietary Microsoft IE code. There's nothing to suggest DivX have ever tried similar tactics.

[hello_hello]

Exactly my point - people converting DVDs to Divx (and especially those that didn't do this themselves :p) helped DXN earn a lot of money. As a corollary, they also made people not converting pay more for features they don't need, but are trendy.

How that was your point, is beyond me.

[hello_hello]

Yes, Xvid can use VBV only when encoding in two passes.

But it looks like DivX can do it also in constant quantizer mode (that would mean the specified target quantizer is overridden for sections that violate VBV requirements).

If you're correct, then it's disappointing someone hasn't taught XviD to do that.
It still doesn't change my earlier argument of all compliant encodes being playable on all DivX certified players, even if it's easier using the DivX encoder. Maybe that's the solution for kurkosdr if he doesn't want to do 2 pass encoding with XviD.


Now, can we get back on topic of how to produce divx compatible encodes using non -Divx Inc programs?

I guess... as long as those who don't encode refrain from interrupting the thread with off-topic speculation. ;)

I answered your question in detail in the AutoGK thread. Although Jawor seems to think the DivX encoder can control VBV while running single pass encoding. If that's correct, then maybe DivX is your answer, or it's 2 pass encoding with XviD.

I haven't used the DivX encoder in years. Anyone know if there's any major quality differences compared to Xvid?

[Jawor]

Although Jawor seems to think the DivX encoder can control VBV while running single pass encoding. If that's correct, then maybe DivX is your answer, or it's 2 pass encoding with XviD.
A quick test confirmed my suspicions - VBV can be used by DivX in constant quantizer mode (we just have to select the profile). AVINaptic clearly shows that in my test clip VBV values for the selected profile are respected (and quantizers were simply increased where it was necessary to satisfy VBV requirements).

I haven't used the DivX encoder in years. Anyone know if there's any major quality differences compared to Xvid?
The quality difference isn't as huge as it was many years ago, but Xvid can still beat DivX, even when using the feature-restricted “DivX HT” profile (and the slowest motion search algorithms in both codecs). Supposedly DivX's multithreading scales much better than Xvid's on 4+ cores... I can't test that, I've got only 2 :mad:

[kurkosdr]

I was hoping that Xvid would have a "two pass constant quantizer" (aka "two pass set quality percentage") mode, but since it doesn't i ll go with two pass target size.

[Jawor]

The two-pass mode's goal is to reach constant quality at the specified target size and within VBV constraints. Quantizers throughout the movie will be similar (the codec will have to increase them only for sections that otherwise would violate VBV constraints).

[kurkosdr]

Looks like youtube beat us: I found this (http://www.youtube.com/watch?v=vKHX8FGKFgw) and it's quite useful (just set the bitrate to 4000)

[Jawor]

Selecting the appropriate profile in Xvid's VfW GUI will enforce VBV restrictions just as well, but much faster ;) Sure, the capability of using user-specified VBV values can be handy sometimes (that's why I included it the VfW GUI of my Xvid builds), but usually that's too much hassle (almost all built-in profiles use VBV values from “DivX Home Theater” anyway)...

[Chetwood]

Given the relative low cost of todays "high-def" capable network media players along with their ability to play just about any audio, video and container format.
If only! The WDTV Live e.g. has several subtitle bugs (like PGS in m2ts but not in mkv, trasparent outline, etc.) and does not support all losless audio formats.

It's like MS-DOS vs. DR-DOS, or Netscape vs. IExplorer.
As already said it's not like that at all. M$ used asshole underhand tactics to deliberately cripple a competitor's product and got convicted for it (not that it helped those competitiors). Divx seems to have more ethical standards.

Now, can we get back on topic of how to produce divx compatible encodes using non -Divx Inc programs?
So were Staxrip and Handbrake already tested in that regard?

[SeeMoreDigital]

If only! The WDTV Live e.g. has several subtitle bugs (like PGS in m2ts but not in mkv, trasparent outline, etc.) and does not support all losless audio formats.
You're moving the goal posts quite a bit wider now... You don't get support for these with a "disc spinning" DivX compatible stand-alone either!

EDIT: Sufficed to say, as with any hardware playback device, it's up-to the end-user to generate encodes that will play in-accordance with the hardware players specification.

[hello_hello]

Looks like youtube beat us: I found this (http://www.youtube.com/watch?v=vKHX8FGKFgw) and it's quite useful (just set the bitrate to 4000)

I can't speak for every XviD build but the version which comes with AutoGK automatically provides the appropriate VBV values according to the profile you select and doesn't seem to let you change them anyway.
The main difference I can see between following the video's instructions and encoding with AutoGK is you're selecting the target bitrate rather than the target file size, but it's still 2 pass encoding only you're guessing as to what the quality will be.

The video suggests using Average Bitrate 2 Pass encoding while setting the bitrate to 1500 or 2000. Is that really any different to 2 pass encoding while selecting the file size except you're selecting it in a different way? Seems to me like it's a method which also predetermines the quality, while potentially wasting a lot of bits or reducing the quality through restricting it too much.

I've got a bunch of AVIs on my hard drive waiting to be burned. The largest has a bitrate of around 1800 while the smallest has a bitrate of 730, yet they were both encoded to the same quality setting (AutoGK's 75%).

