CCE 1pass VBR test results

[telemike]

Here are the results of using DVD2SVCd to encode a 2:24 clip from a dvd source using CCE 2.50 and changing only the Q factor. 480 x 480 svcd

Standard Bileinear resize. audio is 160k dual channel.

Q SIZE MINUTES FOR 800MB
5 39506 50.63
10 38392 52.09
15 36220 55.22
20 33574 59.57

[UltimateDBZ]

So? :confused:

[telemike]

Well, you can see the amount of compressabilty with each increase of "Q" with a 1-pass VBR. Much like TMPGenc's CQ mode.

[UltimateDBZ]

But what's the point of this topic? Just informational?

[telemike]

Informational.

As you can see by changing the "Q" factor from 5 to 20, I estimated that 10 minutes more per CD can be added

[DDogg]

As a slightly OT aside, I noticed D2S had in his to do list a size prediction for 1pass vbr. I would guess that would be a lot like the process used for TMPG CQ that is in the program now. I don't know if it will ever happen since he seems to be moving away from much further development, but it would be a nice addition to have.

Maybe it might happen sooner if somebody/us could work on some type of calculation and prediction algorithm that he could just plugin as the multipass prediction code is already in the program presently for TMPG CQ mode.

[UltimateDBZ]

Originally posted by telemike
Informational.Gotcha. Just making sure there wasn't a question/problem that I missed, is all.

[telemike]

Now, for someone to make that into a plugin or program! There is a program for TMPGenc called ToK over at www.kvcd.net. However, no program for CCE.........

[DDogg]

I expect that tylo might give it a shot if he has the time and sees the worth of doing it. I was always expecting that RB would come up with something like this as it seems down his alley.

@Bach, thanks M, I had never tackled that thread as I thought it involved dicking around with all that hand bit allocation. I should have known that would have not been acceptable to you. Best regards.

Later: I tried the above and it worked fine. Did a 1:50 hr movie in less than 3 hours and the quality is good. I found it interesting to see the quick result of using light filters to reduce size instead of just using Q. One test run with no filters using Q16 came in at 16,950. I added Deen() and undot() and it dropped to 13,250. This allowed me to then reduce Q to 14 and still make my target for a two disk encode.

@DVD2SVCD, well there is the prediction code and you already have these routines built in for TMPG. How about re-waking from the dead and add this for CCE 1-pass VBR?

[tylo]

Did I miss something here? D2Sroba does exactly what Bach describes here, and it is fully automated. Wow - that documentation must suck bigtime. Anyone like to create a new webpage?

@DDogg: When you reduced Q to 14, you actually assumed that the quatization level curve was linear, which is almost correct. Be aware though, that it's not as linear when you come closer to the min/max bitrate settings.

@DDogg (2): Did you really try to read the D2Sroba doc at all? You just admitted to have skipped one off the most interresting theads in this forum...;)

(Edit): For your information, I will add an input box, giving the choice of using 1, 1-2, 2, 2-3, 3, 3-4, or 4 CDs. When selecting e.g. 2-3, it will estimate Q for both 2 and 3 CDs, giving you the choice of selecting number of CDs based on estimated Q and bitrates for these two.

[DDogg]

You would have to understand the history of 1 pass and myself. I got very involved with 1 pass vbr encodes a long time ago...long time as in my old 800 Athlon slot A classic was state of the art, we were still on EZboard, and our resident composer was doing the fifth symphony (never mind) :)

Anyway I pestered D2S to add CCE 1 pass VBR mode and as you can imagine, the resulting tech support from newbies trying to use it was a freaking nightmare. Then I pestered him into adding the TMPG 1 pass CQ mode and more support nightmares came along. Those two 1 pass modes caused so much aggravation to D2S and myself that I kinda developed an aversion to the whole topic, hence that thread held no appeal for me whatsoever.

Back to your plugin ... yeah, I actually did read them, but as said, I must have not understood them completely :p Actually I think I only did one encode with your plugin. I got a 2 disk encode on a longer movie and did not like the quality so I did not pursue it as I should have to find out how to force a three disk encode (I am assuming you could do so).

I do hope you add the front section you mentioned to your work, although I still would like to see native support for Bach's concepts incorporated directly into D2S. Anyway, don't hesitate to PM me if I can test or help in any way. Maybe I can make suggestions about the user interface that might make it slightly more intuitive for users who do not read long convoluted threads and docs :D

[tylo]

Thanks, DDogg. I know you are one of the greatest contributors to keep these forums interesting.

You wrote: I got a 2 disk encode on a longer movie and did not like the quality so I did not pursue it as I should have to find out how to force a three disk encode

You should rip a chapter or two and encode it with a set of Q values. The one with the lowest quality that you find acceptable, is the one u put in D2Sroba. That would create 3 discs in this case. In theory, with this Q, you should always get that quality for any movie.

Ok, I'll start updating my plugin as soon as I find some time.

[DDogg]

Great news and if I can help with the docs let me know. I wonder if that frontend prediction piece would not be real handy for many folks as a standalone?

[tylo]

New release: D2Sroba v2.1.3
http://forum.doom9.org/showthread.php?s=&postid=326241#post326105

Feedback on GUI, docs, and bugs, please.

[Holomatrix]

Hi, I was just trying the new version, started it up ok but when dvd2svcd was in the middle of extracting the audio, D2Sroba disapeared. The status window is gone plus the taskbar icon is gone. Is that supposed to happen? dvd2svcd is still doing it's thing though (converting the audio)

[tylo]

You didn't by any chance minimize the DVD2SVCD window? This should only happen if D2Sroba does not see the DVD2SVCD app anymore (i.e. minimized or closed). D2Sroba does not do anything before it sees the CCE encoding window, so you can actually restart D2Sroba while DVD2SVCD is working. Please, try again and see if it happens again.

[Holomatrix]

Woops :) I do minimize it because I use my work PC right now to do other things, so I should not do anything with my computer while using D2Sroba ?

[tylo]

Well...,
Before CCE and DVD2SVCD is terminated by D2Sroba, do not minimize DVD2SVCD. After that, you may minimize CCE 2.66+ when using EclCCE, but don't touch the keyb+mouse during input simulation by DVD2SVCD and D2Sroba. While encoding the main movie, you may use your PC.

[homerjay]

Originally posted by tylo
New release: D2Sroba v2.1.3
http://forum.doom9.org/showthread.php?s=&postid=326241#post326105

Feedback on GUI, docs, and bugs, please.

hi there i am interested in trying out this following discussion with a friend but i dont appear to have access to the thread
what i am really wanting to do is experiment with creating 1cd conversions up to about 2 hours

i have managed to produce excellent results with a final Q of between 63&65 using various plugins but despite selecting 1 cd it is going onto a second to the tune of between 150 and 250 mb depending on filters used

if you can give me a link to an active thread or advise me here i would appreciate it
i have downloaded the latest version 2.44 from your webpage and am using cce2.50 and avisynth 2.5

thanks

hj