Low Bitrate MAF Encodes using CCE and Tylo's 1 pass Plug-in

[telemike]

Yes, I give IMMENSE credit to KWAG and the crew from kVCD.net as well as the others here :)

Thanks to y'all here and over yonder at www.kvcd.net, I am making really good SVCD's with 1-pass CCE!
:D

[Holomatrix]

Originally posted by jorel
@ DDogg
:)

Little fix in the script:
as SwitchThreshold can be egual to 4 and the limits for
temporalcleaner parameters are (10,15), the correct line is :

TemporalCleaner(6+nf,11+nf)

!

Do you mean change the end 2 to a 4?
8=SwitchThreshold = (Width<=352) ? 4 : (Width<=480) ? 3 : 4

[jorel]

Originally posted by Holomatrix
Do you mean change the end 2 to a 4?
8=SwitchThreshold = (Width<=352) ? 4 : (Width<=480) ? 3 : 4

no ,i wrote "can be" egual to 4 or less:
SwitchThreshold = (Width<=352) ? 4 : (Width<=480) ? 3 : 2
means
if width is <=352 will chosed 4,if width is <=480 will choosed 3
the switchthreshold "auto choose" the values.
then use
TemporalCleaner(6+nf,11+nf),remember the "limits" of the filter.

ahhh.....sorry my poor english!
:o

[homerjay]

Originally posted by Holomatrix
@homerjay - What would happen if you didn't add that line at the begining?


colours were messed up:confused:

[Holomatrix]

Oh, that's right, I forgot that those Divx codex's installed screwed things up for me. I did an encode and the people were blue :) I just added SwapUV() and all was fine. I don't do AVI anymore so haven't had that problem.

[DDogg]

Jorel, so I THINK :) you are just saying, Change TemporalCleaner(6+nf,13+nf) to TemporalCleaner(6+nf,11+nf). Is that all? If not, please post the corrected line as a CODE segment - bracket CODE bracket then end with bracket /CODE bracket.

The reason I even brought up the second script was because it seemed to give better compression than the semi-auto-MAF and thought others might like to test that. I personally will probably settle in on the Semi-auto MAF because I do not really want to deal with setting anamorphic true or false and all that BS.

However, that being said, compression is the name of the game when doing 90-120 minutes on one piss-ant 80 min cdr. Same goes for IDCT factor - If IDCT 7 is slower but gives better compression, then it is what I would use. My new machine is fast enough not to notice much anyway.

Speed and compression are nearly mutually exclusive. BUT, that being said, if Tylo's plugin AND these MAF scripts can give us an accurate and fast 1pass vbr 1 cd encode using CCE on a 90-120 min source...well then you kinda get the best of both worlds.

Actually the real reason I brought this stuff up was because I was bored and wanted to hear myself expound :p

[jorel]

Originally posted by DDogg
Jorel, so I THINK :) you are just saying, Change TemporalCleaner(6+nf,13+nf) to TemporalCleaner(6+nf,11+nf). Is that all? If not, please post the corrected line as a CODE segment - bracket CODE bracket then end with bracket /CODE bracket.

The reason I even brought up the second script was because it seemed to give better compression than the semi-auto-MAF and thought others might like to test that. I personally will probably settle in on the Semi-auto MAF because I do not really want to deal with setting anamorphic true or false and all that BS.

However, that being said, compression is the name of the game when doing 90-120 minutes on one piss-ant 80 min cdr. Same goes for IDCT factor - If IDCT 7 is slower but gives better compression, then it is what I would use. My new machine is fast enough not to notice much anyway.

Speed and compression are nearly mutually exclusive. BUT, that being said, if Tylo's plugin AND these MAF scripts can give us an accurate and fast 1pass vbr 1 cd encode using CCE on a 90-120 min source...well then you kinda get the best of both worlds.

