FCP6 and Quality Size questions

[TheJadeArcAngel]

Hello! Let me start by saying I read ( and used ) the AmvApp program via Cross-Over. This managed to handle a few issues I was having but only started a zillion others so I would like to start at the beginning:

1) I am needing to archive an entire series for the purposes of making AMV(s). I would like to rip/encode the whole series so if I need more material in the future I dont need to re rip-encode whole dvd's again, etc... With the H264 codec and a simple .mov container, a single episode comes out to a little less than 1 gig

*Note: this was achieved simple with MPEGstreamclip, selecting the 'Export to Quicktime' option and then selecting H.264 as the codec, 100% quality, ( no selection of multi-pass or anything of that sort ) and no size altering, thus, 854 x 480. The quality is 100 percent

With this method the show as an ... edge ... around anything black. An obvious ugly pixilation. the color space is also a little nasty however that can be changed, easily. Its mainly this horrible treatment of blacks that catches my eye and is so obviously not in the raw VOB files itself. I understand encoding handles the footage and it will come out a little less quality but .... eeeeeh

ANYWAY, this results in 1 gig files for 22 minutes of footage, more-or-less.

The AMVapp process ( with the huffyuv<spelling?> ) got me files well into the 5 gig range and that was for 1/2 the episode. Were looking at file sizes of 10+ gigs for 20 minutes of footage.

@.@ That seems ... not normal. Given even the demuxed files ( while not interlaced ) are about a gig, something seems wrong when these files get so big

Anyway, so I am looking on the site and I see a lot of PPC Mac users talking about Avidemux. I looked on the net and found Avidemux2_gtk?? I loaded it and saw a lot of settings but no (obvious) way to export anything and its crashed more than a few times while reading the .vob files.

back to AMVapp and VirtualDubMod ...

So, I start playing around with it and decided to test a 12-second clip just to play with the settings and what-not. I find things like this happens QUITE a lot when the file is finally exported ...



This is with the following settings:

VIDEO MODE: Full Procesing
COMPRESSION: Huffyuv v2.1.1

----> Huffyuv settings are: PREDICT MEDIAN and [<--- YUY2]

Also, even with de-interlace options ( of the MANY i find ) their is always some remains of material not de-interlacing properly



@.@ And, at the end, a single episode is 14 gigs ...

So, I guess, I have become embroiled in a program everybody seems to use and relies upon but I *KNOW* I am getting the settings wrong, or missing a step, or something

<_< problem is, nobody explains everything they do when they use VubMod so they must have steps that are not in guides or 'rules of thumb'

Help? Whats the means of getting these 'lossless' quality clips/files everybody uses but without a series being appx. 100+ gigs of material?

[TheJadeArcAngel]

EDIT: Okay, so ...

After looking around the forums, guides, and the rest of things other people have posted on the site I have gatherd the following:

1) Big-ass file sizes are just par for the course. I was thinking the 'normal' size wouldnt be over 10 gigs but, eh, thats life and thats what external HD's are for

2) One of the problems I was having w/ VDM was because I applied the deinterlace filter line on one run, it crashed, and then I re-did it without the filter line and was like 'ragh, it no worky, hulk-smash!'

3) I downloaded the wrong version of avidemux2 and now have the right one, the qt4 version. IT WORKS LIKE A DREAM .... with one exceptional error ....



This still pops up ....

@.@ any ideas how to get rid of it? It seems like such a random problem since the rest of the footage is awsome

[TheJadeArcAngel]

EDIT: ... more updates

Turns out ( of all things? ) that 'error' was hard-set into the physical VOB file? I re-ripped w/ DVD Decrypter and it loaded fine

[Qyot27]

I don't know how FCP handles editing in colorspaces other than RGB, nor if what I'm about to say would work at all (although in theory it should; Kionon or someone else that has ready access to the apps in question would be better to conform it), but:

If you have Perian installed, then through Crossover, I'd recommend using a combination of ffdshow and VirtualDub to encode to YV12-mode HuffYUV, preferably with the Adaptive tables option enabled and the prediction set on either Plane or Median. This will dramatically cut the size down, although they'll still be pretty huge. ffdshow's VFW interface allows for compressing in this format. Alternately (if ffdshow fails to work right), this can be done through mencoder's command-line*. You would need the Windows version of mencoder to do this, because it can accept AviSynth input.

*The mencoder command-line would look like this (I don't know if automatically putting it in MOV here is better than saving as AVI here and using SimpleMovieX to transfer to MOV):
mencoder testscript.avs -o testscript.mov -nosound -of lavf -lavfopts format=mov -ovc lavc -lavcopts vcodec=ffvhuff:context=1:pred=2

[TheJadeArcAngel]

Oooooo ( copies the names of the programs down )

Thanks!

After *seriously* thinking about it I'll likely be switching over my anime editing to Windows for a while, just try it all out. I love love LOVE the higher quality I am able to get w/ VDM and I really want to play around with it

My BIGGEST hurdle is the script-based clean-up. In the guides they mention a lot of different scripts and some look like they're ones now natively apart of AvsP or Avisynth, etc... plus the sheer amount of 'play time' required to test/play/refine these scripts PER SHOW is staggering

I can't get over how much time it takes to manage one clip, let alone ALL the material somebody might want for a big AMV project ...

STill, this is all so much fun

[Qyot27]

My BIGGEST hurdle is the script-based clean-up. In the guides they mention a lot of different scripts and some look like they're ones now natively apart of AvsP or Avisynth, etc... plus the sheer amount of 'play time' required to test/play/refine these scripts PER SHOW is staggering

It just has a comfortability curve. When you know what the individual functions do, and how strong to set them, once you've seen enough variation in sources you can really just guess from experience and it'll usually be okay. Most new titles (like, from the past 5-6 years) have very little wrong with them, and could probably just benefit from the basic IVTC operation, maybe the necessary aspect ratio correction if you prefer not to make the editing program compensate for it, and a little bit of light smoothing, enough to get rid of any mosquito noise or other barely-noticeable problems. It's the older titles that will make you want to rip your hair out (not to say there aren't just plain badly-authored releases from the last 5-6 years - *cough* R1 Mai-HiME *cough* - but those are increasingly the exception, not the rule).


Also, to be clear, when I said ffdshow, I mean the current builds that go under the name of ffdshow-tryouts. I generally go with the most current generic SVN build by clsid. The beta versions I'd stay away from, not because they're buggy*, but because they get outdated fast.
http://sourceforge.net/project/showfile ... _id=173941


* Well, except that one from December '08, which will spazz out on you if you try to use a CLI program and ffdshow is called somewhere in the process (for instance, the program takes AviSynth scripts, the video you're trying to open is normally decoded by ffdshow, and you're using either AVISource or DirectShowSource in the script which will make ffdshow do the decoding) - that issue was fixed in January, but there's not been another beta release that I know of. To be precise, it made it to where you had to specify the exact path to the executable if the command-line options exceeded 256 characters or something like that; it'd be immediately noticeable if you do really insane-level encodes with x264 or xvid_encraw or mencoder, etc.