Best Compressor 4.4 Settings for FCP 10.3.3 Project?

Hey there guys,
I hope that this post finds you all in excellent spirits, and that you are having a great C&NY period.

We are now in the process of compressing the 110GB Master File in Compressor 4.4.

We were just wondering if any of you had any thoughts on the best video compression settings for a dual-layer DVD (movie duration: 130 minutes).

We are trying the following settings which we got from the book DVD Studio Pro 4: Third Edition:

Two-pass VBR (best)

Average Bit Rate: 4.5

Maximum Bit Rate: 7.5

Dialogue Normalisation: -31dBFS

Our main concerns are:

1. That we get the highest video quality possible which will still fit on a dual-layer DVD; and

2. That the Dialogue Normalisation setting will not interfere with the music and sound effects in the film.

As ever, any guidance is sincerely appreciated. Thank you for taking the time and trouble to read this!

FYI, we have just completed two attempts at compressing the Master File:

1. The first attempt - with the automatic bit rate settings and single-pass BVR - came in at 3.77 GB and was way, way too blurry.

2. The second attempt had an average bit rate of 4.5 and a maximum bit rate of 7.5. It comes in at 4.42 GB but is still too blurry.

Any ideas on the best settings? We have around 3 GB to play with, because the sound is coming in at just 188 MB.

AlexMaxim wrote: ....best video compression settings for a dual-layer DVD (movie duration: 130 minutes)...We have around 3 GB to play with, because the sound is coming in at just 188 MB.

I don't understand the 3GB restriction since a dual-layer DVD should hold 8.5GB. Is it possible you are only burning it as a single-layer?

I've never burned a dual-layer DVD but here's some info from DVD Studio Pro about this: documentation.apple.com/en/dvdstudiopro/...tion=10%26tasks=true

joema wrote:

AlexMaxim wrote: ....best video compression settings for a dual-layer DVD (movie duration: 130 minutes)...We have around 3 GB to play with, because the sound is coming in at just 188 MB.

I don't understand the 3GB restriction since a dual-layer DVD should hold 8.5GB. Is it possible you are only burning it as a single-layer?

I've never burned a dual-layer DVD but here's some info from DVD Studio Pro about this: documentation.apple.com/en/dvdstudiopro/...tion=10%26tasks=true

Yes, you are absolutely right: a dual-layer DVD has a capacity of around 8.54 GB, although apparently it holds only 7.95 GB, as your excellent link confirms.

What I meant to say was that there is an additional 3 GB on the disk that we could in theory use, because an average bit rate of 4.5 and a maximum bit rate of 7.5 produced a 4.42 GB movie file. Does that make sense?

Yes. However I don't understand why an average 4.5 and max 7.5 mbps rate would be that blurry. The DVD MPEG-2 encoded standard def bit rate is often around 6mbps, it seems in line with that. If you have the option of encoding the master file to an MPEG-2 file without burning it, that might provide faster turnaround on tests, plus you could peek at the quality at that intermediate point.

Another troubleshooting approach might be to take a smaller piece of the master file -- IOW encode from Compressor 60 sec of it, then go through DVD Studio Pro at the same output bit rate to an MPEG-2 file and examine that. This would provide very rapid turnaround on trying different parameters without re-encoding that huge file each time and burning the DVD.

joema wrote: Yes. However I don't understand why an average 4.5 and max 7.5 mbps rate would be that blurry. The DVD MPEG-2 encoded standard def bit rate is often around 6mbps, it seems in line with that. If you have the option of encoding the master file to an MPEG-2 file without burning it, that might provide faster turnaround on tests, plus you could peek at the quality at that intermediate point.

Another troubleshooting approach might be to take a smaller piece of the master file -- IOW encode from Compressor 60 sec of it, then go through DVD Studio Pro at the same output bit rate to an MPEG-2 file and examine that. This would provide very rapid turnaround on trying different parameters without re-encoding that huge file each time and burning the DVD.

We 100% agree with you re: encoding the MPEG-2 master file without burning it. We are exporting the compressed file to an external hard drive. Funnily enough, it's not actually taking that long to encode - between 59m and 71m, depending on how high the bit rates are.

Like you, we have no idea why the picture is so blurry. There is no practical difference at all between the Master File (110 GB) and the 'ordinary' 12.29 GB H.264 file we exported from Final Cut Pro 10.3.3, but the quality difference between the H.264 file and the compressed MPEG-2 file (6.59 GB) is really obvious.

Is it worth trying the 'Send to Compressor' option from within FCP X, or is this simply not a solution to the problem?