Actually the real reason I brought this stuff up was because I was bored and wanted to hear myself expound :p

i had to quote your post for evident reasons.
see that i wrote that my english is horrible in my last post.

see the "full" line changed:
9=ScriptClip("nf = round(YDifferenceToNext())"+chr(13)+ "nf >= SwitchThreshold ? unfilter(-(fmin(nf*2,100)),-(fmin(nf*2,100))) : TemporalCleaner(6+nf,11+nf) ")
from your post: "Last edited by DDogg on 7th August 2003 at 19:39"
like i wrote before!

only this my friend,cos it give clean results and short final size!
:)
best regards.

[DDogg]

Thanks, Jorel. The above is edited as you suggested.

[telemike]

Couldn't get motion filtering to work with CCE 2.50 :(

I did get this script to work:


[AVISYNTH_STM_BilinearResize]
0=BilinearResize(^TargetWidth,^TargetHeight)
1=Undot()
2=Asharp(1, 4)
3=STMedianFilter(^S_FilterThreshHold, ^S_EdgeThreshHold, 0, 0 )
4=TemporalCleaner(6,11)
5=MergeChroma(blur(^blur_chroma))
6=MergeLuma(blur(^blur_luma))
7=AddBorders(0,^BorderTop,0,^BorderBottom)
^S_FilterThreshHold=8
^S_EdgeThreshHold=32
^blur_chroma=1.50
^blur_luma=0.1


Any suggestions?

[Holomatrix]

Originally posted by telemike
Couldn't get motion filtering to work with CCE 2.50 :(

I did get this script to work:


[AVISYNTH_STM_BilinearResize]
0=BilinearResize(^TargetWidth,^TargetHeight)
1=Undot()
2=Asharp(1, 4)
3=STMedianFilter(^S_FilterThreshHold, ^S_EdgeThreshHold, 0, 0 )
4=TemporalCleaner(6,11)
5=MergeChroma(blur(^blur_chroma))
6=MergeLuma(blur(^blur_luma))
7=AddBorders(0,^BorderTop,0,^BorderBottom)
^S_FilterThreshHold=8
^S_EdgeThreshHold=32
^blur_chroma=1.50
^blur_luma=0.1


Any suggestions?

Upgrade your CCE version :)

[DDogg]

Telemike, I checked 2.5 and it is working for me with the adaptive script above. When you say "Couldn't get motion filtering to work with CCE 2.50", what exactly do you mean? What happened?

[r6d2]

DDogg,

I cannot tell you how much I appreciate you were bored. This CCE+SemiAutoCropAndResize+MAF (spaces arguably important), has me really surprised.

Just encoded Solaris (the movie, not the Sun Microsystems Operating System :)), and I got the following:

CVD, Q=6
SVCD, Q=12

Both length=01:32:46 (credits cutted), Video BR=1073, Audio BR=96, only ¡1 CD!

Just unbelievable! And in terms of quality they compete hand to hand to Kwag's 1 CD TMPGEnc encodes (which I barely got to reproduce, anyway)!

I cannot think of the bunch of 2 and 3 CD encodes I have on my library... What a waste of CDs and storage space!

Kudos to you, man, and remain bored, please.

[tylo]

Now, where the hell IS Bach when you need him? :-/
Now, who needs Bach when you have DDogg? :D

[jorel]

Originally posted by tylo
Now, who needs Bach when you have DDogg? :D
:p

ha,ha,ha...
yeah,who?!?!?
hey DDogg,
stay "bored" and smoke some "crack" too.
(only kidding,you know,...i can't resist)
;)

@ r6d2
a long time i try to show that i got the
same results like you posted here my friend!

:)

[kwag]

Just unbelievable! And in terms of quality they compete hand to hand to Kwag's 1 CD TMPGEnc encodes (which I barely got to reproduce, anyway)!

I assume you are comparing MPEG-2, right ?
Because I am aware that CCE's MPEG-2 quality is better than TMPEG's MPEG-2 quality (at bitrates > 2,500Kbps). But quality wise, TMPEG's MPEG-1 is still not matched by any present encoder!
So on MPEG-1, there's still no encoder that can match TMPEG on low bitrates!
And TMPEG's MPEG-1 quality IS better that CCE's MPEG-2 quality at low bitrates. And that IS an industry fact ;)
Just a note ;)

-kwag

[dvd2svcd]

@Kwag: You know better than making posts like that. It's bound to start a flame on what's best and that is _not_ needed in this thread. Any further "this or that is best" in this thread will be deleted or edited by me! And if you'd show some good faith I'd prefer you to edit your post.

