new giude for DivX4

[dado023]

http://www.med.uoc.gr/~diamantop/divx/index.html

for new users it will be very informing

best wishes

[kheperi]

hmmmm...
It seems you had a close look (cut and paste) at Toms hardware site when designing your website :-)

[MxxCon]

haha, indeed look like tom's hardware :)

dado023, nice work. you spend so much time tweaking video quality and you totaly ignore sound quality. your VBR settings of 64-192 will give pretty crappy quality. i suggest you check out --alt-presets by dibrom

[everwicked]

It's actually my guide :-) (page title?)

Nice comment, it was time for someone to point something to correct. I will do some tests on it, thanks.

[Tr!n!]

damn u dado, seems like u cross posted :eek: :(
there's just same thread in General/Newbies section

[everwicked]

And he linked to the guide, not the page :-/

Oh well...

[Tr!n!]

m'kay, nevermind then...
but would appreciate it if some1 would take a look @
the questions i asked in the General/Newbies thread, tnx ;)
And btw....have a good time tonite all :p:p:p
cya

[dado023]

sorry for bad english

something is wrong here,,,,this is not my guide...i m not capable of that.....on the bottom it sayz "Copyright everwicked© - 2001"
I noticed this guide(post)few days ago at http://www.divx.com/forums/index.php and i assumed that others would like to read this master p ,just wanted to share it with all other members.All i wanted to do is to get more people to read this guide and yes i cross posted because why wouldn`t newbies learn something about advanced divx4 encoding...






:(

[kheperi]

thats ok. You didnt claim that it was your guide. i assumed that falsely. I just couldnt resist of posting my rather superfluous contribution... :-)

[everwicked]

Hehe, site is about to change dramatically, hold on tight :D

[NeVeRLiFt]

Originally posted by everwicked
Hehe, site is about to change dramatically, hold on tight :D

Your lame!

[everwicked]

I couldn't possibly fit the description of yours in just one post, maybe i will write a book about it sometime! :D :D :D

[NeVeRLiFt]

:D thats was just best come back!
/me slaps everwicked upside his warped head with my Nandub encoded DivX movies!

[everwicked]

Time will tell.

Stop being a little baby.

[everwicked]

Lovely news for you :)

Domain is up at http://www.everwicked.com

[NeVeRLiFt]

bah!

[MxxCon]

it's all nice and dandy, but why did you get .com? :)
your not a company or corp..that's just silly when ppl reg domain w/ their nick and .com
also don't you think that vbportal is a bit overkill?:) you don't have that many news or other stuff going on so use even 50% of it's features :)

[Tr!n!]

man where are u talkin about?
just like every .com site is a company..
and about the vB portal: like it here and its alot nicer than much
sites on the net with ugly frams and stuff

Nice work everwicked :)

[Peters]

@everwicked I'm still waiting your amswer...


Did you have a look to analyse.log?
If you do , then you will notice that all the nice work and efforts of Dmitry are useless since Quantizer's limits are not respected

look at the KF quantizer, did you get Quantizer equal to 2?
I don't think so..
Try Q min of 4 and Q max of 4 by ex for DF
Did you get Quantizer equal to 4?
I don't think so..

If i'm wrong, i will be happy to see your logs..

[Acaila]

Actually, the only time the log file changes is if you use different quantizers.

Settings at 2-pass 1st pass:
Bitrate: No Effect
Rate control averaging period,frames: No Effect
Rate control reaction period,frames: No Effect
Rate control up/down reaction: No Effect
Maximum quantizer: Logfile changes completely
Minimum quantizer: Logfile changes completely

By changing quantizers you change quant, texture,motion and complexity in the resulting logfile. The change in quants is NOT used for 2nd pass (because if you set them to 31 the 2nd pass log still won't have quants of 31), but the rest IS.

So in Everwicked's guide it's useless to set bitrate to 6000 and the rate controls to 200.000/50/5 because they don't change the logfile AT ALL.

[Peters]

Originally posted by Acaila
Actually, the only time the log file changes is if you use different quantizers.

