Mencoder Extra Settings Makes Quality Worse?

[Fluffbutt]

From Pal dvd to Pal vcd (kvcd).

vlelim=-4
vcelim=9
lumi_mask=0.05
dark_mask=0.01

I got from another site that listed mencoder 'must have' setting (didn't say WHY though!).

I tried these, (and worked out what the hell they do!!) - results are not good, very bad blocking over plain surfaces. Although file size drops by 10%, quality goes down by about 25%.

For those that don't know, I gather that the two mask's set up a threshold for whisch to consider areas for quality encoding (as in 'everything below (or above) this is one area"), and the lim's tell mencoder how much to vary the quality for those masked areas.

(corrections welcome)

Leaving these out of the bat file raised the size from 600 to 655, but gave a quality increase of less-than-vcd to above-svcd (almost dvd!!).

I've also tried vrc-eq=avgTex, but didn't see any difference.
Also used :vhq as well, results seem better, less blocking on edges (??)

Anyone else with ideas and setting to try?

[incredible]

As you may have already seen, here's a Thread where we (mostly biul and DigiDoc and Riu) have tested a very lot of the mencoder parameters.
http://www.kvcd.net/forum/viewtopic.php?t=9074

There you can find a lot of answers .... but its loooong

[yaz]

Quote:

Originally Posted by incredible

... there you can find a lot of answers ...

ehrmmm ... yep ... & there u can find also the opposite of them. i must say (with the greatest respect of mine !) that thread is a big confusion for a noob. it just hurts.
that's why i've just started a systematic investigation of mencoder options. first results are very interesting but i'm stopped. my present problem is that i couldn't find a reliable mpeg2 stream analyzer. imho, bitrate viewer simply cheats. i've tried 3 diff versions & each(!) gave me different results. diff was more than significant, sometimes about 50% (!?!?!)
what do u guys use or recommend for such a purpose? i just wanna check the bitrate (min/avg/max) & the quantizer distribution.

if i have sg relevant i will pile it up here (or where?)
the bests
y

[incredible]

Youre right, thats why I said ".... its loong"

Well a resumeé as I see can be found in Vmesquitas MencoderME Templates as they do contain the summary of that thread there, ... as I understood this.

Quote:

bitrate viewer simply cheats

I do quote KIKA (online transl.Tool used)

Quote:

Because a high level means only that a block contained many same values after quantization, and which can mean also result of a perfect interaction of quantization and RLE coding! And that again means that in such a case highly one quantized, the image quality however on the highest possible level was! Since however hardly someone really understands, what has it with the q-level actually on itself, I leave rather the fingers of interpretation attempts.

From this site: http://www.edv-tipp.de/gastbeitraege/kika001_dct.htm

Thats also the reason WHY I dropped Bitrateviewer to see the encoding Quality in mathematical "translation" ... dont know to say it better in english

CCE outputs at th same avg bitrate almost a double of Q factor heights. But the result checked by eyes is just a bit worse .... so

There are several mpeg analyzing tools (I do saw some of them on the www but the good ones do cost a HELL!

[yaz]

@incredible
thx, man !
kika : wufff ... i feel i need a heavy brush-up on my german
options : is there any interest in a systematic investigation here around ? imho, mencoder really do worth a try.
btw, mencoder ... u do still owe me an athlon compile accepting ffdshow avis (just to remind u )
the bests
y

[Fluffbutt]

Umm, yaz - i'm using an athlon compile here that accepts ffdshow avis - it's the main base for my movies. Dvd2avi then ff (oh, sorry its ffvfw..) oops

Incredible - yas, saw that forum. But as far as I read nearly all the discussion is about dvd output (kdvd), not k(s)vcd. I'd imagine the setting must change for such low (compared to dvd) bitrates, after all we're talking a min-ave-max of 550-900-1850, not 3000-5000-8000...

[incredible]

Quote:

Originally Posted by yaz

btw, mencoder ... u do still owe me an athlon compile accepting ffdshow avis (just to remind u )
the bests
y

And you do owe me a possibility to get d2v sources working with mencoder!

