PDA

View Full Version : 2CD versus 1CD


Gazza
16th May 2003, 05:25
On the subject of opting for 1CD as opposed to 2CD. Is there any concensus that a 2CD (2 pass) encode with no filters and default or minimum xvid settings is equivalent (in terms of quality only) to a 1CD encode (2 pass) but with filters and tuned xvid settings (b-frames, qpel, VHQ,....)?

Teegedeck
16th May 2003, 07:59
I don't quite understand. You mean whether an 'unstable'-build-1CD-encode can look like a 'stable'-build-2CD-encode or better?

majerle
16th May 2003, 08:08
@Teegedeck

i think means that an unfilterd(undot/convolution...) without mpeg-tools (bframe/gmc/...) 2 cd encode is similar in quality as a one cd rip with heavy use of filters and tuned xvid settings (b-frames, qpel, VHQ,....)

Personally i usually aim at 2cd and so i prefer not using any flters (or at last a lazcos resirer and convolution hiq) but mpeg-tools is proof for me ("cum grano salis" as latin will say)

Regards

Andres

Gazza
16th May 2003, 08:17
Apologies for confusion.

I'm wondering if the quality of a 2CD encode (with the latest unstable build) and no b-frames, no qpel, no cm, no filters is equivalent to the quality of a 1CD encode (with the latest build) with b-frames, with qpel, with cm, with VHQ and filters, etc.

If there is an equivalent quality level, then you have a choice between fast speed and large encoded size (2CD) versus slow speed and small encoded size (1CD)? Does any of the developers know what the equivalent settings are?

After all, don't the new xvid features aid compressability in some way?

Andrey
16th May 2003, 08:48
As I understand :) a situation:
1 CD rip - lower resolution, most "additional" tools (like VHQ) must be used to achieve acceptable quality.
2 CD rip - if it has 2.35x1 aspect, then 704x304 can be used and compression ratio is only slightly stronger than DVD, so no need of special tools (like using H.263 quant and so on).
2 CD rip if aspect is lower (16:9 as example), lower resolution a bit and the rest is the same.
All above is for ~2h long film.
So there can be two goals, as I think:
1. Fit the film on one disk.
2. Get quality nearly as MPEG2 have. (It must be achived, because compression ratio is practically the same as on DVD for widescreen movies).
Speed of compression is not the main goal, especially when it will not differ more than 2 times :).

Correct me, if I'm wrong at something.

kaitsuburi
16th May 2003, 08:49
For me 2-CD means AC3 sound, where 1 cd is usually --alt-preset 128 Mp3 and sound is a pretty big factor in considering quality.

But I see the point of your question. I often find that with long hard-to-compress movies (2+ hours) I have to use advanced features and filters in order to fit them into the space left over after adding ac3 track and subtitles. With such movies (LOTR, Minority Report), I bet the 1-CD tweaked encode would look less detailed than the corresponding untweaked 2-CD encode.

-kaitsuburi

cult
16th May 2003, 17:11
I can get movie in 1cd+ac3 with vhq4+bframes+filters

lighty
16th May 2003, 19:36
Originally posted by cult
I can get movie in 1cd+ac3 with vhq4+bframes+filters
Sure you can but:
1. How long is the movie?
2. Is it 6ch AC3 or 2ch?
3. How low do you set your resolution?
4. Do you like macroblocks?

:devil: :devil: :devil:

iago
16th May 2003, 20:00
I can get movie in 1cd+ac3 with vhq4+bframes+filters"I can get any movie in 1cd+ogg with vhq4+bframes+filters" would be a more reasonable statement I guess! :D

Koepi
16th May 2003, 20:17
hehehe :)

kilg0r3
16th May 2003, 20:40
This is such a pointless discussion. I have sources which are about 2 hours long and fit on 1 cd with full res plus two soundtracks. Then, there are other sources which are 100 min long but which do not even fir on one cd when reducing the resolution. 'Atempause' was one and 'O Brother where art thoua was another.

'Atempause' has been worse, though; very bright movie, lots of detailed pictures with inherent motion/transformation, plus a hell of a lot noise. Although it is only 107min it had a first pass size of 2.5 GB _after denoising_ in full res. I had used mpeg quant. Nonetheless, iam currently trying to squeeze it onto one cd by reducing the vertikal resolution by a third plus using soft mpeg matrix. Yes, I will have to resize it during playback but I don't care. Zoom Player and ffdshow provide the funcionality.

Of course I am not using the ac3 track. :P

iago
16th May 2003, 22:03
Originally posted by kilg0r3
This is such a pointless discussion.Well, kilg0r3, of course it is, even the most pointless discussion of all times; still a never-ending one! ;)

Btw, for "1CD-ac3" encodes I strongly recommend using "DctFilter(1,1,1,1,.5,0,0,0) + cpu=6 mpeg2dec3 + TemporalCleaner + UnFilter + UnDot + BilinearResize(512*xxx) or (576*xxx) depending on the AR + vhq4 + b-frames 3/150/100/255 or even more depending on the compressibility test results! :D

And don't complain about the softness of your encode! :D

cult
17th May 2003, 15:33
lets see :movie election,resolution 672x272,audio 5.1ac3.I dont use anything except ac3.
But my point was that I am now able to get 1cd 4:3 movies(usally yes they have 2.0ac3).I couldnt do this before.Dont forget that 4:3 movies are hard to compress because of the ar

kilg0r3
17th May 2003, 16:39
@cult

it may sound a bit queer, but for 4:3 movies, i'd really suggest to reduce vertical resolution of the source and strech the film during playback. AFAIK, horizontal res is more important for the visual experience than the vertical res.

cult
18th May 2003, 15:58
@kilg0r3
Yes,I had thought of it some time ago and posted in gk forum,but their answers made me to drawback...Maybe it was not a bad idea after all?

precjednik
4th June 2003, 00:14
Originally posted by kilg0r3
@cult

it may sound a bit queer, but for 4:3 movies, i'd really suggest to reduce vertical resolution of the source and strech the film during playback. AFAIK, horizontal res is more important for the visual experience than the vertical res.

Interesting...

Do you recomend this on captured material also?
I made some test-encoding with horizontal reduction.
Captured (7xx-576) material have almost doubled redundancy in horizontal direction (if PAL signal is somewhere between 352 and 384, in theory, in practice even less). So, I was thinking that I could reduce just horizontal res to achieve better compression-per-resolution (??). Maybe wrong though.

Does anyone try experimenting with horizontal downsizing and what are the results?

Bad english haaa! :) Sory :(