Toast vs. Compressor & DVD Studio Pro

I'm making DVDs from an 80 minute DV-PAL Quicktime movie. Toast 8 has yielded only bad looking DVDs. The Quicktime movie is 16 GB and looks great so I don't see why a 4 GB DVD must look awful.
I have Compressor 2.1 and DVD Studio Pro 4.1 but have so far been unable to make them work. Is there a simple tutorial for using them -- no menus, nothing fancy, just to make a DVD with a quality picture?

Dennis Couzin
Berlin, Germany

The quality of the DVD is dependent on how you encode the MPG-2 video and audio. If you just drop the .mov (Quicktime) file into DVD Studio Pro, you will get an average looking DVD.

Using Compressor is the key to getting great looking video. Toast is only useful for burning copies of DVD's with the Video_TS folder.

Toast seems to be quite simple to use (but i avoided using it due to the lack of control over encoding settings).

>The quality of the DVD is dependent on how you encode the MPG-2 video and audio.

I agree with John Foley on this. A good knowledge of encoding Mpeg2 is important to getting that video onto DVDs. A VOB is just another wrapper for Mpeg2. I think a lot's been posted on this.

Thanks, I can understand how exporting straight from FCP to Compressor without passing through a DV Quicktime movie might yield a better MPG-2 compression, but I don't know where your certainty comes that your encoding settings will be smarter than those of Toast 8. Setting bitrates seems the big deal, and there's much vague writing on it, but Toast will set them as high as possible subject to the constraints of fitting on a DVD and the DVD being generally playable. Anyhow my problems come after Compressor. I can't get a playable DVD from Compressor's output with DVD Studio Pro. Guess I must read that yuk.

Dennis Couzin
Berlin, Germany

> Is there a simple tutorial for using them -- no menus, nothing fancy, just to make a DVD with a
> quality picture?

Delete the menu from the Outline pane, select the disc, and set first play to track 1, chapter 1.

Thankyou Strypes. It worked! I also had to lay the imported .m2v onto track V1 and the imported .ac3 onto track A1. This yielded a playable the DVD with stereo sound. The picture is still noisy. I'm not sure the picture is better than a Toast8 DVD. This was the original question. I used bit rate settings 6.5 and 8.5 in Compressor, and this resulted in a 4.1 GB DVD. Do I dare to use somewhat higher bitrates?

Dennis Couzin
Berlin, Germany

> I used bit rate settings 6.5 and 8.5 in Compressor, and this resulted in a 4.1 GB DVD.

I'll push the bit rate setting up and down to 6.8/8.0 for better overall compatibility with players. If you're doing this, make sure your video is under 90 mins. Also, select an "open" GOP.

Are you exporting a QT or exporting it straight to Compressor off the timeline? Exporting it straight to Compressor allows Compressor to render straight off your timeline. Exporting a QT movie from Final Cut renders all elements to the sequence codec.

strypes, thanks for your continued advice.
I happened to make a 7.0/9.0 DVD and I could see its superiority over the Toast8 DVD, especially as less noisy blacks. I agree that a 7.0/9.0 DVD invites problems with various players, and will next make a 6.8/8.0. The video is just 80 minutes. I'm exporting from FCP to Compressor and then using DVD Pro to mux and burn. I'll select "open GOP" in Compressor.
There are too many MPEG-2 parameters in Compressor for me to succeed by trial and error.

Dennis Couzin
Berlin, Germany

A general guideline... I used to think that a short IP GOP setting would have been better for the video, but after a bunch of trial and errors, I've come to realize that even if you minimize on interframe compression, and due to the upper bitrate limit on dvds, compromising on interframe compression means adding intraframe compression- your I-frames get compromised, resulting in lower quality than you'll like, since the rest of the GOP has to work off the information from the I-frame. The cons is having long dissolves blocking out, and noise around dark areas. So you have to be careful with that.

My current conclusion is a 2 pass, open 15 frames IBBP setting, with compression markers carefully marked out. The benefits of having 2 passes is pretty clear- it takes twice as long, as it only starts proper encoding on the 2nd pass, but it brings a noticeable difference in quality.

An additional method you could try out- is to export a QT movie, include compression markers and set it as 10 bit uncompressed (if you have the disk space). Then import that into compressor and see which is better. Exporting straight off your timeline allows compressor to render your clips straight off your timeline (so rendered clips, text and graphics are not rendered to your sequence codec), but recently I've found Compressor messing around with my Boris text 3d. One alternative is to render it out uncompressed and see how final cut does in converting footage to uncompressed- text should look better especially if the sequence codec is in DV.