[telemike]

@ddog: I got Kwag's Adaptive Script to work! :D I must have scripted mine wrong.

In my test I noticed a HUGE compression increase over a non-filtered encode. Wow.


[AVISYNTH_Adaptative Filter_Anamorphic]
0=nf=0
1=GripCrop(^TargetWidth, ^TargetHeight+^BorderTop+^BorderBottom, overscan=^overscan, source_anamorphic=^source_ana, dest_anamorphic=^dest_ana)
2=GripSize(resizer="BiCubicResize")
3=Undot()
4=Asharp(1, 4)
5=STMedianFilter(^S_FilterThreshHold, ^S_EdgeThreshHold, 0, 0 )
6=MergeChroma(blur(^blur_chroma))
7=MergeLuma(blur(^blur_luma))
8=SwitchThreshold = (Width<=352) ? 4 : (Width<=480) ? 3 : 2
9=ScriptClip("nf = round(YDifferenceToNext())"+chr(13)+ "nf >= SwitchThreshold ? unfilter(-(fmin(nf*2,100)),-(fmin(nf*2,100))) : TemporalCleaner(6+nf,11+nf) ")
10=GripBorders()
11=function fmin(int f1, int f2) { return (f1<f2) ? f1 : f2 }
^overscan=1
^source_ana=true
^dest_ana=false
^S_FilterThreshHold=8
^S_EdgeThreshHold=32
^blur_chroma=1.58
^blur_luma=0.1
?=

[r6d2]

Originally posted by kwag
I assume you are comparing MPEG-2, right ?

Hi Kwag, I'm so glad you pay this forum a visit. Yes, I'm talking about MPEG-2. Unfortunately, I've not ben able to successfully mux MPEG-1 with DVD2SVCD, but I have done some attempts to use MAF with DVD2SVCD and TMPGEnc using MPEG-2 (as you suggested on a thread for DVD2SVCD), but I ended up with terrible quantization, above 10 in most of the movie.

I have seen screenshots in your forum to support your "industry fact" statement :), and I really admire your encodes, but I have found it difficult to match the quality you get, even starting from scratch and going by the book.

I invite you to try this MAF script with CCE+MPEG-2. You may be surprised.

[Holomatrix]

Originally posted by r6d2

I invite you to try this MAF script with CCE+MPEG-2. You may be surprised.

I think by asking KWAG to try CCE would be going against his Religion :) hehe

[r6d2]

Originally posted by DDogg
Now, where the hell IS Bach when you need
him? :-/

Seems that Bach found a new place to install his tent...

http://forum.doom9.org/showthread.php?s=&postid=355500#post355500

[jorel]

Originally posted by Holomatrix
I think by asking KWAG to try CCE would be going against his Religion :) hehe

this is not the case holomatrix!
the real reason is the quality of the results
with mpeg1 using MAX bitrate 2500 in tmpgenc.
you got fantastic quality in one cd with very long time source.

i ever use dvd2svcd with tmpgenc encoder to got my kvcds mpeg1
320x240 or 352x480 and the reults.....you know!
great progs,great scripts.....great results.
:)

[DDogg]

Hey, no hassles guys. Tmpg is great as well as the great sleeper, Procoder. I love Kwag's strong opinions as that drives a passionate search for quality as he has long demonstrated (and frankly did not get all the respect he deserved for it).

But, because of those strong opinions, I wanted to concentrate on giving dvd2svcd, CCE and one pass vbr a shot as these are a very fast method to produce a more or less standard svcd. This is very on topic for this forum whereas non standard xvcds are definitely not.

Also, I would greatly appreciate hearing Kwag's opinions on low bitrate encodes using ProCoder, both mpeg1 and mpeg2. Kwag, have you tried Procoder? I have a feeling it might make you take a second look.