Settings at 2-pass 1st pass:
Bitrate: No Effect
Rate control averaging period,frames: No Effect
Rate control reaction period,frames: No Effect
Rate control up/down reaction: No Effect
Maximum quantizer: Logfile changes completely
Minimum quantizer: Logfile changes completely

By changing quantizers you change quant, texture,motion and complexity in the resulting logfile. The change in quants is NOT used for 2nd pass (because if you set them to 31 the 2nd pass log still won't have quants of 31), but the rest IS.

So in Everwicked's guide it's useless to set bitrate to 6000 and the rate controls to 200.000/50/5 because they don't change the logfile AT ALL.

You're wrong... by ex look at divx.log with a bitrate of 500 (without changing other parameters) and with a bitrate of 6000

Of course, they are different...why don't you think we put the calculated bitrate in the 2pass-1st pass? :)

[Acaila]

You think I would post that if I hadn't tested it?

My log file was absolutely identical whether I used 6000 or 1000 bitrate.
The only time I got a difference was when I changed the quantizers.

[Peters]

Originally posted by Acaila
You think I would post that if I hadn't tested it?

My log file was absolutely identical whether I used 6000 or 1000 bitrate.
The only time I got a difference was when I changed the quantizers.

you're wrong, i repeat...
And of course, i've tested and why don't you think that bitrate can only be changed + or - 20% in the 2nd pass.

So, try again with a res like 640x272, a low bitrate like 500 and a bitrate like 6000, you will see...:)

[everwicked]

Originally posted by Peters


You're wrong... by ex look at divx.log with a bitrate of 500 (without changing other parameters) and with a bitrate of 6000

Of course, they are different...why don't you think we put the calculated bitrate in the 2pass-1st pass? :)

First pass is with a bitrate of 6000 and quant 2/2 so what you propose does not make sense.

I am sorry that i haven't posted a reply to you but i have been on the site too much... I will do a test and post results soon hopefully.

[Peters]

Originally posted by everwicked


First pass is with a bitrate of 6000 and quant 2/2 so what you propose does not make sense.

I am sorry that i haven't posted a reply to you but i have been on the site too much... I will do a test and post results soon hopefully.

Are you talking to me ? :)
I just reply to Acaila who says that bitrate has no influence on divx.log but only quantizer and i disagree with that, that's all
Don't be confusing...i don't propose anything, i just told you that following your guide quantizers for KF and DF are not respected in regard of what you ask

[everwicked]

Hm, there should be no difference between several bitrates in the first pass, it should be the second one that makes a difference..

Dmitry says that there incompatibility issues with divx4log 0.38 and divx 4.12? Anybody knows more about that?

[Peters]

So, everybody is crazy? :)

Don't forget to tell Doom9 to modify his divx guide. No matter to put the calculated bitrate in the 2ndpass-1st pass! lol :)

[everwicked]

I'm talking about what SHOULD be. Every calculation should be done in the 2nd pass but apparently it's not.

Sorry for the confusion, i shouldn't post stuff when traveling for 20 hours :(

[Acaila]

Well, I hate to disappoint you Peters, but I've run the test again, and checked and rechecked, but I get exactly the same logfile whether I use a bitrate of 1000 or 6000.

Maybe my system is screwed?

Can anyone please confirm either Peters' or my own results?

And in case anyone dares say it, "No, I'm far from a newbie"

[Peters]

Acaila, i don't tkink you're a newbie and i trust you when you say you got same logs with a 1000 and 6000 bitrate (maybe let try to lower 500?) , very strange

But everywhere, in every divx4 guide, it's indicated to put the bitrate in the 2pass_1stpass ans in the 2pass_2nd pass, right?

Well, just the 6 first frames of a 2pass_1stpass (from a small clip)

One encoded at 800 kb and the other 6000 (Divx 4.12)
All the other parameters (Quantizers Max=12 Min=2, etc...) are left as default.

