DVD Quality problem
 

If anyone has already posted this question - please point me to the right direction.

I seem to be having quality issues in my final product - i.e. DVD-5 . I an editing with FCP 5.1.4 , from captured 1080i/60 footage. I've tried exporting as a QT movie at current settings ( with noticeable artifacts ) and with the AIC codec ( with a bit fewer artifacts ) .

Can someone PLEASE assist me with recommended work-flow to get the best quality product to DVD possible ?? Step by step would be appreciated.

I only have 25 minutes of footage and need an auto-play DVD. I've burned the last two times with iDVD. Should I use DVD SP to burn the projects ?

I have to deliver this project by week's end, Anyone know what I'm missing ?


Thanks !!
Darin

Home Run !!
 

I hate to answer my own question- but maybe it will help someone else as well.

I just burned the best quality SD DVD that I have ever burned ! Just what I needed - an auto-play , looping DVD from HDV footage.

Here's the links to the step-by-step instructions:

HDV video to SD DVD using Compressor and DVD SP

DVD Studio Pro: Authoring a DVD that plays automatically without a menu

[ Be sure to use the "Bit Budget" calculator - Awesome ! ]

That's all it took !! Thanks Ken Stone & Apple !

I don't think there's any way of getting high quality results using mac workflow for the HD down-scaling to SD - one has to dip into a bunch of esoteric utils - but the quality is well worth it.

HD to SD: Precomposed Blog

That's completely incorrect and in fact the built-in tool, Compressor - which that article does only but to slam it - is quite capable of making very high-quality encodes and downscaled HD-to-SD finals. There are two main problems with Compressor that cause this wide-scale misconception:

1. Most people do nothing but use the built-in templates for making their encodes, whether it's downscaling or not. In fact, the title of some of the built-in settings say things like, "best quality...", but in fact it's very, very far from being the best quality Compressor is capable of because the deeper settings for both scaling, retiming and motion estimation are not turned on and or not optimized. And the reason Apple did this leads into the second reason;

2. Compressor is painfully slow when you activate all it's advanced settings to their maximum quality settings, and the same people who simply use the built-in template settings also lack the patience to allow Compressor to complete it's task. It's one reason why so many third-party apps have sprung up in Mac-land for final encodes; not because Compressor *can't* do a quality job, but that it's very inefficient at doing it taking far too long for the average "I want it yesterday" user.

As posted in my review of the recent Apple-based encoders:

encoder shootout

Episode is the king of the hill for both speed and quality. In fact, no software-based encoder, not even the PC versions are capable of making a higher-quality final encode than Episode. Only a hardware encoder can do a better job - if you can afford the price-tag. Episode's main failing? The cost: just $4 dollars shy of the full FCS suite.

So can Compressor do the job as well as the PC encoders? Absolutely if not better, you just have to be patient and let it do it's job.

For a quick and full understanding of all Compressor is really capable of I always refer people to this book: Compressor 3 Quick Reference Guide; Brian Gary.

Robert,

What is the best HD to SD work flow when using compressor? I've read in other thread that you need to use a bicubic spline downconverter or Lanczos rescale. I don't fully understand what those are but it seems like that is what most people are using to get the best quality. Are these available in compressor? or something similar? At this point using compressor is the most cost effective solution, I'll just have to be patient.

I am shooting a wedding the first week of March and want to shoot it in HD so that I can have the flexibility to put HD footage online and use the footage for demos but the clients need the final product on DVD. Any help with this work flow would be great! Thanks.

Chris

I'm in the same kind of boat. I will also be shooting a wedding in March and would like to know the best workflow. Everywhere I look, people are using Cineform NeoHd or programs like Episode 5. I find it really hard to believe that FCP Suite can't do the things I want and make it look good. I must be missing something HUGE!

I'm so glad Robert said his piece because I read the linked article too and thought that's complete rubbish - I get good results using Compressor but didn't really have confidence to trash the article.

Joel and J it isn't that tricky - look here for a thread on the issue:

http://www.dvinfo.net/forum/dvd-auth...ny-advice.html

Robert, that's good news and would simplify my foray into WindowsXP everytime I need HQ HD -> SD conversion.

Your 'shoot-out' seems mostly concerned with speed over quality and your test file does not address issues of stair-step or jaggies common in converting interlaced HD footage to SD. A problematic area is titles - you make no mention of quality of CG titles. FWIW I should also declare I work in PAL.

