02-25-2004, 12:56 PM
|
Free Member
|
|
Join Date: Jan 2004
Posts: 341
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
I hope to have time to start testing today on my company's PIII-500 laptop w/192 MB RAM and 2GB free HD space.
Just to fit into your testing standards, what tools do you use besides Bitrate Viewer? VirtualDubMod? Others?
I already downloaded BV and VDMod for install after work. Please tell if if there's something else to install.
@vhelp: I've already downloaded your sample, haven't seen it yet.
My version is PAL.
Bilu
|
Someday, 12:01 PM
|
|
Site Staff / Ad Manager
|
|
Join Date: Dec 2002
Posts: 42
Thanks: ∞
Thanked 42 Times in 42 Posts
|
|
|
02-25-2004, 01:01 PM
|
Free Member
|
|
Join Date: Jan 2003
Posts: 1,009
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
ok, great.
"The Fifth Element" ...
But, just for your info, I ripped the dvd's chapter 6 and 7 for these tests. I
found it worth using in my tests, because of the bitrate spikes issues. I
recall jellygoose working w/ this issues of b-spiking too, and even did the
same scene, and got the same results (w/ respects to b-spiles) I think that
jellygoose is PAL too.
-vhelp
|
02-25-2004, 05:46 PM
|
Free Member
|
|
Join Date: Jul 2003
Location: Valencia (España)
Posts: 741
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Hi friends,
I didn't want to, but at last I installed another tool in my PC: bitrate viewer.
Well bilu, I think you were waiting for the results in that explosion-blocking scene:
- when using vbitrate=vmax=5000, vqsquish=0, vrc=tex, max_b_frames=1, --> max bitrate 4574 (avg Q 2.15)
- with vbitrate=vmax=8000, vqsquish=1, vrc=tex, vqblur=0, max_b=2, --> max bitrate 4340 (avg Q 2.22)
- with vbitrate=vmax=8000, vqsquish=1, vrc=tex, vqblur=0, max_b=2, noise added, --> max bitrate 4408 (avg Q 2.66)
- and finally, I tried vb_qfactor=1 and vb_qoffset=0 (rest of parameters as last one), --> max bitrate 5414 (avg Q 2)
YES bilu, you knew that. After decreasing Q in B frames (through vb_qfactor/offset), average Q decreased and bitrate raised. Filesize grew from 6499 to 7745 kb. And visually got much less blocks, still noticeable on PC screen (hope we'll be able to get rid of them).
What I do next, what I do next, what I do next?.
Taking the mood this encoder .
Mencoder didn't take profit of all the bitrate available (up 8000), maybe can't do that better and already completely compressed the image, but now there are less blocks (but still they are)
@vhelp
the scene I'm refering to is between frames 4854-4876. There's a scene some minutes later, when the jedis are following the person trying to kill queen Midala, that I believe is still more bitrate demanding. Maybe I'll ready tests with this new sample, to see if mencoder can raise bitrate to limits available, without going over.
Well, don't want to make it too long, but we have new parameters to play with.
@ all,
also PAL land here
|
02-25-2004, 06:00 PM
|
Free Member
|
|
Join Date: Apr 2003
Location: Chinese Democracy starts now!
Posts: 2,563
Thanks: 1
Thanked 0 Times in 0 Posts
|
|
Hi all,
Since J3lly is in Germany he is definitly in PAL Land as well as myself.
Sure would like to have the same movie you guys are trying.
That way we could do more in less time in terms of testing.
Have we given up on finding a good clip in the INet for testings?
That sure would be great.
Cheers
__________________
Rui
|
02-25-2004, 06:08 PM
|
Free Member
|
|
Join Date: Jun 2002
Location: Germany
Posts: 1,288
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Talking about me behind my back huh :P
Yep, PAL sources only here. Well I'd just propose everybody should get that Fifth Element DVD, and encode exactly that explosion scene. It's perfect to test these spikes. And it's a great movie anyway so you won't regret you got that on DVD.
__________________
j3llyG0053
|
02-25-2004, 06:18 PM
|
Free Member
|
|
Join Date: Jan 2004
Posts: 341
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
vhelp's scene is from 5th Element, digitall.doc's scene is from Star Wars.
Just to avoid filling this thread with unrelated stuff, could someone give a hand here?
http://www.kvcd.net/forum/viewtopic.php?t=9260
thanks,
Bilu
|
02-25-2004, 07:50 PM
|
Free Member
|
|
Join Date: Jan 2004
Posts: 341
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by digitall.doc
What I do next, what I do next, what I do next?.
|
Hmmm... you already have tried the default matrix, which is h.263.
Quote:
Then I remembered that nocht matrix is supposed to cut frequencies and lower bitrate, so I tried using default
matrix instead of the Notch matrix. File size grew, but no image improvement.
|
Why not try the other one?