Ok, ... I did provide in the very end of that compile thread one of the latest (one week ago) builds .... but one week is much in mencoder developing time I think ... and Vmesquita talked about a new version which also got some more IVTC bug fixes and addings.

According to FFdshow .... just tweak the codecs.conf file ... even you did tetermine a internal codecs configuration, when starting the .bat mencoder searches for that config file anyway ... I did find that out by just deleting one codec description in the codecs file ... he counts one codec LESS when showing the log! So mencoder does take not whats IN that codecsfile on your disk.
So do use the ffdshow configurations of the config file as mentioned by RDS_Correia in that compile thread!

The bests also to you

I hope you understand that I wont get away from my ffvfw makeavis as my system is running perfect ... and as we know ... never change a running system as I do not got these Issues riu talked about which do occur when using the ffvfw one instead of the ffdsho one.


BTW... that codec workout (VFAPI or d2v) was mainly intended to see if any windows codec driver could be used for mencoder to decode inputs no matter which one IF its supported by Windows installed codec library/dll's.

It would be really a big SHOT to get d2v directshow codec working in mencoder as by this people using Win98 can directly encode from VOBs as FAT32 doesn't support a "one movie VOB" in its big size.

VFAPI wouldn't be my recomm. in a case of Win98 as VFAPI converter and its internal processing works with RGB!!
But by the way its faster than makeavis!

[digitall.doc]

Quote:

Originally Posted by incredible

VFAPI wouldn't be my recomm. in a case of Win98 as VFAPI converter and its internal processing works with RGB!!
But by the way its faster than makeavis!

Inc,
is it already done?. Is there a way to tweak codecs.conf or config.conf to work with .avs files directly within mencoder, using the same VFAPI we use within TMPGEnc or DVD2AVI?.
It would be great, for those of us that learned with avisynth (I asked in other thread about mencoder filtering, but it seams we don't manage still well with those filters...)

[incredible]

I think theres no way to get avs "directly" working with mencoder.exe (IMHO but that could change) ... as I think mencoder needs FourCC descriptions in the inputted container. And that provides for example makeavis or VFAPI reader/converter generated avis , as it ends up in an recognisable "avi container" for mencoders eyes.

I think even DVD2AVI does a diff. way than directly frameserving via VFAPI as this would mean IF d2v would be frameserved via VFAPI ... only RGB streams would be the result when importing that d2v in avisynth

[Fluffbutt]

Umm, am I doing something 'wrong' with the process I'm using?

dvd2avi d2v - avisynth with filters - ffvfw fake avi - mencoder - mpg
mux with headac3'd mp2 from ac3 from dvd2avi - vcdeasy to cdr.

I get the impression from digitall.doc ("..It would be great, for those of us that learned with avisynth..") that avisynth filters don't work this way (they sem to, DCT really slows mencoder down to about 2 fps (from 8 )

[yaz]

hmmm ... maybe i'm just dead-brained but there seem to be some confusion here.
mencoder can't eat anything but 'containers' which are identified exclusively by their fourcc. the proper decoder is read from the codecs.conf according to this. so, there's no way of feeding scripts like avs or d2v directly.
the solution is a 'wrapper' prog (like makeavis, vfapi, ...) they simply mount a fake header onto the script with a unique fourcc. it's not framserving but wrapping.
but ... d2v can be fed directly into some prog (e.g. vdubmod, tsunami, mpc,...). they all have their own parser/interpreter for handling such scripts & some of them can serve (vdm, tsunami, ... ) however, it's just 'by-passing' of decoding to raw avi.
so ... inc., my dear, there's no way of direct feed for d2v. sorry, pal ! the more i think the more i'm convinced that i simply fed m2v instead of d2v & i was very proud of ... nothing
but, as prodator pointed out, there's not much sense of forcing it as d2v can be served via avisynth. on my athlon xp it does not make much difference as regards speed. encoding from vobs runs about 44-45fps & from a vob-d2v-avs-avis chain it's about 42-43 (no filtering, of course).
@digi.doc : the way of hacking codecs.conf is in the 'how to compile ...' thread (we haven't found a better place to hide it more )

the bests
y

@inc. the newest menc from vmesquita is not optimized for athlon so ... it's your turn again

[Prodater64]

But why quenc can?

