@ DigiDoc

So you do deal with 1500kbit in average? anamorphic? 704x576?
And where do you see those blocks (8x8)??? During the whole framecount or only where fast moving OR dark scenes do occur??
Whats the source... DVD or mpeg4 compressed avi?

Maybe your source already got some higher quantisation on those parts which you just in this encoded state cannot see?

So a ...
vf=yuvcsp,pp=hb:100:100/vb/dr
... could do a little postprocessing in mencoder for you?!
Or if you feed your source via Makeavis use the BlindPP(CPU=4) deblocking in your AVSscript.

Best would be to show a pic of your frame of the source and the SAME of the encoding! I will then treat the source pic in Photoshop just to see whats going on deeply in the source.

Ok, I just finished encoding a 99 minute movie with 2-pass mencoder and an average bitrate ov 2120kb/sec. The quality is superb. According to BV the highest Bitrate peak is ~8000kb/sec. I'm so amazed by the quality.
Haven't tested against CCE, but I won't need that program from now on I fear. 8)
I used a simple script without filters except BlindPP(cpu=4), Undot(), a Blockbuster Line with variance 0.4 and DCTFilter. Speed was about 28fps on my XP2700+ :!: :!:
This is the encoder I dreamed about the past years! :lol:
Thanks for all your work and testings everybody! :)

@bilu
This is not 3pass in the same sense CCE would do. It is 2 PassVBR with a audio-only extra pass. :roll: CCE 3-Pas VBR means encode three times and keep updating the log file. I was never able to understand how can a encoder discover better ways to distributing bitrate between passes, but that's the way it works. :wink:

@inc
I was doing my tests visually, using Avisynth'a stackhorizontal to do frame-by-frame compare between the output of different options. I just posted the BV quantization sum. I didn't knew it was inaccurate, also. :(

Go to the mencoder filtering thread or read the mencder manual and you will se that exactly the filters of your purposes are also present in mencoders internal engine! :D
http://www.kvcd.net/forum/viewtopic.php?t=10242

Means ... it gots a postprocessor, a spatial filter and a noise generator and more like temporal denoiser etc.

... so we should try speed/Quality comparisons between direct vob inputs/menocder filters and the known Makeavis inputs/avisynth filters.

:D

@ DigiDoc

Here I found a special "starwars" test on the mencoder postprocessings :wink:
ftp://ftp.mplayerhq.hu/MPlayer/tests...s/highres-png/

@inc,
thanx a lot for the feedback. Whenever I ask for help, you're one of those that are always there (maestro).
Well, my tests where with average 1500 kbit, and 704x576 anamorphic.
Related to microblocks, I'd say that they appear all over the framecount, but are more noticeable in certain scenes as those you commented (dark scenes, dark flat surfaces, fast scenes). In fast scenes also appear some macroblocks.
I'll give it a shot, since it looks like the images you pointed me in your next post.
... so this pp settings would be equal to BlindPP(CPU=4), more or less, isn't it?.
It would be nice we had (for those of us not very accostumed to read this terms: I understand little in man_page) a kind of traslation of mencoder filters into avisynth filters. For instance:
-scale is like resize. And sws=2 is bicubic and sws=9 is lanzcos
-noise is like blockbuster method="noise", and parameters "x" in mencoder have similar strength to "y" in avisynth
-hqdn3d is a spatialtemporal filter like STMedianFilter (?, or whatever)
-unsharp is like asharp...

Don't know if is that easy, I imagine not. But it would be very useful a guide like this, for those of us that learned (me little) filtering with avisynth.

EDIT: before reading your post I started a new 2pass test without trellis and with scplx_mask=0.3 (previous 0.2), and this way I obtained less microblocks, but still present. Don't know if it will help you help me :wink: . Next I'm testing postprocessing...