Conspicuously absent from your shoot-out and post here are specific settings that will 'unlock' this hidden quality of compressor. It may well be capable, and I will invest some time testing it (again) with careful attention to frame controls and "scaling, timing motion".

What I can say is the PC-based AVISynth rescaling of HDV produces far superior DVDs to any of the other Mac-only workflows where the authors' seem completely satisfied with their results. I.e. each of us has different standards and for instance I am completely dissatisfied with the results following one such workflow:

Exporting HDV Video from the Timeline to Standard Definition DVD

Well, the article isn't rubbish - 'cos it works - plain & simple. Whether comparable results can be obtain from a simpler Compressor workflow is the subject of debate.

You mention you get good results (not great?). As mentioned my post above people have different standards.

Maybe your subject matter is not demanding of the HD to SD process. Do you use a lot subtitles on-screen? Do you have high contrast diagonal lines like a dark men's jacket lapel against white business shirt?

I would respectfully suggest you try a test with above sort of content 1) in compressor using the method you're happy with 2) following Jon Geddes workflow.

Appraise and then share your thoughts. If you haven't tried the PC workflow your view that it's complete rubbish is of little merit.

Mark, I should have qualified my argument - I meant that the contention that the results from Compressor were bad was rubbish, not that the workflow suggested would not produce very good results. Indeed I haven't tried it because the complexities of it for a Mac user are prohibitive unless you've got a lot of time. This does matter.

And you are right, true appraisal cannot be done without proper comparison. All I can say is that I haven't felt the need to do it because I am happy with the Compressor workflow (though not the rendering times as Robert points out). Maybe my standards are lower but I feel we can get so bogged down in technicalities and an obsession with visual 'quality' that we forget the content.

Geoffrey,

Fair enough. I admit to obsessing about the encodes as I'm responsible for producing internal video comns for large govt department featuring the agency head. I had a requirment to use VMWare Fusion to do our WMV encodes with Expression encoder...so I took a morning to try the PC utils.

Compressor with frame controls ON not productive - my faffing about in WindowsXP is quicker based on the 1min HDV test I'm trying today - just incredible 2x 1 min encodes are taking something like 2 hrs!

Indeed, the shootout is specifically about speed, because those who know how to tweak and setup Compressor properly never complain about it's quality of output but it's snails-pace at doing the job.

Suffice it to say that Compressor has been used a great deal in the commercial world even in film workflows to and from the Telecine process with great success. Very slowly, but successfully. And the "best settings" question has been posted - and answered - to *ad nauseam* on this forum, so for those asking that question I always respond with two options:

1. Use the "search" feature on the forum to lookup those posts so that it's not being re-posted again for the um-teenth time and/or;

2. Get the book by Brian Gary that I recommend to anyone serious about learning all the tricks Compressor has to offer. You certainly can't learn all you need to know hoping a Q and A session on a forum will cover *everything*. (^_*)

At some point when I get caught up on everything else I'll post a "best settings" for various Compressor tasks on the review site and hopefully it will get posted as a sticky here, but until then use the free tools available to you and get the info freely floating around here.

Thanks - Robert - indeed I enabled the frame controls and can see the there is no problem rescaling. I added comment on Jon Geddes blog to this effect as I was incorrect about Compressor's ability.

I'm still sorting out some flicker or jittering prob related to my field dominance (I was sure lower/odd was it for SD PAL DVD).

Unfortunately the speed penalty will probably count it out for all but the odd job...our 2.26 8-core MacPro isn't enough.


UPDATE:
As I eat my words on Compressor’s ability I thought I’d share my comparison of PAL HDV to SD scaling/MPEG2 compression of Compressor 3.5 vs a workflow using a Windows utils as outlined at the PreComposed blog referenced above.

Once I tweaked Compressor the results were nearly the same as using a freeware PC encoder (HC Encoder v0.23) and resizing with Avisynth.

Both encoders were set to two pass, 6.2 Mbps VBR Best quality, ceiling of 7.7 Mbps, with Compressor Motion estimation = best.

I preferred the HC Encoder video – it showed slightly better details in terms of small facial imperfections -not a good thing aesthetically – but this a technical appraisal.

NB: I left Compressor frame-controls anti-alias and detail on 0, since Apple specify these parameters were applicable to improving up-scaling.

