[Glenn Camhi]

So to ask the question that always gets asked yet again, but in more detail…

Source footage is 1080 24p (23.976, obviously, just shorthanding).

Final cut is same.

For the DVD…

Is there any advantage to converting to 29.97 interlaced? That’s what the post house did in the encode, but I thought we should keep it 24p. Might that play better on some DVD players, even nowadays?

Here’s a more specific question:

For anyone who has an HDTV (most folks now) and a current DVD player (say, PS3), would a DVD authored at 24p play directly as 24p without any conversion/interlacing?

And thus, would it look better than if the footage had been encoded as 29.97i?

It’s a fairly major post house, and the folks I spoke to there said I was wrong that most Hollywood DVDs are 24p. Am I? If they’re not 24p, why does advancing one frame always show a new frame, whereas advancing one frame on the interlaced DVD they gave me only shows a new frame every 2nd and 5th time?

Thanks for any expertise here!

[Glenn Camhi]

Yeah, it is odd. Thanks for confirming.

Can’t imagine what could be happening in the export of the AIFF file, but something is.

[Glenn Camhi]

Thanks Mark. Weird, eh? Could Stereo Downmix have caused an issue? Maybe that’s the version we’ve been testing most recently? I’ve tried both. (Though I thought Channel Grouped does a stereo downmix anyway with stereo pairs.)

[Glenn Camhi]

Thanks again, Angelo.

For the record, I got the 1-pass CBR version via Sonic from the post house, and have a verdict:

It looks MUCH better than what Compressor produced. Mostly. There are some areas where it fares a bit worse: most notably, thin lines (like phone wires) and computer-generated graphics exhibit very noticeable aliasing that’s less pronounced or wholly unnoticeable in the Compressor-encoded version. However, it may be that because the Sonic-encoded version is sharper, any slight aliasing may simply be enhanced. What’s weird is that in some places, it’s the reverse (more aliasing in the Compressor version than the Sonic one)! No clue why, but overall, the Sonic-encoded video is far superior.

[Glenn Camhi]

Well, interesting. When an AC3 is made right off the QuickTime file, the LtRt mix sounds great whether played as straight stereo or decoded ProLogic II.

However, when an AC3 is made from an AIF file exported from Final Cut, it sounds lousy when played as straight stereo. Something happens in the export process from Final Cut that messes up the file.

Could it be some setting? I’ve set the tracks as a stereo pair in Sequence Settings / Audio Outputs, and tried both Stereo Downmix and Channel Grouped, and in the export AIFF options I’ve set it for Stereo Mix.

[Glenn Camhi]

Thanks to you both for your replies.

Angelo: Yep, pro quality. I haven’t been able to find any reliable documentation on what 2-pass CBR does over 1-pass CBR, except that it analyzes what the optimum high bit rate should be and then sets it. If you have room for a CBR of 8.0 Mbps (which seems to be the highest that’s generally considered safe for most dvd players) then there’d be no point in 2 passes, right? But if 2-pass CBR does something else, is there any documentation anywhere that explains it?

I thought the point of multiple VBR passes is to analyze where your footage need higher bit rates and where you can save room with lower bit rates. If you don’t need to save any room, is there still some advantage? What else is it doing? BTW, I thought quantization errors occur in the other direction, when going from analog to digital, no?

Brad: Similar question — if 8.0 Mbps is your max as you say, and if there’s room on the DVD for the entire video to be at 8.0, why use VBR? That is, what advantage could there be in using lower bit rates in some places? Does something else happen when you use VBR that a constant 8.0 bit rate fails to do? I’ve never seen any documentation on that so I don’t know. Thanks.

[Glenn Camhi]

Thanks Angelo! Interesting. FWIW, I’m not getting the machine up and running, a major finishing place in L.A. uses it. I’ve assumed whatever they did would be superior to what I could do at home with Compressor.

For 30 minutes on a DVD, is there much advantage to using multipass over 1-pas CBR? Are you talking 2-pass CBR or VBR? If VBR, wouldn’t 5-10 passes be better? I know very little about this (clearly), but I thought if you’re using a very high bit rate like 8.0 on a DVD, 1-pass CBR is as good as you can get as far as bit rate’s concerned.

Anyway, if my options are Compressor at home or the Sonic, which can give me the better looking DVD?

[Glenn Camhi]

Thanks, right, it’s not reencoding.

After several tests at various bit rates, I’ve found that encoding in Compressor looks worse overall than encoding in Toast, in that it introduces new flaws. (At least for ProRes4444 going to H.264.) Among the flaws that Compressor introduced:

– blockier/aliased thin and diagonal hard lines
– some slight but odd color shifts
– a rainbow (moire?) pattern that appears in a thin black fence that does not appear when I encode in Compressor (see below)
– white text on a colored background gets some slight dark “shadows” on one side, not unlike when you over-sharpen or reduce the size of an image. (Same text looks pretty clean when encoded via Toast.)

Maybe there’s some other setting I’m getting wrong? FWIW, one color/post house said they also get better results sending QT files directly to Toast rather than encoding in Compressor first when making test Blu-rays. Since some of you experts here have the opposite experience, could it possibly depend on what the source material is?

Anyway, more tests to do….

[Glenn Camhi]

Thanks, will do, John. Running a few other tests too, will report back.

[Glenn Camhi]

Thanks, all. As I mentioned in my first post, I did try encoding in Compressor as well, to no avail.

But given what you’ve all said, perhaps I need to adjust the settings in Compressor. By which I mean, maybe Toast is overriding the “Never” reencode option, because it can’t handle the settings of the h.264 file that came out of Compressor? Is that plausible?

John: Thanks, that’s interesting. The source material is indeed PR4444, from an Alexa.