Notice that only the 2 first frames of each log are equal

Divx800.log
------------

##version 2
quality 5
Frame 0: intra 1, quant 4, texture 169295, motion 0, total 169295, complexity 2116187
Frame 1: intra 0, quant 4, texture 87114, motion 3341, total 103616, complexity 1088925
Frame 2: intra 0, quant 5, texture 65007, motion 3319, total 76582, complexity 1300139
Frame 3: intra 0, quant 5, texture 65637, motion 3292, total 75862, complexity 1312739
Frame 4: intra 0, quant 6, texture 42051, motion 3102, total 51343, complexity 1314093
Frame 5: intra 0, quant 6, texture 51730, motion 3172, total 60299, complexity 1616562


Divx6000.log
-------------

##version 2
quality 5
Frame 0: intra 1, quant 4, texture 169295, motion 0, total 169295, complexity 2116187
Frame 1: intra 0, quant 4, texture 87114, motion 3341, total 103616, complexity 1088925
Frame 2: intra 0, quant 4, texture 86206, motion 3492, total 99266, complexity 1077575
Frame 3: intra 0, quant 4, texture 83634, motion 3317, total 95245, complexity 1045425
Frame 4: intra 0, quant 3, texture 110010, motion 3414, total 123697, complexity 733399
Frame 5: intra 0, quant 3, texture 123420, motion 3529, total 135001, complexity 822799

To complete, if you choose Qmin=2 and Qmax=2 of course the bitrate don't matter.
If i do a 2pass_1st pass with these quantizers, i obtain a different log from the 2 previous but in fine when the 2nd pass has ended the files created with 6000 (Maxq=12, Minq=2) and (Maxq=Minq=2) are exactly the same (same size, same bitrate)

Well, if you want, i can put you a link for the little vob i've used
(No Copyright), just to compare my results with your's


see you soon...

[Acaila]

Okay, that was stuuuupid!

Yeah, I had MinQ=MaxQ in all my tests, so that's why I ended up with the same log however I changed the bitrate.

So changing bitrate from 6000 to 500 wouldn't change a thing if MinQ=2 and MaxQ=2. But if MinQ=2 and MaxQ=12 then changing from 6000 to 500 would change a LOT.

I still don't understand how DivXlog would be any help though, because you'll make a logfile using CBR. In my opinion VBR 2-pass would always result in a better quality, unless you change texture/motion/complexity for every frame, instead of whole zones, manually.
But I'm not an expert on this program, so I don't know exactly how it works.

P.s. I just spend the two weeks off I had around christmas and new year on DivX encoding. I can say with certainty now that it screwed up my perception and logical thinking BIGTIME. Remind me to get a life :D

[Peters]

Originally posted by Acaila
I still don't understand how DivXlog would be any help though, because you'll make a logfile using CBR. In my opinion VBR 2-pass would always result in a better quality, unless you change texture/motion/complexity for every frame, instead of whole zones, manually.
But I'm not an expert on this program, so I don't know exactly how it works.


I agree with you and in fact all the tests i made show that Divx4log is a nice program but don't reach his goal (keeping the quality asked)
Notice that Blacksun , the first to make a divx analyser, stops working on this prog for the same reasons...

[pulsedriver]

Everwicked:

In your divx guide you suggest to use "fast recompress" instead of "full processing mode" for both pass 1 and pass 2. I've always been using the latter. Is there any difference in speed / quality?

[FactorM]

@pulsedriver:

you'll get a noticable speed improvement and no quality loss, because all filters are handled by avisynth. you need full processing mode in virtual dub if vdub makes the filter work.

[Dionysus]

everwicked, what's the deal with your site. I haven't been able to get on it for a while now. I want to look at your guide for the Mpeg4cd but it says that they cant fidn the site. if you get this is there any chance you could email it to me. blair_locken@hotmail.com thanx man. if not just tell me when or if it's fixed. thanx man.

[Tr!n!]

hmmz, seems to be offline :-\
btw there is only a GKnot guide...

[Acaila]

In your divx guide you suggest to use "fast recompress" instead of "full processing mode" for both pass 1 and pass
2. I've always been using the latter. Is there any difference in speed / quality?

If you use full processing mode in VDub then your video will be converted from YUV to RGB and back. This slightly changes the video, but it's hardly noticable. When you use fast recompress (and no extra filters in VDub) then you won't have this conversion and final quality will be more like original video.

[everwicked]

Was the site down sometime? When was that?

I have to spank my monkey admins then :D

[pulsedriver]

Well, if you're busy "spanking your monkey" I can understand if the site goes down every now and then... :-)

Sorry. I just had to post that.

[everwicked]

monkeys=admins of www.everwicked.com hosting company

I haven't started spanking them yet but i should i reckon :D :D

[pdontthink]

the site appears to be down again... :(

[everwicked]

There was a downtime yesterday. I apologise for that.