Here's an idea! ISO Video Support

[qtipextra]

Now this might be difficult, as I don't even think there is currently any codecs on current PC OS'es that does this.

Basically what I am thinking is support to play DVD ISO image files. This way there is no converting of video files, all a person would have to do is create an image file (even shrink it down to 4.7 GB) and load it on their device running Rockbox, and be able to play it like a DVD video.

We could create a program that would read the contents of the ISO archive file. If the contents contain DVD files, then play it as such.

So basically, all we would need to do is compile an ISO reader, and send it to a DVD codec on the media player.

What do you think?

[linuxstb]

Not even the fastest current Rockbox device (the Gigabeat F) could decode DVD resolution MPEG-2 video in realtime, so there's little point in adding support for DVD images.  This is why you need to re-encoded (and resize) video for Rockbox.

Some future targets (e.g. Gigabeat S) may be able to, but then you'll run into the 4GB maximum filesize on FAT filesystems.  There's no real advantage to using .iso images anyway, it would be easier to just copy the contents of the "VIDEO_TS" folder from the DVD.

[qtipextra]

Not even the fastest current Rockbox device (the Gigabeat F) could decode DVD resolution MPEG-2 video in realtime, so there's little point in adding support for DVD images.  This is why you need to re-encoded (and resize) video for Rockbox.

Some future targets (e.g. Gigabeat S) may be able to, but then you'll run into the 4GB maximum filesize on FAT filesystems.  There's no real advantage to using .iso images anyway, it would be easier to just copy the contents of the "VIDEO_TS" folder from the DVD.

Very true and valid points. The GB S is the specific target I had in mind for now as it has some really good hardware specs, and over at it's current "new ports" thread, it is currently possible to run Rockbox on it. Very good point about the DVD folder vs ISO. I personally like ISO as I can keep things better organized, however it is not necessary, but it wouldn't take much more code to implement ISO image reading. Another great point about the 4GB file size limitation on the FAT file system. GB S has a FAT32 file system, but the 4GB size limitation is still present. The only way would be to implement NTFS file system, which, of course, is a big NoDo, but still may be possible on some devices.

[Llorean]

It would make MUCH more sense just to copy the .vob file off the player for the title you're interested in. This is often split into 2GB files anyway, when presented to the host computer. Then you'd just have to have mpegplayer play these (this would require adding support for AC3 audio and selecting the appropriate audio stream), play DVD video (scaling and correcting aspect ratio) and move on to the next .vob file in the series if the title is split among multiple.

Then just organize it in folders named after the video. It's not quite as nice as ISO files but it means you don't have to deal with the filesystem limitation, you don't have to care about detecting which title on the DVD is the one to actually play, and you don't have to code support for DVD menus (or title selection) if you decide guessing is a bad idea (which it really is)

[qtipextra]

It would make MUCH more sense just to copy the .vob file off the player for the title you're interested in. This is often split into 2GB files anyway, when presented to the host computer. Then you'd just have to have mpegplayer play these (this would require adding support for AC3 audio and selecting the appropriate audio stream), play DVD video (scaling and correcting aspect ratio) and move on to the next .vob file in the series if the title is split among multiple.

Then just organize it in folders named after the video. It's not quite as nice as ISO files but it means you don't have to deal with the filesystem limitation, you don't have to care about detecting which title on the DVD is the one to actually play, and you don't have to code support for DVD menus (or title selection) if you decide guessing is a bad idea (which it really is)

Yeah, another good point. Almost all DVD video files are split into multiple files so you would not have to worry about the 4 GB size limit. So really all we would need is the AC3 audio and DVD video (.VOB) support.

[gevaerts]

Yeah, another good point. Almost all DVD video files are split into multiple files so you would not have to worry about the 4 GB size limit. So really all we would need is the AC3 audio and DVD video (.VOB) support.

dvd video is easy. That's plain mpeg2 which we support. The main issue will be scaling

[Llorean]

If I understand, our AC3 isn't the fastest codec on the block either. So we're decoding 4.5 times the number of pixels we normally do and decoding a slower audio codec. So performance is definitely going to be a significant question as well.

[Chronon]

It's not covered in the CodecPerformanceComparison and SoundCodecs claims that it works "very efficiently".  Do you have some idea of how fast it is compared with some of our other, better tested codecs?

[Llorean]

No, that's why I said "If I understand." I'm just going off the impression I got from talking with amiconn earlier. But I expect that just getting DVD-resolution MPEG-2 playing realtime is going to be a considerable hurdle. We have approximately twice the speed of the Gigabeat F and we need four times the pixels plus scaling. Not to mention many DVDs have multichannel AC3 (5.1 surround) which, even if AC3 is as efficient as MP3, means we have to deal with quit a bit more data.

[saratoga]

The S's CPU is tremendously faster per clock then the F's, and it does have a coprocessor that can probably handle the scaling without bothering the main CPU.  Its certainly possible to do this, but a lot of work.

Adding scaling to mpegplayer would be a great first step.

[RotAtoR]

Not to mention many DVDs have multichannel AC3 (5.1 surround) which, even if AC3 is as efficient as MP3, means we have to deal with quit a bit more data.

I believe on most DVDs the AC3 audio is going to have 48kHz sample rate, so it'd currently need to be resampled as well.

[delt]

.... or, maybe porting ogle ( http://freshmeat.net/projects/ogle/ ) to all platforms running rockbox?

</smartass_mode>

[GodEater]

I'd like to point out that there's another overhead here as well - which is cracking the CSS which is in the .VOB files - we'd need a port of libdvdread (or similar) in order to do that, and it's yet more processing that needs to be done.

[Llorean]

I think the assumption is that the .vob will be unencrypted by the time it gets to Rockbox (we'll not discuss it further, and shall assume that we're talking about homemade DVDs and/or the few ones sold encryption free).

[cool_walking_]

Wouldn't including CSS decryption violate the DRM playback NoDo?