A Quality Conundrum

[magic144]

so here it is

if I have a movie that I want to back up in movie-only mode and it's going to require only about 98.5% compression (by which I mean the target will be practically the same size as the original),

is there ANY value in using RB/CCE? I mean APART from time...

will the re-encoding make the video in anyway softer or blurrier than a transcoding?

will the use of matrices make a difference at that tiny compression requirement? - as far as I have read (and I've never used matrices other than the default), they are mostly for much more significant compression...

will some high-bitrate scene stand out as poor with transcoding (i.e. fast motion or some kind of water effect or complicated scene/explosion)

or can I sleep easy and just always do RB coz I've got the time and not worry about it looking any worse than transcoding...

don't want to open a whole qualitative can of worms here, short answers if you got'em :-)
is there a ratio above which you might always choose transcoding (e.g. >95% say?), or will it always depend on the nature of the source to some extent and therefore always be intangible without trying it both ways!

there, big question for short answers!

m

[spyhawk]

With that little compression, might as well use Shrink or Recode. You won't tell much of a difference with the original. I prefer Recode over Shrink in regards to speed.

But if you prefer RB, I would choose CCE 2-pass, High_High(4000+) matrix if the avg bitrate is > 4000. Or you can use the matrix extracted from the movie using RME.

It all comes down to how good your eyes are. Just don't be addictive in zooming to 400% to see if there's any pixelation. Enjoying the movie is more important.

[magic144]

thanks spyhawk

I must admit I've never used Matrices or RME - whether it be 65% compression or 90%! - I have a v. big screen RPTV and apart from a general 'softness' with higher compressions, never been unhappy with an RB/CCE encode - I HAVE seen TRANSCODES produce nasty (well annoying anyway) macroblocking on too-highly compressed a disc in certain scenes

so I've always been using the RB matrix "Encoder Default" - is that something built into CCE (if using CCE as encoder), or is that one of the standard RB ones (I have a feeling it will be the former - if so, does CCE ever choose a different default depending on bitrate, or is it fixed)

I tend to think of the prospect of playing around with matrices as an "if it ain't broke, don't fix it" kind of thing - along the lines of the chroma bug that supposedly exists in my DVD player - if I don't know what to look for, and I can't see it, why would I want to find out???!!!

I guess I'm more concerned about the likelihood of a transcode producing these annoying macroblocks and how I can begin to predict when that might happen (or if, like I said time is not an issue, I should just always run everything through RB overnight!)

cheers again

[manono]

Hi-

If it were I, I'd not use DVD-RB at all, but just demux the cell containing the end credits (or opening credits), reencode that, and then fit it back into the DVD using VobBlanker, leaving the movie itself untouched.

Alternatively and more easily, shrink a small part of the menu using MenuShrink. Depending on the menu, you may be able to get away with just shrinking a chapter menu or 2, or the extras menus.

Alternatively, since you removed all the extras, you may be able to blank out those menus entirely using VobBlanker and gain back the space you need.

The last thing I would do would be to reencode the whole movie. Just the act of reencoding the movie, no matter the matrix used or the amount of compression, will negatively affect the quality.

And even a transcoder at 98.5% will do a better job, as the original matrix (matrices) will be retained. I wouldn't worry about blocks at that high of a compression, unless they're already in the source.

[jdobbs]

I personally back up all movies using DVD-RB.

It is true that the lower the compression needed the harder it is to see the quality differences between an encoder and a within-the-compressed-domain transcoder... but they are still there. Why take chances and be disappointed when you finally watch the film, only to notice blocking in a few high demand scenes?

[spyhawk]

Ya'll are right! I think magic144 just wants to do movie only. I was just giving quick solution and alternatives. Looks like he has already chosen RB method.

To each his own so long as one is satisfied with the output. :)

[FredThompson]

I personally back up all movies using DVD-RB.You just a tad bit biased, don't you think? :P

[jdobbs]

Maybe... but the point is valid. If you need you backup in 30 minutes instead of two hours I guess there's a reason to transcode. I can't think of another one.

[FredThompson]

If the target and source size are about the same, it could be argued that requantizing is closer to the original quality because the I-frames are the same. If you're keeping, say, 95% of the B and P vectoring, what are you really losing?

[jdobbs]

Yes... but the argument wouldn't have much merit. You have to make up the loss of reduction in the I and P frames by more reduction in the B frames (at 95% I don't think the P frames would be modified). So 3/4 of the frames are bearing 100% of the reduction... the result is a stream in which the frame quality isn't consistent... and that shows itself more in full motion playback. No argument that the I frames and P frames will look great -- but the stream as a whole will suffer more than a fully encoded stream. That is one of the reasons a frame-by-frame comparison between an encoder and a transcoder isn't a valid quality measure.

[Fishman0919]

I did a test with this awhile back... 95%-85% reduction Movie-Only between DVD Shrink and DVD-RB w/CCE Basic showing the results to people at the Home Audio/Video store I worked at who had NO IDEA what I was doing. I asked them to tell me what movie look better to them... I displayed the 2 discs with the encoded movie on 2 of the same TV's set side-by-side (I believe they were Mits 55-inch WS HDTV's)with 2 of the same DVD Player showing the movie and if I remember correct it was something like 22 out of 25 liked DVD-RB w/CCE Basic better. I did this was 4 or 5 movies and the result were about the same with all of them.

[Harrysmiith]

[QUOTE=jdobbs]I personally back up all movies using DVD-RB.

Just to take this a little further - would you still back up a movie with DVD-RB which required no compression ? Being lazy I have tended not to do so but wonder if the use of undot and deen etc is reason enough to always use DVD-RB.

[magic144]

thanks for all the responses, comments and suggestions

I thought this would provoke quite a discussion

I wouldn't personally think it would be worth running a flick through RB when no compression was required - unless the production values on the original were really bad for some reason, or you have some particular dislike of grainy effects!

but then I've never used Undot or Deen so I can't attest to their effects - I figure I always want the output to resemble the input as closely as possible

m

fyi, I ran that 9x% movie-only backup thru RB and the output is perfectly fine to my eyes - I think I'll just do RB for all my backups and save any deliberating

[jdobbs]

Just to take this a little further - would you still back up a movie with DVD-RB which required no compression ? Being lazy I have tended not to do so but wonder if the use of undot and deen etc is reason enough to always use DVD-RB.In that case it really doesn't matter since you're not changing the source MPEG stream.

If you are going to use any filters you would, of course, have to reencode so "No Compression" doesn't apply.

[magic144]

@jdobbs

can you use RB to strip down a disc to movie-only
such that if the result requires no compression, RB will simply
strip the unwanted menus/extras and reassemble the target files
without the need to invoke an encoder (i.e. like doing a no-compression
re-author in Shrink, usually takes about 3-5 mins)???

or does RB always invoke the encode step?
is it smart enough to detect the need to reencode if you have chosen filters?

m