Quote:
mpeg_quant -> use MPEG quantizers instead of H.263.
(default:disabled) (i.e. use H.263 quantizers)
|
NOTE: No garantees of improvement, out of ideas for now
Bilu
|
02-26-2004, 04:20 AM
|
Free Member
|
|
Join Date: Jul 2003
Location: Valencia (España)
Posts: 741
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Well bilu,
I already changed Notch matrix with standard matrix. I think I posted it, when I tried to improve image in my tests and raise bitrate. In fact, my last tests changing B frame quantizer where done with standard matrix.
Related to mpeg quantizers, I tried to use it. I just included in the command-line the function mpeg_quant, but mencoder gave error and didn't encode. I tried then giving it a value (mpeg_quant=0 or 1) and didn't work also, so I dropped it. How can I activate mpeg_quant.
|
02-26-2004, 04:40 AM
|
Free Member
|
|
Join Date: Jan 2004
Posts: 341
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by digitall.doc
How can I activate mpeg_quant.
|
Maybe you can't and this parameter is only used for MPEG-4 encodes
Bilu
|
02-26-2004, 09:13 AM
|
Free Member
|
|
Join Date: Jan 2003
Posts: 1,009
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
I tried this under w98gold, and it just crashes mencoder
If I put a 0 (or no) as param value, it works (cause 0 or no, mens disabled)
..which tells me that we are using this command param correctly.
Also, on last line of my error, it same:
Quote:
[mpeg2video @ 01427200]mpeg2 style quantization not supporetd by codec
Could not open codec.
FATAL: Cannot initialize video driver.
|
-vhelp
|
02-26-2004, 11:17 AM
|
Free Member
|
|
Join Date: Jul 2003
Location: Valencia (España)
Posts: 741
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Yes, vhelp, that was the error I got when tried to employ mpeg quantizers. Don't know if we're doing something wrong, because under ffvfw you can activate mpeg quantizers, so the codec can work with them.
|
02-26-2004, 12:16 PM
|
Free Member
|
|
Join Date: Jan 2004
Posts: 341
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Quote:
Originally Posted by digitall.doc
Yes, vhelp, that was the error I got when tried to employ mpeg quantizers. Don't know if we're doing something wrong, because under ffvfw you can activate mpeg quantizers, so the codec can work with them.
|
Or ignore them
I've seen no reference to H.263 quants in FFMPEG docs, just in Mencoder docs.
Bilu
|
02-27-2004, 05:33 PM
|
Free Member
|
|
Join Date: Jul 2003
Location: Valencia (España)
Posts: 741
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Hi ,
as it seems that this became our "official" mencoder thread, I'll post here my last experiences:
you know from my previous posts I've been testing an scene in StarWars II where I was getting blocks...
I've been trying 2pass encoding... well, first of all it doubles encoding time, but maybe it worth.
Command-line:
Code:
mencoder -of mpeg -ovc lavc -nosound -lavcopts vcodec=mpeg2video:vpass=1:
mbd=2:vrc_minrate=300:vrc_maxrate=8000:vqscale=2:vrc_buf_size=1835:vmax_b_frames=2:
keyint=15:aspect=16/9 D:\Temp\1.avi -o D:\Temp\encoded1.mpg
mencoder -of mpeg -ovc lavc -nosound -lavcopts vcodec=mpeg2video:vpass=2:mbd=2:vrc_minrate=300:vrc_maxrate=8000:vbitrate=8000:vrc_buf_size=1835:
vmax_b_frames=2:vqblur=0:keyint=15:aspect=16/9 D:\Temp\1.avi -o D:\Temp\encoded2.mpg
As you can see I tried to make it simple, removing quality functions.
I get less bitrate peak in this scene (about 4700, in 1 pass with other settings got 5400), with Q about 2 (same as 1 pass) but visually almost the same blocks (or even less ). And later comes a more bitrate-demanding scene, that with ffvfw got bitrate above 10000, and with mencoder 2pass I get bitrate 7700.
To sum up: very satisfied.
How can we speed up the proccess (if possible) or are we to stay in 1 pass and improve it?.
... suggestions
|
02-27-2004, 05:42 PM
|
Free Member
|
|
Join Date: Apr 2003
Location: Chinese Democracy starts now!
Posts: 2,563
Thanks: 1
Thanked 0 Times in 0 Posts
|
|
Hi digitall.doc
I am very impressed with this encoder's quality under 1 pass.
I sure will try 2-pass but not right now.
I think it would be better to understand a bit more about mencoder on single pass and then tweak arguments to shoot at 2-pass.
Don't know if others agree with me or not.
Everybody lease post your oppinions on this issue.
Cheers
EDIT: BTW you're using GOP=15. Encoding at (720/704)x576?