[incredible]

@ yaz

Ill check the new CVS tonight ...

@ Prodater

I think Nic (just assuming) did implementate a core where an internal AVS to AVIcontainer generation will be proceeded --- thats what he calls there in the thread "... The medium of dance ... " maybe based on a source of makeavis or that "AVS2AVI.exe"

@ Digi.Doc

... so as you read above ... we should check that avs2avi as it maybe will a bit faster than the makeavis? --- as every fps will count on very slow/old machines/CPUs of some members

[digitall.doc]

Quote:

Originally Posted by Fluffbutt

I get the impression from digitall.doc ("..It would be great, for those of us that learned with avisynth..") that avisynth filters don't work this way (they sem to, DCT really slows mencoder down to about 2 fps (from 8 )

Hi Fluffbutt,
your way is OK, above all if it works well for you. As Inc stated, the important thing is find a way that things work, and don't touch it (well, you'll have to change mencodeme and mencoder.exe versions).
I don't really know what did you understand from my post. The thing is that I learned (little) filtering videos with avisynth filters. And, apart from MA script and Static Script v4, we all keep some scripts we used on an special situation, or for some sources (captures, interlaced,...). And if we need to change a thing, we (more or less) know what filter to apply or tweak... I learned (little) this with avisynth. Mencoder has several filters in, that I'm not used to and don't know how to tweak. That's why it would be great if I could keep avisynth in mencoder, without losing speed...
This is what I wanted to say, but more explained know.

Quote:

Originally Posted by yaz

the solution is a 'wrapper' prog (like makeavis, vfapi, ...) they simply mount a fake header onto the script with a unique fourcc. it's not framserving but wrapping.

yaz, understood now.
But are wrapers supposed to slow down the proccess a lot?. Because I tested just a resize with and without avisynth, and seemed to me that was faster with avisynth... but when tried with a hole script, it really slowed down a lot (slower than TMPGEnc).

Quote:

@digi.doc : the way of hacking codecs.conf is in the 'how to compile ...' thread (we haven't found a better place to hide it more )

Quote:

Originally Posted by Prodater64

But why quenc can?

Well, you can read Inc explanation... anyway I'd like to have it also implemented in mencoder or mencodeme...

Quote:

Originally Posted by incredible

@ Digi.Doc

... so as you read above ... we should check that avs2avi as it maybe will a bit faster than the makeavis? --- as every fps will count on very slow/old machines/CPUs of some members

Where am I to begin looking for AVS2AVI?

[incredible]

Quote:

Originally Posted by digitall.doc

Quote:

Where am I to begin looking for AVS2AVI?

The download:
http://www.divx-digest.com/software/avs2avi.html

As I thought ... a commandline app. ... easy to integrate in GUIs


BUT! We should see which Fourcc and therefore which decoder has to be used in mencoders codecs.conf as till now I didn't use it .. so Ill look into it this evening.

[vmesquita]

If I remember correctly, AVS2AVI is not like MakeAVIS, it's more like a command-line AVI encoder.

[digitall.doc]

Quote:

Originally Posted by vmesquita

If I remember correctly, AVS2AVI is not like MakeAVIS, it's more like a command-line AVI encoder.

Does it mean it will encode first an avi from avs, and then feed it to mencoder?. If the answer is "yes", that's not good

[vmesquita]

Quote:

Originally Posted by digitall.doc

Does it mean it will encode first an avi from avs, and then feed it to mencoder?. If the answer is "yes", that's not good

Not really. It's an AVI encoder, you can use to encode to DivX or even to encode to mpeg1/2 using FFVFW, but only that. Imagine it as a "command-line very stripped down virtualdub"

[incredible]

Yep! It is ... sorry

[digitall.doc]

Quote:

Originally Posted by vmesquita

Not really. It's an AVI encoder, you can use to encode to DivX or even to encode to mpeg1/2 using FFVFW, but only that. Imagine it as a "command-line very stripped down virtualdub"

Not useful for our purpose