WM9 for DVD-Rip [Archive]

Ramirez

16th April 2003, 02:03

Damn it, as usual you're always one step ahead me bro,:) anyway I'm planning quite a big comparison between WMV9 vs. RV9 too, its gonna be BLADEII, I'll try to compare about 20 frames at 600kbps.No filtering for RV9 because RV9 kinda hates filtering plus the source is very clean, for WMV9 I'll go for CPU=4, this baby has its own PP just way less aggressive then Real's.

P.S
Karl knows about that problem with walls, so I'm thinking perhaps it’s a good idea to delay it for a while perhaps they'll release a fix for that problem soon, that way this comparison will be fairer.

P.P.S
I'm not sure about XVID,you see XVID funs kinda hates critics,and if the results will not be more then satisfactory...(You've seen that right?);)

Sirber

16th April 2003, 02:30

Originally posted by Assault
I think you shouldn't use the same filters for the encodes but the encoding speed should be nearly the same. For example you could use more filters / slower filters for the XviD encode than for the WMV9 encode because XviD is faster.´

Regards
Assault

I don't care about speed, I want results. I want to compare apples with apples, so I use raw codecs, which meen no filters. At 300kbps, WM9 is not good. RV9 is not too bad (The original MPEG is worse).

I'll cut it and post it maybe JEUDI (What's it in engligh? I don't remember)

@Ramirez

Please don't use filters, or use the same. Like I would use DeComb as IVTC only for both. I can't wait to see your comparison :)

iwod

16th April 2003, 12:36

i only encode Anime..... so may be different to the test you want. I haven't done any very serious testing yet. But below are the results i have so far.

(Notes: I don't have the figure with me right now i will post it tomorrow or few days later )

WMV encode a slower than RV9

WMV uses higher resources to play.

WMV provide similar results @ low bitrate ( 350Kbps VBR ) But the result files size are larger than RV9. ~20%

WMV scale better than RV9 on higher bitrate. @ >600- 700 Kbps. ( I am still testing on this one )

Does any body know if WMV has built in post processing like RV9?

And WMV advantages are starting to shown up. Such as WMV in AVI, and porting WM to linux platform.

Assault

16th April 2003, 13:41

Originally posted by Sirber
I'll cut it and post it maybe JEUDI (What's it in engligh? I don't remember)

Thursday ;)

Sirber

16th April 2003, 14:16

Originally posted by iwod
And WMV advantages are starting to shown up. Such as WMV in AVI, and porting WM to linux platform.

Real is already on Linux, and Microsoft won't port it on linux.

Tommy Carrot

17th April 2003, 01:07

All right, we all know RV9 and WM9 is better at low bitrates than xvid. But what about higher quality/bitrates (>700 kbit)?

In my tests both real and WM was more blurrier, with less details than xvid.

This is why they are useless for many purposes. The streaming bitrates has no appeal for me. :D

Sirber

17th April 2003, 03:42

My goal in life isn't to get a girlfriend, or a car, but to fit DVDs on 1 700 MB CD. :D That's why I use RealVideo. I have to change Cds when watching a movie :mad:

Tommy Carrot

17th April 2003, 05:32

Originally posted by Sirber
My goal in life isn't to get a girlfriend, or a car, but to fit DVDs on 1 700 MB CD. :D That's why I use RealVideo. I have to change Cds when watching a movie :mad:

Yes, you're right (on the CD-switching issue:D), but i store the movies on my hdd, or burn 'em onto DVD, and i want PERFECT quality.

int 21h

17th April 2003, 08:49

Originally posted by Sirber
Real is already on Linux, and Microsoft won't port it on linux.

http://www.vnunet.com/News/1140128

CaptainCarrot

17th April 2003, 09:41

@int 21h:
It IS correct that MS is porting the wm9-codec to linux, but only to be used in consumerhardware based on linux, since there are quite some players running linux and MS wants wm9 to be the HDTV-Standard (i think 2 wm9-encoded HDTV-movies have already been published). So you probably will have to wait until someone hacks one of those hardware-players to get wm9 on desktop-linux.

ookzDVD

17th April 2003, 11:40

Originally posted by Sirber
Hi

Can someone do a test with RM9, WM9 and XviD at the same bitrate and the same filters? That would be great. If someone is interrested please share with us.