[jorel]

of course DDogg,
no need preocupations.
;)

telemike is my friend here and in other forum too!
Kwag post lots of goods in kvcd forum about d2s too.
he is developing one good program called CQMatic
and have intentions to use CCE too.

the intention is good results and friendship.
i got more from friendship cos
all my results came from you all,not came from my own knowledge.
this is the idea.
problems....never.

:)

ps:
see how i'm "expert"...look my avatar!
:p

[DDogg]

@ telemike, something I did notice that may be interesting to you. I was using cce 2.67.00.11 for most of all my tests and I got to know Tylo's Q32 first pass bitrate yield pretty well (in my sleep).

When I tried 2.50 I noticed the first pass as about 200 higher. I only did it once so maybe it was a glitch, but that would tend to indicate 2.50 would not be as efficient as the later versions. Just a FYI for you to take a look at.

@r6d2Just encoded Solaris (the movie, not the Sun Microsystems Operating System ), and I got the following ...That is an incredible result! However, as we all know it depends on the source. I am trying to do the usuual sspxts which is 4:3 and having hell getting anything under 60. These are the encodes that I would guess would benefit by that second or third pass. I am just not up to speed on that and my eyes are giving me a lot of problems so catching up and reading 50 page threads are difficult.

[Holomatrix]

Originally posted by DDogg

@r6d2That is an incredible result! However, as we all know it depends on the source. I am trying to do the usuual sspxts which is 4:3 and having hell getting anything under 60. These are the encodes that I would guess would benefit by that second or third pass.
I second that, because I am trying to get the movie 'Phonebooth' below a Q of 63 with is only 80 min which I know would give me better results by doing a multi-pass encode instead. I just don't get why D2Sroba would pick a high Q on such a short movie.

[DDogg]

Holo, I know you do know damn well why that is :p:) All Tylo is doing is measuring the size of the pass. The size of the pass is based upon compressibility of a given source.

Take "the mtrx" for example. Green and dark 16:9, it compresses well so a pass is smaller. Take the favorite ball-buster SPR, damn near impossible to compress properly because of all the grain, fast jerky pans and water and high action shots. Pass is much bigger because of that. Tylo has to keep increasing the Q to hit the project bitrate he needs for sizing. I think I said that right :)

BTW, SPR might well be an excellent test subject for MAF.

[Holomatrix]

Originally posted by DDogg
Holo, I know you do know damn well why that is :p:) All Tylo is doing is measuring the size of the pass. The size of the pass is based upon compressibility of a given source.

:) Yes that I know, I don't think I type everything I mean to say, I had a discussion a little while back which I can't remember right now (memory is going weary) about what I thought the '?' in D2Sroba meant, I was under the impression that the '?' in amount of CD's would allow me, or D2Sroba tell me that "I have determined that your movie will fit on 1 CD at a Q of 63 or 2 CD's at a Q of 50, What do you want to choose?" then encode at that, or automatically pick 2 CD's if the determined Q is higher that your picked worst Q. If determined Q is worse than my worst Q then I wouldn't mind D2Sroba automatically doing more passes to determine the Q for 2 CD's. Does that make sense? :)

[Crackhead]

@holo and ddogg
got the same problem with Tomb Raider(88mins) and The Crimson Rivers(90mins)[cutted the credits of both movies] , D2SRoba won't go under a Q of 63!

The results of Tomb Raider weren't so bad, a lot artifacts around the faces, but not noticeable on TV.
Didn't finish The Crimson Rivers cause of the high Q.:mad:

I will do some more tests this week, but i think i won't get a q of 12 like r6d2 got:angry: :o

Greetz, Crackhead

[r6d2]

Originally posted by Holomatrix
I had a discussion a little while back which I can't remember right now (memory is going weary) about what I thought the '?' in D2Sroba meant,

I would not call it a discussion, just a conversation, and it is on the old thread. Anyway, what I tried to tell you then, and I quote myself here :), is:
Originally posted by r6d2
In "?" mode, it will let you choose only if two Qs will satisfy your requirent. If only one does, it won't ask and just proceed.