__________________
Rui
|
02-27-2004, 07:40 PM
|
Free Member
|
|
Join Date: Jan 2003
Posts: 1,009
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Hi guys.
I'm also w/ rds. 1 pass should be all. But, I'm not pushing 2p over the edge.
You guys make the decision - - either to encorporate or not to inc.
I prefer to stick to 1p just because its less steps. Maybe is quality is really
your main concirn, then by all means, push it to the limit, if you can. But,
then, I gotta think about TMPG for this maybe. No, not ME using TMPG under
2p. I would never cross that boarder - thus far.
OT 1: I've ben wondering just how many other video sources can we
open (or is limited under MEncoder) cause, so far, from my understanding,
we only have the ability to either:
* open from an true .avi source (no frameserving or vdub/avisynth)
* use makeAVIS (either by using .reg hack or XP)
So far, W98 can not open/process makeAVIS sources. It can only open true
.AVI sources. I've spent approx a week trying to figure this one out, and
then in the middle of this week, I really got into it, because for me, my sources
are always from frameserved references (ie, vdub/avisynth) and so far, I
can't open these, which pretty much makes my work in mencoder useless,
unless I don't mind double-duty work of re-writing a filtered new .AVI source,
which I'm not about to start doing
Currently, I'm pretty disapointed at this limitation w/ mencoder not being
able to accept any form of frameservers (ie, vdub/avisynth) under W98
I can't see anybody actually wanting to continue testing mencoder, if they
have to rewrite true .AVI source files. I have two or three so far, and I'm
not about to make any more. Its just too inensive a job to have to do, let
alog having to do a whole movie, every day This is not what I'm going
to be doing.
OT 2:
On the other hand, through my trial and tribulations w/ mencoder's lack of
frameserving abilities ( W98, others ?? ) I've ben able to figure out how to
get mencoder to open my DV files.. and they actually encoded.. quite well
I must say. These DV source files were my ADVC-100 captures of my VHS
SW WS SE that I still have on my hd. I was pleased that the quality was
pretty good from vhs sources but still, even if I were able to open up
these DV files in mencoder to encode, at this given juncture, you can't do
any filtering (from frameservers) and I have plenty of good filtering processes
and methods that I use lots, under vdub
Well, that's ben my long and busy day's jeorney thus far
@ rds,
looks like mencoder for W98 is proving hart-ship for me (and others here,
I'll assume) And, if things don't get better, it looks like I might be heading
for a new direction (or hobby to toy with elsewheres)
-vhelp
|
02-27-2004, 07:52 PM
|
Free Member
|
|
Join Date: Jul 2003
Location: Valencia (España)
Posts: 741
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Hi,
I wasn't really proposing you to move to 2pass: it means doubling the encoding time. Just posting my experience: it works, and works well. But I'm focused in improving 1pass. With your help. I've posted several settings for 1pass I don't know if you tried, and how can be improved.
@Rui
yep, encoding at 704x576
|
02-27-2004, 08:21 PM
|
Free Member
|
|
Join Date: Apr 2003
Location: Chinese Democracy starts now!
Posts: 2,563
Thanks: 1
Thanked 0 Times in 0 Posts
|
|
@vhelp and digitall.doc,
Hi guys,
I'm very concerned right now in finding out if vhelp is the only one having problems
under Win98 or if it's an OS specific problem.
I can't find anyone else with Win98 or WinME willing to give it a shot, right now.
Unfortunatly, at least here in Portugal, Win98 gave place to WinXP pretty fast.
So, the few guys I know with Win98 have no time right now to assist me on some trials...
If mencoder has an issue with Win98 and WinME it's an enormous step back and I'll start aiming
more at ffvfw instead because with the latest vbv buffer developments by Kwag it could be 100% working
in the next...few hours?
Ok, maybe days or weeks, but that would be in a short period and that wouldn't mean we have
to stop testing it because we'll be able to patch every movie we've been encoding since the begining.
So, right now, I'm working on a small guide for mencoder to see if we can catch some more guys attention.
Hopefully some of them are 98 or ME and we'll have more feedback besides vhelp's.
I hope I have it finished by tomorrow, but since I have so many "real life" things to do during
the weekend I really can't promise anything.
Keep up with the testing spirit and maybe we'll find a way to it under every Win32 OS.
I wonder if Jorel has tested it yet .
I remember he has a lot of Windows "images" created with Ghost or something like that.
I bet he could try it. Maybe he'll even like it if it works
So 1-pass or 2-pass right now is maybe not a so big issue.
But for those with Win2000 or WinXP I'm not saying they should stop testing it
I'll "see" you soon.