To give you an idea of the quality difference:

Above is the compressed MPEG-2 file.

Below is the Master File (the H.264 looks identical):

Attached is the Master File/H.264 still (I can't get it to display).

AlexMaxim wrote: ...we have no idea why the picture is so blurry. There is no practical difference at all between the Master File (110 GB) and the 'ordinary' 12.29 GB H.264 file we exported from Final Cut Pro 10.3.3, but the quality difference between the H.264 file and the compressed MPEG-2 file (6.59 GB) is really obvious....

If both 12.29GB H264 and 6.59 MPEG-2 files are the same 130 min program length, the MPEG-2 file is essentially 1/4 the bit rate. You lose a factor of two in the less efficient MPEG-2 compression and another factor of two from the file being 1/2 the size. There might also be "generational" quality loss from re-encoding an existing compressed file.

It would help if you posted the file attributes of all three files using MediaInfo or Invisor. I personally like Invisor because it allows spreadsheet-like side-by-side comparison of video file attributes. You just drag/drop the files to Invisor and it displays them all in a grid. Or in Finder you right-click on the file, select Services then Invisor or MediaInfo.

MediaInfo: mediaarea.net/en/MediaInfo/Download/Mac_OS
Invisor: itunes.apple.com/us/app/invisor-media-fi...or/id442947586?mt=12

joema wrote:

AlexMaxim wrote: ...we have no idea why the picture is so blurry. There is no practical difference at all between the Master File (110 GB) and the 'ordinary' 12.29 GB H.264 file we exported from Final Cut Pro 10.3.3, but the quality difference between the H.264 file and the compressed MPEG-2 file (6.59 GB) is really obvious....

If both 12.29GB H264 and 6.59 MPEG-2 files are the same 130 min program length, the MPEG-2 file is essentially 1/4 the bit rate. You lose a factor of two in the less efficient MPEG-2 compression and another factor of two from the file being 1/2 the size. There might also be "generational" quality loss from re-encoding an existing compressed file.

It would help if you posted the file attributes of all three files using MediaInfo or Invisor. I personally like Invisor because it allows spreadsheet-like side-by-side comparison of video file attributes. You just drag/drop the files to Invisor and it displays them all in a grid. Or in Finder you right-click on the file, select Services then Invisor or MediaInfo.

MediaInfo: mediaarea.net/en/MediaInfo/Download/Mac_OS
Invisor: itunes.apple.com/us/app/invisor-media-fi...or/id442947586?mt=12

Firstly, thank you so much for your kind and informative response, which is sincerely appreciated!

Re: Invisor, that looks like an amazing app and we will try and get hold of it as soon as we can.

However, we just noticed a couple of things in the Compressor settings which may help:

1. We shot this movie using a Canon 7D in Europe using the PAL standard. However, the Colour Space setting is set to NTSC (Rec.601). Could this be causing a problem?

2. Secondly, under 'Quality' we have now noticed two settings for 'Resize Filter' and 'Retiming Quality'. We have now set these to 'Best'. Surely this would make a difference?

AlexMaxim wrote: ...1. We shot this movie using a Canon 7D in Europe using the PAL standard. However, the Colour Space setting is set to NTSC (Rec.601). Could this be causing a problem?

2. Secondly, under 'Quality' we have now noticed two settings for 'Resize Filter' and 'Retiming Quality'. We have now set these to 'Best'. Surely this would make a difference?

I'm not familiar with those, so maybe someone else can comment.

Another major issue not yet discussed is interlacing. I've only used a DSLR to shoot progressive video, but PAL and NTSC both are interlaced. By definition DVDs are also interlaced. For DVDs the assumption is the playback chain will deinterlace -- ie the hardware player or the software player. However in today's post-interlace era, it is very easy to play an interlaced file on a playback tool which does not deinterlace, and that produces poor visual quality. Yet if that same file is burned to a DVD and played on a hardware player, it may look much better.

So this raises the issue of how you can trust what you see from an image QC standpoint. This is even more confusing since some players (like Quicktime 10) will automatically attempt to deinterlace in when playing some video files but not others.

One way to avoid this uncertainty is using VLC for playback, which defaults to no deinterlacing but has adjustable deinterlace modes. IOW you manually turn on deinterlacing then select various algorithms. This is in VLC menu Video>Deinterlace and Video>Deinterlace Mode.