Originally posted by Holomatrix
If determined Q is worse than my worst Q then I wouldn't mind D2Sroba automatically doing more passes to determine the Q for 2 CD's. Does that make sense? :)
It makes a lot of sense, this is known as range mode: 1-2, 2-3, even 1-4.

Anyway, the "Really Worst Q" factor will eliminate all this hassle, if you're really going for quality. BR, even if it is low, is a result of the Q you set, not otherwise. I don't know if I'm being clear enough on this new parameter concept. If anyone has a question please welcome.

[Holomatrix]

I think I'm getting it now :) My brain doesn’t work like a sponge to much anymore, I think sometimes like a wall :) Thanks

[kwag]

Deleted! You were warned!

[dvd2svcd]

@Kwag: Next time I will strike you. I will not have your strong opinions start a flamewar! I have always been fair to you, so be fair to me too.

[r6d2]

Originally posted by Holomatrix
I am trying to get the movie 'Phonebooth' below a Q of 63 with is only 80 min which I know would give me better results by doing a multi-pass encode instead.
Beware: Multipass (2 pass, for instance) will not give you signifficantly better quality than OPV if you still confine the movie to 1 CD. Neither n-pass would.

Multipass will spread a little better the scarce BR available, but that's it. On 1 CD rips there's no much BR to spare.

So you may say, "I'll go for multipass 2 CDs then", to which a good answer would be: go for OPV, 2 CDs. Half the time, comparable quality.

[r6d2]

Originally posted by r6d2
Just encoded Solaris [...] and I got the following:

CVD, Q=6
SVCD, Q=12

Both length=01:32:46 (credits cutted), Video BR=1073, Audio BR=96, only ¡1 CD!

Just unbelievable!

Guys, I have terrible news for us all. You know, when something is incredible, unbelievable, wonderful, and all those spectacular first impressions you get, I always keep on thinking on it for a while. And most of the time I find out that too good to be true stuff is really too good to be true, if you know what I mean :)

So I did some more checking. I tried the movie above without any filtering, and with the simple, quicker and worthy UnDot()+Deen() companion. Here are my findings. The columns are sorted by descending Q.

Movie: Solaris
Length: 1:32:46 (133468 frames, 23.976 fps)
(no credits)
Audio BR 96 kbps Stereo
Target size 1 CD
Target MPV size 713 MB
Target Video BR 1.073 kbps
D2Sroba Default values

Setup Resolution_________________ SVCD SVCD SVCD CVD
Determined Q_______________ 14 12 10 6
Estimated Video BR_________ 1.123 1.073 1.092 1.071
Filters SemiAutoCrop & Resize______ Yes Yes Yes Yes
MAF________________________ No Yes No Yes
UnDot()+Deen()_____________ No No Yes No
Results Actual MPV size [MB]_______ 726 704 719 700
Encode Duration____________ 1:03 2:45 2:09 2:04
Target Size/Actual Size____ 98% 101% 99% 102%
CCE Speed__________________ 1,47 0,56 0,72 0,75

First of all I must say that the quality of all encodes is great. Even the unfiltered one. Let's see some conclusions:


MAF takes 1.3 times more than UnDot()+Deen(), and does not reduce Q as much.

In MAF, by its adaptative nature, actual sizes end up being lower than target sizes (you will not overflow), but error is bigger.

The use of UnDot()+Deen(), since not adaptative, predicts better. It cleans and homogeinizes the whole source in the same way, and probably more than MAF, so it allows to predict more precisely. It seems to overflow a little, though.

When you have a very clean and compressible movie with a low Q, not doing any filtering will save you a lot of time!

The incredible low BR of this Solaris encode does not yield on the MAF script, but on the compressibility of the movie (dark and green, and 16:9, same as DDogg's "the mtrx"). (So, what else is new :))


I'm certain this is a one movie only test, but at least clears up the "less that wonderful" results after all.

Things like this remind me of people still trying to build perpetual motion machines, turning lead into gold and stuff like that :)

The bottom line is, IMHO, MAF is great, but not much greater than simple known already filtering, and the most important thing is it will not do any magic, nor is the panacea to 1 CD rips. Just compresses the source a bit, and takes a lot of time. :)

[DDogg]

Ain't that just some doody in your oatmeal! :mad:

Just dropped my Q by 10 points using my standard deen(), undot() and undot() again after the import line of Gert's semi-auto over using semi-auto with MAF. Well, at least I know this encode like the back of my hand so I can do a good compare when I screen it.