The 10-bit uncompressed QT is about 120 GB which I can deal with. I used this method when making a DigiBeta tape from the DV PAL project. The results were unwonderful. FCP needs to fudge the uncompressed data, and it seems to flatten the upper end of the tone scale. Also it introduced banding artifacts in some colored titles. Also it could not make the lower field dominant.

DV PAL has lower field dominant, but regardless of how I set the sequence settings the 10-bit uncompressed QT comes out with upper field dominant. My FCP 5.1.2 did it and the laboratory's FCP 6 did it too. The laboratory was able to swap fields when making the DigiBeta, but the 10-bit uncompressed QT is not usable for much else. Should I have made a de-interlaced 10-bit uncompressed QT?

If this is an FCP glitch, someone more expert than I should report it in another forum.

The DVDs I'm making now are for noncommercial use (although I want to make them as good as I can). The distributor will be responsible for the final mpeg-2. But I must provide him with a good QT!

Dennis Couzin
Berlin, Germany

> The 10-bit uncompressed QT is about 120 GB which I
> can deal with. I used this method when making a
> DigiBeta tape from the DV PAL project. The
> results were unwonderful. FCP needs to fudge the
> uncompressed data, and it seems to flatten the
> upper end of the tone scale. Also it introduced
> banding artifacts in some colored titles.

I wasn't so sure about changing the compression setting within FCP, or FCP's ability to handle codec conversion. I should some tests on this. Currently I'm pretty certain that I will not use FCP's QT conversion to export a mpg4/h.264 file as the quality is an issue when compared with QT pro/compressor. (more combs on interlaced mediums, less sharp image). I'm not too sure about decompressed footage or footage exported with the export QT movie option, though...

>Also it
> could not make the lower field dominant.

No no. Try to keep it native to the codec (upper field dominant in our case), as various hardware/software might not be able to cope with a shift in field dominance. So when switching between DV PAL and SD PAL, you need to shift the fields. I was surprised Compressor didn't do it instinctively- when i export a DV PAL into uncompressed, it needs to understand that it's a lower field dominant format, and it needs to shift fields before it converts it to SD PAL.

>
> DV PAL has lower field dominant, but regardless of
> how I set the sequence settings the 10-bit
> uncompressed QT comes out with upper field
> dominant.

I think i get your picture. Does the footage suffer from field tearing? I usually add a shift fields filter before exporting. Final Cut messes up when you're working with speed ramps or mixed media. I'd want my SD QT export to be upper field dominant, with no field tearing when i export DV footage into that codec.

My FCP 5.1.2 did it and the
> laboratory's FCP 6 did it too. The laboratory was
> able to swap fields when making the DigiBeta, but
> the 10-bit uncompressed QT is not usable for much
> else. Should I have made a de-interlaced 10-bit
> uncompressed QT?

No. Keep the interlace. Only de-interlace for purposes that do not require interlace (web cast/ film screenings, etc). De-interlacing reduces quality.

>
> If this is an FCP glitch, someone more expert than
> I should report it in another forum.
>

I think field misinterpretation has been an issue with Final Cut. I was concerned about the

> The DVDs I'm making now are for noncommercial use
> (although I want to make them as good as I can).
> The distributor will be responsible for the final
> mpeg-2. But I must provide him with a good QT!

What kind of QT? The word has been to export it uncompressed. (i suppose through Compressor). My advice is to include compression/dvdsp markers, as with these markers, compressor reads where the edit points are, so it adds I-frames at those points, irregardless of where it is on the GOP.

Thank you again strypes for the detailed reply.
Previously I used FCP's Export > QuickTime Movie and Export > Using Quicktime Conversion to make 10-bit uncompressed. (For the first way I changed the FCP sequence settings to 10-bit uncompressed.) I found the first way to yield better color/tonality than the second way. Now I will try your suggestion to Export > Using Compressor.