Cheers
__________________
Rui
|
02-28-2004, 05:03 AM
|
Free Member
|
|
Join Date: Jul 2003
Location: Valencia (España)
Posts: 741
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Well, Rui, I begun testing ffvfw, and if I focused more on mencoder was due to its speed (faster than ffvfw), and that bilu (hi Bruno, how are your testings) managed to explain functions in a way I understood better mencoder than ffvfw and knew better what I was doing (and this is essential for me, to employ any encoder). I won't leave testings with ffvfw, but I'll keep mainly in mencoder (even I don't know right know how to improve it, need all your help).
And my main problem with ffvfw (apart from encoding speed) was with bitrate (not with vbv that we solved adjusting bitrate slide earlier. Now, with poerschr way of encoding, you're right we need to change vbv, but I guess is a question of time managing to ffvfw doing this without having to mess the m2v file) too high/too low. As I still didn't try poerschr (or was marcellus: sorry if I'm wrong) way to tweak ffvfw, I don't know if the bitrate issue is solved. But I still find we don't really know how to employ ffvfw (maybe Milan's help...)
So, here I'm back again to mencoder
|
02-28-2004, 08:21 AM
|
Free Member
|
|
Join Date: Apr 2003
Location: Chinese Democracy starts now!
Posts: 2,563
Thanks: 1
Thanked 0 Times in 0 Posts
|
|
Hi digitall.doc,
I also think that mencoder could really be the way to go.
It's got quality, it's got speed, it adapts to all our needs of encoding (KVCD and KDVD)
although it's a lot better for higher bitrates.
Most of all, it's still under development by a team.
This could come to be a very well succeeded encoder in the near future if one of the guys can
do a good package with it similar to dvd2svcd.
That would put mencoder in a very strong/top position.
But we must remember that the mentioned developer team is aiming almost 100% to Linux.
And we know how these guys forget about win32 guys like us. Not that they mean to, btw...
So I'll do the same as you, stick more with mencoder with an eye at ffvfw, if I can find
a way to keep vhelp and all the win98 users in the team.
I must remind you all that, so far, I haven't seen any feedback from win98 or winme users,
that are testing mencoder.
This is somehow very important for us so if some of you could post us your experiences...
Cheers every1
__________________
Rui
|
02-28-2004, 09:23 AM
|
Free Member
|
|
Join Date: Jan 2003
Posts: 1,009
Thanks: 0
Thanked 0 Times in 0 Posts
|
|
Morning everyone
I've got some gusto power under my belt this early morning
@ rds,
I have some new ideas that I'd like to throw at this MEncoder problem some
of us (if not ALL w98/se/me) users' are experiencing. My guess or huntch
is that these users have given up on MC since it's very rebelious in its features.
Well, that's just another one of my assumptions. Anyways..
Here's what I am proposing that we (w/ yours and others help here) do.
I have a press pc sitting behind me that is pressing upon me to give him a
try. He keeps telling me that he is not tainted w/ drivers and various versions
of mencoder/ffdshow/ffvfw. In fact, he tells me he does not have any of them
.
.
I'd like to take this 2nd pc (equipt w/ W98Gold) and install ONLY THE
NECESSARY MEncoder SOFTARE APPS/DRIVERS/CONFS etc. on it and
run some testings on it to see if it will encode my files from makeAVIS.
.
.
I need to be sure that I have ONLY the necessary software/apps for this pc.
It's an untouched pc.
.
.
I have a theory that maybe perhaps my other pc that's given me the trouble
is far too tainted w/ drivers and colorspace issues and so on, that the cause
for makeAVIS not working could be that very reason. But, I need to rule
this out by first trying on an untouched pc.
If we can do this, then I'll see about imaging it now and we (you) can
gather a listing of files (latest versions) for me to install on it. Should anything
go wrong, I can de-image it to it's fresh state I left
I've ben toying w/ the codec.config file and have a better understanding of
it's params now. So, I'm ready for it too
I also belieave that the codec.conf is the key (or partial) to getting
makeAVIS to work on w98. That's why I have ben messing with it since
yesterday, ALL DAY. Heck, I even developed a front-end tool to just it
alone, to help me find the right setting.
But, I still think we should try and figure out a way to get vdub to work with
mencoder to. And, also avisynth scripts to (w/out the need for makeAVIS)
But, as I was saying, codec.config is key because its the "driver" for MEncoder
to open .avi sources and, having the right driver on your pc and no issues
w/ colorspace and whatnot, might be key also but only add'l testing will
tell.
But, I agree with you.. mencoder shows great promise. All it really needs
is a good pair of paints/shirt/tie (GUI) which I'm still developing, (others
can do the same, as they wish, too) and sky is the limit
The 2nd PC to the test.. .. ..
@ all - - Please, ..do get me a list preapred for work
Well, let me start by getting myself a fresh cup of coffee, and begin pulling
more of my hair out.
Chereo all,
-vhelp
|
All times are GMT -5. The time now is 07:06 PM — vBulletin © Jelsoft Enterprises Ltd
|