I think Doom9 is working on it.
I just can't wait to see the conclusion.

iwod

17th April 2003, 12:03

Originally posted by Tommy Carrot
All right, we all know RV9 and WM9 is better at low bitrates than xvid. But what about higher quality/bitrates (>700 kbit)?

In my tests both real and WM was more blurrier, with less details than xvid.

This is why they are useless for many purposes. The streaming bitrates has no appeal for me. :D

i will say @ >1000 Kbps xvid or divx is till better than WMV.
I haven't test the WMV pro which allow HD, coz my computer can't play it smoothly.... >_<

I am not sure if RV9 was tuned for high bitrate. As i have stated b4 that RV9 was for Streaming, so @ >700 their quality are being outplace by mp4 rivals such as Xvid and Divx. This is simliar to why RV9 is better tha xvid and Divx becoz it is tuned for low bitrate. And that is the reason why RV9 doesn't scale so well @ high bitrate.

Ramirez

17th April 2003, 13:20

Originally posted by iwod

I am not sure if RV9 was tuned for high bitrate. As i have stated b4 that RV9 was for Streaming, so @ >700 their quality are being outplace by mp4 rivals such as Xvid and Divx. This is simliar to why RV9 is better tha xvid and Divx becoz it is tuned for low bitrate. And that is the reason why RV9 doesn't scale so well @ high bitrate.
Well,it's isn't quite true, have you tried latest build of producer? Try it and you'll be surprised by the quality of the resulted output, I've played allot with this build recently and came across the following, at 900kbps and up almost no blurring/smoothing occurs
There is some amount of /blurring/smoothing + RV9 kinda hates walls, but in general the quality is awesome! Last time I've tried it at 980kbps/720x384 and the resulted rip came out amazingly detailed and sharp.

Sirber

18th April 2003, 14:44

midiguy

21st April 2003, 03:21

meh, IMO, you guys use RV9 at too high a resolution.. I use the same res for RV9 as I would for DivX or XviD (maybe one notch higher), but when you go 640 x 480 @ 600 kbps or something like that, you are sure to get a blurry result (atleast for most sources...)

Sirber

21st April 2003, 04:00

I made Stephen King's It, a 3h07 movie, in 640x. The result is so great for a movie that long. Almost no bluriness, at 450kbps for video and 64kbps for sound. The source was noisy so I used Convolution3D at light. It tooks me (or It mouhahahaha) 30 hours to encode it (It again! Want a baloon?).

It's my best encode right now...

For ~600kbps:

I have Monty Python and the Holy Grail, 650kbps with 2x96kbps sound track_. It's sharp and nice, IMO better than the DVD because there is no noise, but noise and RV9 makes sometimes some filcktering... :(

But in generals all my best rips are in RV9 (3h07), in second XviD for 2h20 movies.

feliz

21st April 2003, 06:54

My vote is for wm9 for dvd rips at >2000 kbs. I have less artifacts such as ringing.

Sirber

21st April 2003, 15:20

Is more a rip about 500 - 1000kbps?

feliz

23rd April 2003, 03:19

sorry, I like to rip my dvd's with ac3 so I get about 2 rips per dvd+r which ends up around 2000 kps. I figure might as well keep the quality up if I'm saving the ac3 in avi. I like the quality with wmv with ac3 better.

ooops!

29th April 2003, 16:39

I've been experimenting with WM9 for some time and have ripped about 20 movies so far into this format. My goal with all the rips was to produce files that were no larger than 700MB.

I find WM9 is much more accurate when it comes to working out the finished file size, than say DivX.

When carrying out 'video' only tests, I have found that at 800kbps - 2000kbps WM9 and DivX are fairly evenly matched in terms of quality. Above 2000kbps DivX is better than WM9. Below 800kbps DivX begins to drops away and at 600kbps a WM9 2pass is much better.

In one of my tests I have managed to encode the whole of Star Wars 2 (137mins PAL) to a file of just 680MB. Using a 2pass video setting of 615kbps and a 2pass audio setting of 64kbps.

In my opinion, at low bit rates WM9 out performs DivX..... But maybe RM9 is a better option below 600kbps. I am yet to try.

I've never really liked Real Media stuff. As it always wants to hi-jack the settings in your PC and connect you to a load of it's crap on it's 'web site' - or should that be 'web shyte'!

