workflow from Betacam SP PAL to DVD

Hi,

I would like to share my first-time experience with getting from a Betacam SP PAL tape to a DVD with anyone who would read it. I'm not sure I followed the most optimal course througout, and am concerned that the media may have suffered because of it. Hopefully, someone will be kind enough to share their wisdom.. I'll try to pitch in to the brain trust, by providing some details that could be useful to other beginners, like myself.

The project involved synching the video material (of a live concert) from a newly-surfaced Betacam SP tape to a pre-existing audio mix. The Betacam contained a composite edit- clearly live- of several camera feeds. The audio was a composite of the best live performances from several different nights. Few moments offered the hope of simply 'lining up.' (Eek!) Of the 90minute program, we only wanted 10minutes-worth of highlights.

1. Betacam SP PAL --> DVCAM NTSC (via Alchemist Pro)

For converting between NTSC and PAL- performing a standards conversion- the Alchemist Platinum is currently the best device. I was advised to digitise to DVCAM bc it theoretically offers the best chance for a longer lifespan, and can store more information. Albeit the differences (between DVCAM, DV, miniDV) are small.

I wonder, if my final product was going to be a DVD, did I really need to perform a standards conversion? FCP is perfectly happy in PAL format, and the DVD could be encoded to play everywhere in the universe.

One oversight on my part at this point related to timecode. The Betacam had several TC breaks, and the engineer (who I didn't know, and the transfer house was also unfamiliar) tried to do the right thing by perserving the breaks in the transfer. Matching the TC, breaks and all, would have been proper if our audio mix was also synched to it. But the audio was not. I would have preferred a straight TC flow, but I worked around it, bc there was plenty of material to work with. It would have been prudent to be present at the conversion, esp since I didn't know the transfer house well.

2. DVCAM --> Hard drive (via FCP 4.5)

I captured via FW, playing out from a deck into FCP, capturing using DV/DVCPro 720x480, 29.97, NTSC (CCIR 601), with field dominance set to none. Then I edited and colour-corrected away (and there was much to do, bc the recording was clearly circumstantial, so skin tones cruised the colour wheel as the lighting changed from song to song.) I put in titles using Title 3d (Gil Sans at 22pt.) All rendering was done using YUV, not RGB.

We tried to capture at 8bit uncompressed, but could not. The capture settings we used seemed accepable. This article at ken stone's gave me heart:
[www.kenstone.net]

3. Quicktime movie --> MPEG

I exported the Quicktime movie using the same settings as the clips.
Then I let DVDSP do the MPEG conversion. I noticed some differences between the QT mov, and the DVD that I can't fully comprehend:

a. In general, the material on the DVD (playing through a DVD player) looks fine on a broadcast monitor.
b. The DVD looks worse on my computer screen, playing through DVDplayer: I notice some artifacts in a few whites (although my levels are below Broadcast danger), and I notice more pixelation, again on whites, than in the Quicktime movie playing in Quicktime player. There are also some interlacing blurs.
c. The Quicktime Movie playing through the Quicktime player (ie, the pre DVDSP converted media) doesn't have the problems mentioned in b., but in exchange the text looks awful.

Is DVDSP making a funky conversion that I can workaround? Or did I set myself up for a problem earlier in the game, through my capture settings or my export settings? Maybe too many filters- is this possible, even though I can't detect the same problems in export? The footage was not great to start with, and the colour correction brought everything to life, but I'm losing sleep thinking that I might have done something to further distort it.

At this point, am looking at what I can do to cover-up some of these problems- a slight Gaussian blur (.1 or .2) maybe De-Interlace the video. Should I try different export settings? Acheiving a balance in quality between between TV screen and computer monitor is extremely important, bc I expect this DVD to be played on both.

My tools:
1 GHz Powerbook (FW800) 2GB RAM, 60GB HD OS10.3.8
FCP 4.5HD (part of Apple production suite)
160 GB Oxford IDE drive (400MHz)

Is this the appropriate list to post to, or should I have posted to the DVDSP?

Thanks very much in already: the knowledge and helpfulness already available at LAFCPUG is amazing!
dan

No need to convert your Pal to NTSC if going to DVD. DVD players dont care from Pal or NTSC, at least the new ones don't.
I wouldn't trust what your DVD looks like on your Computer. Look at it on a broadcast monitor or TV. As far as end product, you can play with the Variable Bit Rate of DVDSP and see if that helps.



Michael Horton
-------------------

hey, Dan, sounds like an interesting show to put together

going to DV might not be a great link in the chain here.

if you want to work with higher res, or rather less compressed, footage, you might be able to get the dub-house to capture to FCP for you.

one it goes down to DV, it gets compressed, and that's that
you cant capture it over FW as uncompressed, anyway,

not that DV is terrible, of course.