[kwag]

The bottom line is, IHMO, MAF is great, but not much greater than simple known already filtering, and the most important thing is it will not do any magic, nor is the panacea to 1 CD rips. Just compresses the source a bit, and takes a lot of time. :)
That's not the case if you are encoding a very high action film. That's where the MA script will really shine, and reduce your file size considerably.
I do agree that if you are encoding a slow action film, some parts of the MA script are really not necessary. Like the "blurring" part. You will still benefit from the "Temporal" MA section of the filter, as it will help clean up very low action parts of the movie. But I would still use undot and a very low temporal filter statically on all frames, to reduce noise spots :)
It's really up to anyone to "Plug-in" whatever filter they want to make adaptive. The framework is already there, provided by AviSynth, so it doesn't necessarily mean it has to be a fixed Temporal/Blurr MA filter. The limit is the imagination ;)

-kwag

[r6d2]

Hi, Kwag, and welcome back to the forum :)

Word of mouth is you're doing some testing with CCE and DVD2SVCD. Did you try the MAF+CCE stuff?

Originally posted by kwag
That's not the case if you are encoding a very high action film. That's where the MA script will really shine, and reduce your file size considerably.

I'll try it with the Back to the Future Trilogy, guess it qualifies as a high action movie. Does it, according to your experience?

[kwag]

Originally posted by r6d2
Did you try the MAF+CCE stuff? No, but because of the nature of the MA design, it shouldn't make any difference no matter what encoder you use.


I'll try it with the Back to the Future Trilogy, guess it qualifies as a high action movie. Does it, according to your experience?
Sure it does! But a better test would be something like "Black Hawk Down" or something similar. That's where you'll see the real file size drop :)

-kwag

[r6d2]

Originally posted by kwag
[B]No, but because of the nature of the MA design, it shouldn't make any difference no matter what encoder you use.
Oh, my... I never though I'd hear you to say that! :D:D:D

If I only could get MPEG-1 working with TMGPEnc and DVD2SVCD to take advantage on the good low BR MPEG-1 encodes you do...

[DDogg]

Yep, my test encode using just standard semi-auto with undot(), deen() and undot() again after the import line was as good or better than the semi-auto w/MAF I did yesterday. Source 147 mins progressive NTSC, Bitrate was 907, Q for 1st run was 48 with bias 10.
480x480 4:3 full screen. Encoding method used was 1 pass Roba which is 2 passes total.

[r6d2]

Originally posted by r6d2
I'll try it with the Back to the Future Trilogy
I guess it's going to be Back to 2 Discs Trilogy... :(

SemiAutoCrop, standard: Q=63 (not by hard limit, but by actual BR)
MAF: Q=62
MAF modified with Deen() instead of STMEdian: Q=58

IMHO, all encodes suck, even on 100% visual you can see the artifacts.

I'd try SemiAuto standard with UnDot()+Deen()+UnDot() as per DDogg's suggestion, but I decided to go 2 discs anyway and put in a second audio track.

(@Kwag, I guess it was not a high action movie after all. Maybe we can ask Zemeckis to incorporate more action on a future remake to get it on 1 CD :D)

[Holomatrix]

Originally posted by DDogg
Yep, my test encode using just standard semi-auto with undot(), deen() and undot() again after the import line was as good or better than the semi-auto w/MAF I did yesterday. Source 147 mins progressive NTSC, Bitrate was 907, Q for 1st run was 48 with bias 10.
480x480 4:3 full screen. Encoding method used was 1 pass Roba which is 2 passes total.

So are you doing this? ;

Undot()
Deen()
Import Semi ...
Undot()

for me file size was only a bit smaller.
Tested a chapter at 50 Q bias 14 with SemiMAF =20,065 meg, Semi Un+Deen+un =20,040 meg

[telemike]

I just did Daredevil. 480x480 SVCD. DVD2SVCd with MAF script. CCE speed = .650 D2Sroba came up with a Q of 38 for a 1 Cd encode.