Other frame-control I settled on :

Resize filter : better (linear filter)
Output field: Top first
Deinterlace : Best (motion compensated)
Adaptive details checked on
Duration 100%
Rate conversion (Fast nearest frame)

In summary, Compressor does a sterling job of processing HD source material into SD MPEG2 but alas the time taken precludes it from being used in our workflow.

My test video was 1501 frames long (~1m), sending 5 iterations to Compressor with various frame-controls settings took 5 hours 41 mins, roughly 1 hour per minute of source to encode.

1min of the same footage took under 7 mins to encode using HC encoder/Avisynth running on VMWare Fusion 3.

So are you saying that it would take compressor approximately 60 hours to encode a 1 hour sequence?

yes, unless extrapolating from my 1 min video doesn't represent Compressor's performance on longer videos. But from Robert's comments of being "painfully slow" above I guess 60 hours would be it based on my 8 core 2.26 machine at least and not fooling with Apple QMaster.

Yes it does take that long - I've done it. Results are very good though. It is the de-interlacing that really ups the encoding time so if this isn't needed then the time is only a fraction of the rate described.

When would de-interlacing not be needed?

I have found that some projects just don't need it due to the nature of the material but more importantly I've noticed that many modern DVD players do a very good job of real time de-interlacing - better than any processing of the file as it doesn't degrade the image in any noticeable way (the Toshiba I've got for instance). My knowledge is not good enough to know how they do it! Again though it does depend on the project and it only takes one awkward passage (normally of fast moving complex motion in my case) to make processing necessary.

Really what I am trying to figure out is if it will be an efficient workflow to try and produce wedding DVDs through compressor from HD footage. Typically the ceremony is anywhere from 20-45 min and the reception events sometimes 20- minutes along with the wedding day trailer. Lets say I have a total of 50 minutes of footage to compress, according to you guys to get quality DVD picture it would take 50 hours. Not to mention that I'm working on a 3.0 ghz duocore iMac. I only have a mac so the PC workflow really isn't a option and 50 hours or longer to compress a single project is insane. Any suggestions?

I've no idea where you're extracting those kind of runtimes from but...

On your machine using the best possible settings Compressor has to offer for MPEG-2 encoding should take between 4-9 hours depending on the actual speed of your system and exactly how deep you go with quality options. Not 50 hours.

As I always suggest to those trying to wrangle and fully understand how to use all of Compressor's settings - and when *not* to use them - pick up a copy of this book which is worth it's weight in Platinum!:

Amazon.com: Apple Pro Training Series: Compressor 3 Quick-Reference Guide (9780321514226): Brian Gary: Books

Thanks Robert, just ordered the book from Amazon. Hopefully this will help clear up some things for me. Thanks for the help every one!

Ha! I just ordered the book yesterday from Amazon. It looks like great minds do think alike. Keep us posted on the settings you use and I will do the same.

Thanks

-Joel

Will do! Thanks!

By the way, I hope that I fall into the "great minds" category. LOL

That figure came from mark's post that, for him it took compressor about one hour per minute to encode his project. Again, I'm a noobie here and maybe his setting and situation doesn't apply to what I'm trying to accomplish but I was simply applying his experience to my projected workflow. Most wedding projects will have somewhere around an hour of footage to encode therefore according to Mark it would take 60 hours at the one minute per hour encode rate.

However I think I'm going to wait and give the book you mentioned a once over and to some practice runs to gauge for myself the time involved to compress a typical wedding video.

You certainly don't have to wait for the book, try doing some of your own test encodes. Take a 10-second clip from any segment and put it into it's own timeline. Fully render it, then send it to Compressor via whichever method you prefer and see how long it takes on your system.

If you run into trouble just let us know!

If you turn on frame controls, rendering times get much longer. But the quality and sharpness also gets better (coming from HD).

I use a preset with a 2-way pass, at 8.54mbps (with a minimum of 6.2 if I recall correctly), with AIFF audio, motion controls on, details set to 20 and motion estimation set to best. For me, this produces the sharpest, crispiest looking .m2v files.

I did a test with both Compressor, Squeeze and Adobe Media Encoder, and the setting above is my 1st choice. AME and Squeeze fail to come close to it. You can add a lot of filters during the DVD encode in Squeeze, maybe that can have some impact.

With that having said, DVD encoding results on a Mac are very poor. On a PC, you can get better results with mostly free software. Adobe and Apple should do better than that.

Read my earlier post in this thread why that statement is false and very misleading.