I am new to video -- my background is in film -- so I'll recite for correction how a beginner understands fields and frames. My PAL camera takes 50 fields per second, the first consisting of even lines, the second consiting of odd lines, even , odd, even, odd, etc. The fields have their definite temporal order, and if the fields were recorded separately and kept separate, there could never be the problems FCP caused for me. So, either in the camera or in the FCP "digitization" pairs of consecutive fields must be combined to make 25 frames per second and just a header added to tell which field in each frame came first. These frames are useful for editing, but they are ugly things with every other line being 1/50 second out of synch. If downstream the header becomes wrong and the display device shows 50 fields per second, there will be big stutters seen with motion. If the display device wants to show 25 frames per second it can massage the ugly frames into better ones, but this also requires correct header information. Likewise the display device can calculate 50 fairly decent frames per second. Everything requires that the original field order not be lost. It's such a simple thing to mark a frame "even field dominant" or "odd field dominant" that I don't see how FCP can get it wrong. I hope Compressor is better.

Dennis Couzin
Berlin, Germany

>Previously I used FCP's Export > QuickTime Movie and Export > Using Quicktime
>Conversion to make 10-bit uncompressed. (For the first way I changed the FCP
>sequence settings to 10-bit uncompressed.) I found the first way to yield better
>color/tonality than the second way. Now I will try your suggestion to Export > Using
>Compressor.

Compressor renders straight off the FCP timeline, so as to ensure that footage do not suffer a 2nd round of compression. Which would be the case if you have added filters or exported via qt conversion. This was helpful in the case of burning DVDs- your text and graphics would not have to be rendered into your sequence codec.

Exporting into a different codec in Final Cut or Compressor technically shouldn't make a difference. But with compressed end user formats (eg h.264/mpg4 exporting via QT conversion in final cut doesn't yields noticeably lower quality than if i exported via Compressor or if i exported a qt movie and encoded that in QT pro, same bitrate, ceteris paribus). This however, makes some sense, as Compressor is a dedicated software for encoding, and that's where it gets its bang for its buck- the ability for stable high quality transcoding. Hell, it's powerful, you can simultaneously perform motion compesated de-interlacing, resize and crop your footage, as well as apply some filters to tweak the colors back for encoding, all in the same process without losing a generation (deinterlace in final cut, how the hell do u crop something to action safe then resize it to the size you want in final cut anyway?)

The only issues I actually have with Compressor is that exporting directly off the timeline messed up the aspect ratio of my fonts once (Boris text, on dvcproHD 720p50 codec). The preview didn't even show it until after the encode, it popped up as a "surprise!" Exporting a QT movie and encoding that in compressor doesn't. I've tried exporting that same clip quite a few times with the same results. Hmm... maybe i could post the project file and someone could check if it was a bug with Boris Text exports from Compressor.

The next issue is that it doesn't encode a H.264 in an mpeg4 wrapper. Also, I've been realizing that QTpro 7.4.1 does a marginally better job at encoding h.264 than Compressor. (Which also could mean that Compressor is falling behind slightly).


>I am new to video -- my background is in film -- so I'll recite for correction how a
>beginner understands fields and frames. My PAL camera takes 50 fields per second, the
>first consisting of even lines, the second consiting of odd lines, even , odd, even, odd,
>etc. The fields have their definite temporal order, and if the fields were recorded
>separately and kept separate, there could never be the problems FCP caused for me.

You got the interlacing part right. Someone else can provide a better understanding of how the fields are stored. There's also something to do with color space in the storing of interlaced mediums...

What i'm gathering, is that interlaced mediums are only recognized with a header. Both fields are kept together in the same "frame". It's no issue in the analog days- everything was upper field first. Digital is different. Isn't that hard to understand- DV is lower, SD in NTSC is lower, PAL is upper, HD is upper or progressive.

>I don't see how FCP can get it wrong.

In 90% of cases, it doesn't. What it seems to be doing, is that it imports the footage in, checks the codec and the codec would tell it how it should be mapped. If it's interlaced, it will assume that it is interlaced as according to what is native with the codec, applying a shift fields filter if you're inserting that clip into a sequence with a different field order. (one of the reasons why I mentioned that you need to keep to the native field order of the codec)

What it doesn't seem to do- When I exported a PAL DV clip into PAL uncompressed. I ended up adding a shift fields filter in the timeline, effectively causing field tearing in the sequence, but it looks okay after export. (can't remember why i actually did it this way) but anyway if you get me here, Final Cut doesn't seem to recognize that it needs to swap field order when transcoding into codecs that requires a different field order. Maybe i could try batch export and see if it makes a difference...

Having speed ramps in DV clips and working in a mixed timeline might pose some problems as well. The speed ramps seem to render straight to the field order of DV, despite the sequence being set to upper field dominant, uncompressed and a shift fields filter applied to the DV clip.

>I hope Compressor is better.
Cross your fingers and pray. It's only just as intelligent at recognizing field order as final cut is...