If you see a significant quality improvement when invoking VLC deinterlacing (esp. on subjects with horizontal movement), that implies the source material is interlaced. Burning interlaced content to DVD is OK and is the standard procedure. Unfortunately when using DVD-targeted media, to get and end-to-end image quality evaluation you must actually burn it then view it on various players: a computer software player, dedicated hardware player and TV, etc. But you can often get a pre-burn assessment of the DVD-targeted material if played in VLC with deinterlacing enabled.

If at any point in the post processing chain the deinterlaced media is hard deinterlaced using incorrect processing it can "bake in" the interlacing so it can't be deinterlaced. That's what you often see on Youtube when comb-like edge artifacts appear. Youtube's normal upload and encoding will deinterlace but only if it can identify the media as interlaced.

Some deinterlace processing if improperly used can discard every other scan line, thus lowering vertical resolution by half.

This is only one possible issue and I'm not saying it's the main factor here. But it is conceivable if you play your current MPEG-2 output file in VLC with deinterlacing enabled and mode = Yadif or one of the other advanced ones, it might look much better. If so this indicates it's just an interlaced file which you've been viewing without deinterlacing, and normally the DVD playback would handle that.

joema wrote:

AlexMaxim wrote: ...1. We shot this movie using a Canon 7D in Europe using the PAL standard. However, the Colour Space setting is set to NTSC (Rec.601). Could this be causing a problem?

2. Secondly, under 'Quality' we have now noticed two settings for 'Resize Filter' and 'Retiming Quality'. We have now set these to 'Best'. Surely this would make a difference?

I'm not familiar with those, so maybe someone else can comment.

Another major issue not yet discussed is interlacing. I've only used a DSLR to shoot progressive video, but PAL and NTSC both are interlaced. By definition DVDs are also interlaced. For DVDs the assumption is the playback chain will deinterlace -- ie the hardware player or the software player. However in today's post-interlace era, it is very easy to play an interlaced file on a playback tool which does not deinterlace, and that produces poor visual quality. Yet if that same file is burned to a DVD and played on a hardware player, it may look much better.

So this raises the issue of how you can trust what you see from an image QC standpoint. This is even more confusing since some players (like Quicktime 10) will automatically attempt to deinterlace in when playing some video files but not others.

One way to avoid this uncertainty is using VLC for playback, which defaults to no deinterlacing but has adjustable deinterlace modes. IOW you manually turn on deinterlacing then select various algorithms. This is in VLC menu Video>Deinterlace and Video>Deinterlace Mode.

If you see a significant quality improvement when invoking VLC deinterlacing (esp. on subjects with horizontal movement), that implies the source material is interlaced. Burning interlaced content to DVD is OK and is the standard procedure. Unfortunately when using DVD-targeted media, to get and end-to-end image quality evaluation you must actually burn it then view it on various players: a computer software player, dedicated hardware player and TV, etc. But you can often get a pre-burn assessment of the DVD-targeted material if played in VLC with deinterlacing enabled.

If at any point in the post processing chain the deinterlaced media is hard deinterlaced using incorrect processing it can "bake in" the interlacing so it can't be deinterlaced. That's what you often see on Youtube when comb-like edge artifacts appear. Youtube's normal upload and encoding will deinterlace but only if it can identify the media as interlaced.

Some deinterlace processing if improperly used can discard every other scan line, thus lowering vertical resolution by half.

This is only one possible issue and I'm not saying it's the main factor here. But it is conceivable if you play your current MPEG-2 output file in VLC with deinterlacing enabled and mode = Yadif or one of the other advanced ones, it might look much better. If so this indicates it's just an interlaced file which you've been viewing without deinterlacing, and normally the DVD playback would handle that.

This is a brilliant idea - thank you!

We are in fact using VLC Media Player for testing purposes, and this is indeed where the losses in quality are being detected.

However, we are currently encoding the same file using different (more exacting) Colour Space and Quality settings and it is taking reassuringly longer to encode: despite having a similar estimated size (6.68 GB), it has taken over 2 hours and 35 minutes so far - and it is only about 60% done.

With any luck, it should have completed encoding in around 2 hours' time. I will post an update here as soon as it is done.

Okay! The file has finished encoding. The total time was 4 hours, 12 minutes and 47 seconds - over three hours longer than the previous encoding operation.

The result? The quality unfortunately is just the same as on the quicker encoding settings. It is like nothing has changed. Still way too blurry. I've never seen a DVD look so low-resolution

The encoding settings are attached - at least, I have tried to attach them.

I have a feeling that there is something wrong with the Quality settings, specifically the anti-aliasing and details options, but I need to do some more research on that front.



Any input would be gratefully accepted!