========================================================
D2SRoBa v2.4.4 - DVD2SVCD plugin, by tylo
OS Version: WIN_XP
--------------------------------------------------------
- DVD2SVCD Shutdown
- 2003-08-10 09:33:49
--------------------------------------------------------
Settings:
- CD size : 800
- Number of CD's : auto
- Worst Q. factor : 40
+ One pass VBR Robshot/Bach : 1
- Sample percentage : 2.0
- Sample GOPs : 1
- Adjust Q. : 0.8
- Safety Zone : 34
+ Do credits bitrate tweak : 0
- Start time : 0:00:00
- Bitrate : 1000
--------------------------------------------------------
- Movie length : 01:43:35 (149030 frames, 23.976 fps)
- Audio size est : 124315983. bytes (160 + 0 kbps)
- CD user data size : 833237920. = (800 - 5)*1024*1024 - 380000
- SelectRangeEvery : every 750, select 15 frames
- Num. sample frames : 2985
--------------------------------------------------------
Calculations:
- Estimated mpv size : 666565608. (Q=40, 857 kbps, 13350992 sample sz)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- Number of CDs : 1
- Target mpv size : 689790437. (887 kbps)
--------------------------------------------------------
- Binary search for Q:
- Estimated mpv size : 759827533. (Q=32, 977 kbps, 15218984 sample sz)
- Computed mpv size : 713196571. (Q=36, 917 kbps, (Q40+Q32)/2)
- Estimated mpv size : 687524668. (Q=38, 884 kbps, 13770792 sample sz)
- Estimated mpv size : 697181213. (Q=37, 897 kbps, 13964208 sample sz)
--------------------------------------------------------
- Q Weight : (689790437. / 49.9262981574539 - 13770792) / (14284988. - 13770792)
- Adjust Q : (38 + 1.0 - 8.82587379954006e-002 - (0.8)) = 38.1117412620046
- Determined Q. : 38
--------------------------------------------------------
Replace values in ecl file:
- opv_q_factor : 38
- opv_brate_max : 1931 ==> 2530
- vbr_brate_avg : 6000 ==> 887
--------------------------------------------------------
- Encoding Movie
- 2003-08-10 09:54:50
--------------------------------------------------------
- Actual mpv file size:
08/10/2003 12:32 PM 690475136 Encoded_Video_CCE_NTSC.mpv

[r6d2]

Originally posted by telemike
I just did Daredevil. 480x480 SVCD. DVD2SVCd with MAF script. CCE speed = .650 D2Sroba came up with a Q of 38 for a 1 Cd encode.

Not bad. I think though that 160 for audio is too much for a 1 CD rip. The 10% rule advises to go for 96 in this case, which would give you a 951 BR, likely to be a Q=34 for video.

[telemike]

I usually use 160k audio for 2-cd encodes and 128k for 1-cd's. Howver, I thought D2sroba was going to do a 2-cd encode since I set 40 as my worst Q.

[Crackhead]

I do some tests on this topic, too, so can someone post this undot(),Deen(), Semi(),Undot()-Script, cause I'm a total newbie on this.
I wouldn't get it work in 5 years!:angry: :angry:

My first results on this topic is, that it doesn't matter what script or how many cds you take, the human eye won't realize any change!:D :D

The TV-screen is so "bad", I didn't notice the artifacts which are clearly visibel on the PC-screen, and I think noone would!:cool: :D

I'll go for a 1cd encode of every movie 'til I get a DVD writer, the options to get a satisfying 1cd encode are quite good, good enough!