@digitall.doc
I think bitrate 1500 is too low for anamorphic 704x576 in any encoder. I would use this bitrate only with 4:3 A.R. (using black borders)

Its exact that similar! :)
(but scale ... it seems it applies a YUV to RGB conversion and that makes me very confusing as its told in the FAQ of mencoder settings.)

I do agree with Vmesquita that 1500kbit is Veery low in case of anamorphic material, even if its 2.35:1 where black borders in the source in its anamorphic state are added.
For me (ok Im a quality freak) a vbitrate in average of 2000kbit in such a case is minum, even when using mencoder. But if the movie is very dark in average like Seven or Matrix .... well you could drop the avg bitrate. ;-)

inc, vmesquita,
thanx pals, that's what I was afraid, too low average bitrate. But it's what I need to fit 4 hours and a half with 4 audio streams and 2 subtitles in a media... if possible. For comparison, I'll try CCE and see if to fit the film in 1442 Mb I can use Q lower than 40. If not, it's just impossible.
Anyway, I also encoded at average bitrate 2000 and got those microblocks :( . They were less at all, but still present.
I'm running now a two pass with postprocessing (it drops speed 4 fps, but if it worth it :roll: ).
What about opening a thread on mencoder internal filters and equivalency with avisynth filters?. I think it would be very useful, but I don't have the knowledge to start a thread like this, just general ideas like the ones I posted. I just can help testing whatever you suggest... :lol:

Well DigiDoc ...

In regular I can fit 2 Streams of each 2h incl 4 Audiostream easely on one DVD-R Media incl. Menues etc. at 2000kbit each. :?
Do you just add the sources AC3s to your encodings or you do reencode them at lower Bitrate?? In my case (and I got a Dolby Sourround Digital decoder) ... 256kbit at 5.1 AC3 sounds good! Means the best compromise where on the other hand the quality of the picture rises :)
Cause 448kbit and 256kbit means a big drop in final filesize of your Audio streams.

ok ... in regular I do a prediction on a whole, means BOTH streams merged in an avisynth script so I'll get the same quality for both streams.
Afterwards I choose the resulted avg Bitrate or CQ of both on both encodings and the summary of all will match. Thats the best approach to get constant quality during the whole DVD-R.

BTW: According to your Microblocks ..... is that still your StarWars Episode youre testing?
To check encoders qualities I do always do preview and check/compare my encodings at 200% in Vdub

I keep original audio track in spanish, and reencode english one in AC3 2.0 192 kbps (I think, I don't remember now the bitrate). That's the way I use to lower audio filesizes.
That's already the way I do (I've learnt from best maestros out there :wink: )
Yep, Episode II. But I didn't compare in Vdub (do you load original and encoded streams, or just one at a time?) since microblocks are really noticeable without enlarging, and aren't in the source (I didn't enlarge it).
This afternoon I'll take a look at my 2000 kb average with postprocessing encode and see.

:cry: It doesn't work :cry:
Even after doing the postprocessing, with 2000 kb average bitrate, I keep seeing those microblocks everywhere.
I tried changing the player, in case it was a player issue (never before, but...), but it keeps appearing in PowerDVD, MediaPlayer, and TMPGEnc preview...
Don't know what to do. Bad thing

Have you thought about encoding non-Anamorphic? You'll save LOTS of space... :roll:

Of course, it could be a way, and I thank you for your support.
I thought I could also resize to 528x576. Or use a higher scplx_mask value (now using 0.2). Or add a heavier filter chain.
But if I was getting good results in my KDVD encodings, anamorphic, 704x576, with very little filtering, and with 2 audio tracks and subtitles, using CCE, I want from mencoder the same, with better result if possible.
And I think that's why we're here, trying to take out the best of mencoder. And the best must be at least the same we already had with previous encoders (of course, always keeping in mind this is a free tool...)

BTW, I tried with a little filtering (hqdn3d=2:3:3), but still microblocks, at 2000 average bitrate... am I the only one noticing this effect at this bitrate?.