Dan:
"The Quicktime Movie playing through the Quicktime player (ie, the pre DVDSP converted media) doesn't have the problems mentioned in b., but in exchange the text looks awful."

i cant comment much on the DVD side, but there's a trick with QT files, (or is it just DV QT Files?)
anyway, you open the files properties in QT (Apple J)
first menu - video
second menu - quality
set it to high

big improvement in the way text looks, esp

but that's ONLY about presentation on the desktop.
that wont help to improve your DVDs,

cheers,
nick

thank you very much Mike, and Nick!

i understand the dangers- inaccuracies- that come with using computer monitors to judge true colours. i used a broadcast monitor, and different TVs (just to see how the colours range.) but we expect many (50%) of our viewers to play this DVD on their computers as well (full screen, no doubt.)

at this stage (making overall adjustments) i feel that i must account for how the video might look on a computer, and find a compromise that works both ways.

based on some more research, and trial-and-error here's what i'm currently trying out:
-copy the clip onto the video track above
-top clip: Gaussian blur (0.1), Opacity (30)
-bottom clip: Gaussian blur (.35), some sublte Desaturate Lows, De-Interlace (Lower), and Broadcast Safe

the basic idea of the above (i'm stll tweaking settings) has solved many of my problems.

hmm... if i had the dub-house re-capture from the DVCAM (uncompressed), could my project use the uncompressed material rather than the DV material? this is new ground for me- what part of the manual discusses this? i suspect those TC breaks will cause me trouble, but it would be nice to be able to still use my work.

what if i were to digitise the Betacam SP tape, uncompressed, into FCP, preserving the PAL format? i think i'd have to start from scratch then.

thanks again (and again),
dan

A lot of computer DVD players will de-interlace on-the-fly for you, so that's another option to consider. There's literally no advantage to capturing DV as uncompressed. You might want to consider rendering the final output before making your DVD uncompressed as that will eliminate a DV compression, but again, the benefits are negligible as DVD is much lower quality than DV.

Taking in BetaSP direct and making a PAL DVD would indeed look best, but an alchemist does a very good quality conversion, so I doubt that's much of an issue.

Graeme



[www.nattress.com] - Plugins for FCP-X

thanks Graeme,

i'm not entirely sure what this means: "consider rendering the final output before making your DVD uncompressed" but I did render in YUV mode, and usually exported the movie using the same (ie DV/DVCPro) codec i captured with.

your msg got me thinking about finding other ways to make the MPEG2 (i deefaulted to letting DVDSP do the work):

I exported the sequence using Compressor, with the MPEG2 60 Minute High Quality Encode preset. while the process took 15 hours to complete (which seems outrageous for 9m10s movie) i acheived the most satisfying results yet. the final DVD img was 504MB (compared to the previous 6 versions, which weighed in at about 330MB.)

to compare, i double checked my prefs in DVDSP, and noted that the Encode tab settings should be reset to 2pass VBR, bitrate 7MB/s, max 7.5MB/s etc. then I exported the sequence from FCP and let DVDSP encode it. the process lasted about 4 hours, the final clip was 525MB, but the quality was nowhere near Compressor's (tiny artifacts in some whites, heavy pixelation.)

Well, if you've edited DV, for instance, renders to DV do some compression and will reduce image quality slightly. If things are cuts only, then there is no image quality loss, but if you do any effects, then before final render, setting the timeline to uncompressed will make things look better. Then export that uncompressed movie (I'd never recommend exporting direct to compressor as it's slow and unreliable) and take it to compressor.

Graeme



[www.nattress.com] - Plugins for FCP-X

hmm... it makes sense that setting the timeline to uncompressed would create a better final render. but if this is the case, it would follow to reason that digitising at an uncompressed setting could squeeze out more quality, even if the original media doesn't have that much information in it (such as DV.)

theory aside, i tried exporting a short clip set to uncomprossed and then making an mpeg2 from it with Compressor. both the quicktime movie and the mpeg2 looked very pixelated, so, not better than previous results.

i couldn't understand why this 'bump' up to uncompressed would look so much worse (i would have expected it to look the same in the worst scenario), but wondered if it had to do something with not being able to fill up all the space of an uncompressed movie frame with the data of a frame at DV compression. as if i was witnessing the data getting spread out too thin?

in any case, thanks for all the wisdom guys!
dan

dan wrote:

>
> hmm... it makes sense that setting the timeline to uncompressed
> would create a better final render. but if this is the case, it
> would follow to reason that digitising at an uncompressed
> setting could squeeze out more quality, even if the original
> media doesn't have that much information in it (such as DV.)

Does make sence but its not the reality although some will argue that. Adamwilt.com has an explanation of this over at his web site. You'll have to search for it.



Michael Horton
-------------------