This is just my opinion, so feel free to persuade me of another opinion, but plz with some understandable, no theoretical or technical stuff!:D :D ;)

Greetz, Crackhead

[r6d2]

Originally posted by Crackhead
The TV-screen is so "bad", I didn't notice the artifacts which are clearly visibel on the PC-screen, and I think noone would!:cool: :D
IMHO you are very close to the absolute truth here, if such a thing exists :D
I'll go for a 1cd encode of every movie 'til I get a DVD writer, the options to get a satisfying 1cd encode are quite good, good enough!
However, why a DVD writer would change your mind then? Your 1CD encodes will continue to look good enough!
This is just my opinion, so feel free to persuade me of another opinion, but plz with some understandable, no theoretical or technical stuff!:D :D ;)
OK, not trying to evangelize here, but my thoughts on the matter are the following (I consider myself just a movie fan and I'd love to hear the experts too in an understandable language):

My next TV will (hopefully) be an HDTV. Its resolution are about 720x1280 (Progressive) and 1080x1920 (Interlaced). That's quite more than double what we have today. My DVD player has a zoom mode (I guess most do). So, when encoding, I do my tests in 2x Zoom mode. If it's not too blurry, my bet is that this encode will look reasonably well on an HDTV.

There are currently players being developed to support this kind of resolutions (as well as DivX! and MPEG-4 in general!)

On sample here:

http://www.keohi.com/keohihdtv/videodevices/momitsuv880/momitsuv880_background.html

As I said, this is my bet. I don't know if it is a good one :)

[telemike]

I have a 43" HDTV.

SVCD's done at 480x480 look great. CVD at 352x480 looks blocky and jaggy.

I don't use my Terapin VCD recorder anymore since it's CBR1150 VCD @ 352x240 and it it blocky but watchable.

[Barker]

I just purchased Gangs of New York 2 CD set which has the movie split into 2 discs, which, when I realized this I thought it would be a great test for this (3 cd's would be a pain to make out of 2 different DVD's; ~1:30 a piece). I used this method, and ended up with Q64 (not surprising, because it's over 4 gb's to begin with (one part), and I'm doing 128 main, 96 comments audio as a test). Other movies that I have done without the MAF script (and without 2nd audio) do not look anywhere near as good as this one, even with lower Q values. I'm not sure what is causing this, but if anyone has this movie and is willing to try, please do, and tell me I'm not crazy?

If this is the results I can get with this script, then I will never go back to the original method I was encoding with.

Thanks for all your suggestions, DDogg, and everyone else in this thread!!!

Regards,
Barker

[DDogg]

Barker, as I am sure you know, it ALL depends on the compressibility of a given source. Some 90 minute sources will look pretty decent, whereas some will not. This is definitely not a one size fits all kind of thing. As mentioned before are the classic examples of "The Mtrx" and "SPRyan". Forgetting the lengths, the first example is material that compresses extremely well, while the second is pure hell and normally requires nearly twice the bitrate for the same section of length (that maybe is slight exaggeration to make the point).

You might want to play around with making a notation of the size of the first Q32 prediction run (D2SRoBa_Q32.mpv) of D2SRoBa on a source you have found to have YOUR minimum acceptable quality on 1 CD for YOUR viewing vehicle. Chances are (not always), if another encode has the same or smaller size for that same prediction pass, it will be ok to at least try to put it on 1 cd (assuming you used the same script and the same settings in CCE). This can save you a lot of useless attempts and wasted time and it takes into consideration both the compressibility AND the length of the source.

Here is an example of my own personal "Compression Index" (I always use the same test script and same test cce settings [Edit: sample size in D2SRoBa for this test was 1%]:

1> "Final Dstination" --- D2SRoBa_Q32.mpv = 8,196KB, Len -- 133341 frames Time 1:32:41 Target mpv size:-(1043 kbps)
2> "The Usual Sspxts" - D2SRoBa_Q32.mpv = 10,380KB, Len - 148070 frames Time 1:42:55 Target mpv size:-(926 kbps)
3> "The Mtrix" -------- D2SRoBa_Q32.mpv = 8,388KB, Len --- 186356 frames Time 2:09:32 Target mpv size:-(709 kbps)


I know from experience that #1 and #3 (barely) on 1 cd is acceptable to me, whereas #2 is not. So, if the D2SRoBa_Q32.mpv is much bigger than 8,300 I normally just go for 2 cds and same myself a lot of aggravation. Note above how the target bitrate is not the indicator of quality!