One thing I've never understood is how AutoGK determines the quality after running a compression test or after it's run the first pass. I gather 75% = Q2.7 but I don't know how the quality is determined. I assume the info is obtained from XviD's first pass log, I just don't know how.
Likewise I've never understood why AutoGK is the only GUI I've ever come across which supplies that info. Why is it not standard for a GUI to run a first pass according to your settings then tell you what the quality will be? I guess the same question would apply to x264 encoding as well.

Thinking about it a bit more (and sorry for wandering off topic a bit) why is it not possible to run a 2 pass encode at a selected quality rather than file size or bitrate?
It's my understanding XviD runs it's first pass using Q2 and then adjusts the bitrate etc for the second pass according to the selected file size. If that's correct, why can't it also make adjustments to achieve a selected quality?
I understand as a general rule it'd probably be pointless (might as well just run a single pass encode at a selected quality) but when it comes to VBV control you're forced to run 2 passes, so being able to do so while simply selecting the final quality rather than guessing at a file size would be nice.
Or even the ability to manually calculate the desired quality/file size based on the stats from the first pass, then use the calculated file size for the second pass to achieve the desired quality. Is that possible?

[kurkosdr]

So were Staxrip and Handbrake already tested in that regard?

Handbrake dropped avi support in their latest versions for no good reason whatsoever (or according to the maintainer's "because avi support makes the code too complicated"). And they explicitly stated it's not coming back. Should I try version 0.9.4 from oldversion.com?

I will try staxrip.

[manolito]

Handbrake dropped avi support in their latest versions for no good reason whatsoever (or according to the maintainer's "because avi support makes the code too complicated"). And they explicitly stated it's not coming back. Should I try version 0.9.4 from oldversion.com?

I will try staxrip.

You would have to go back to Handbrake 0.9.3 to get avi support back...:devil:
And AFAIK the developers stated some more reasons why they dropped avi support for good (the avi format is generally deprecated, their avi muxer was buggy and did not support large avi files and so on...)

But Staxrip is a really good alternatve IMO. First of all you have the choice between XviD and DivX. It comes with templates for both codecs, but you would have to tweak these templates a little (e.g. it defaults to MP3 VBR for audio which is problematic). I also found that it is no problem to use a different version of XviD instead of the version which comes with Staxrip (I prefer Celtic Druid's compiles over Koepi's).

Plus it automates your encodes in terms of cropping and optionally resizing for square pixels. You also can choose between different resizers and deinterlacers. And what I like most is that unlike many competitors it includes all the helper software, no chasing the net to download and install all the needed helper apps. (In the case of Nero AAC I am not so sure how legal it is to include it, but it certainly made my life a little easier.) Too bad that right now it looks like it is not maintained any more...:(


Another option would be DivXAuthor from the DivX company itself. Right now they offer a 50% discount on all their software, which means that I can buy it for 8 EUR here in Germany.

This app really originates from TMPGEnc DVD Author. It authors DivX Ultra compatible files (supports menues, chapters and anamorphic), but you can of course turn these features off and specify square pixels. It has not been updated since 2007, so the encoder is a little out of date. But it basically does what it is supposed to do, and at least for me it has always produced files which play on every standalone I have access to.


Cheers
manolito

[kurkosdr]

Since DivX Networks (now DivX inc) licences the divx use, format and logo for standalones, it's no wonder their encodes work the best on them. :p:p:p

Nah, I don't think the reason non-Divx Inc tools produce excessive bitrate spikes is because of lack of documentation (the bitrates and settings to make a Divx-compatible avi file are well known). More like lack of interest by Xvid. Most non-Divx Inc tools use Xvid, which can't stay within the required bitrates in one pass mode (only in two pass mode), and most tools just use the one pass mode (faster) and just don't care about the bitrate spikes. I now switched to Avidemux, with Xvid as the codec, with "Average bitrate (two pass)" set accordingly (see video above).
I use Divx Plus Converter only for fast encodings when I don't care about the quality of a particular video file.

[kurkosdr]

I can't speak for every XviD build but the version which comes with AutoGK automatically provides the appropriate VBV values according to the profile you select and doesn't seem to let you change them anyway. The main d

Then why AutoGK throws a warning that some standalone compatibility options will be ignored if you set it to "target quality"? If it's not the bitrate, which compatibility options are ignored?

Also, if AutoGK's "target quality" mode uses Xvid's constast quantizer (one pass), how can Xvid keep within the required bitrates? (Xvid can't do that in one pass modes, it doesn't even have VBV options for one pass mode)

[Chetwood]

AutoGK uses a non-standard matrix when standalone compatibilty isn't checked.

[kurkosdr]

AutoGK uses a non-standard matrix when standalone compatibilty isn't checked.

This is not the problem. The problem is that if I set the standalone compatibility to ESS (or MTK, i prefer ESS), and then set it to "target quality", and before i even set the target quality, AutoGK throws a message that some standalone compatibility options will be ignored. No mention what those options are, but if "target quality" is indeed using Xvid's "constant quantizer" mode (one pass), I smell it's the bitrate limits that are ignored.

[nixo]

You'd be right. VBV restrictions are ignored in constant quantizer mode – which is what AutoGK uses. This is explicitly stated under 'Level' in the encoder configuration.

If an encode absolutely must play on a standalone, it's either 2 passes or Divx.

--
Nikolaj