The debate continues...............

Sirber

29th April 2003, 17:59

I did a good rip at 450kbps (video only) with RV9. It took me 30 hours to produce it.

At equal bitrate, like 600kbps, I want to know which codec give the best quality.

ooops!

29th April 2003, 20:21

Well I can safely say (in my opinion). When encoding at low bitrates, DivX is out. Even the new 5.0.5 version using multi pass is not as good as WM9 VBR.

As I write this I am encoding Titanic (187mins) using WM9 at 444kbps video / 64kbps audio. The first pass has taken 4hours and 5mins to complete, the second pass will take three times that. And I'm using a 2.8Ghz P4 PC with 1024MB of DDR RAM!

Encoding DivX is a hell of alot faster than WM9. Performing a manual 2pass encode with DivX at 444kbps took about 6hours. And looked crap. So I am hoping WM9 will look better!

After I've finished the WM9 encode I might give RV9 a go at the same bitrate.

Bye the way. All my finished encodes are viewed on a 42" plasma screen at work. Which really shows the imperfections and means that more than one person is able to pass comment on the finished result!

Sirber

29th April 2003, 20:30

@ooops!

Good! That's what I was waiting for :) Could you give us some screenshots after?

The longest I made was Stephen King's It, 3h07, at 450kbps + 64kbps in RealVideo 9. The quality is nice and a little blury in background. It tooks 30 hours on my 2000+ with 512 DDR@266 (I'm at 333 now). I made a screen:

http://www.webernic.com/sites/sirber/It.jpg
<img src="http://www.webernic.com/sites/sirber/It.jpg">

Here's my filters:

Decomb
Convolution3D at light
Bicubic @ 640x (Or lanzcos, I don't remember)

ooops!

29th April 2003, 23:25

The screen shot looks qood.

But even though I live way over in England. I'll post you some video samples on CD if you like.

Just send me a private email with your address and I'll get right on it!

Cheers

Sirber

29th April 2003, 23:30

Hum...

I will be ok with JPEGs :) You don't have to mail me CDs :D

[edit]

BTW, welcome to the board :)

Atamido

30th April 2003, 02:42

Avoid JPEGS, and rather use PNG's instead. You have the possibility of compression artifacts in JPEGS that could interfere with a comparison between pictures. PNG's on the other hand are lossless, so what you see is what was captured.

Sirber

30th April 2003, 03:36

Yeah but PNG are bigs :(

JPEG at 90% quality are not too bad...

Kb_cruncher

30th April 2003, 04:51

I will do a comparison test with rv9 and wmv9 @500kbits soon.I have only done one rv9 rip so far,Goldmember(90min)at 1200kbits.The results where lightyears ahead of divx or dare i say it XviD:D at the same bitrate.
At the moment i'm testing a new svcd creation proggy http://forum.doom9.org/showthread.php?s=&threadid=51081.Its very good and apparantly "procoder"gives better results than cce,but is is soooo slow(5fps on a 2100+@2.3gig and 512Mb kingston hyperx pc 3000)but it sounds like wmv9 is even slower:eek:

Atamido

30th April 2003, 05:50

Originally posted by Sirber
JPEG at 90% quality are not too bad...
They are not bad, but if you are comparing quality, they are not good. You want an image that shows you exactly how the frame was decoded, and JPEG won't give that to you. Also, PNG's can be big, but aren't always. And in some circumstances, I've had them be smaller than JPEG's. But, with movie images, they are most likely going to be bigger than JPEG, but much smaller than BMP.

iwod

30th April 2003, 11:49

Originally posted by Ramirez
Well,it's isn't quite true, have you tried latest build of producer? Try it and you'll be surprised by the quality of the resulted output, I've played allot with this build recently and came across the following, at 900kbps and up almost no blurring/smoothing occurs
There is some amount of /blurring/smoothing + RV9 kinda hates walls, but in general the quality is awesome! Last time I've tried it at 980kbps/720x384 and the resulted rip came out amazingly detailed and sharp.

You mean M2 of Helix producer. That is the only build i haven't try. I am waiting for AutoRV9 b4 i get a test on it.
I am doing ONLY anime rips . Therefore results may differ to your results.
ON Anime, 900/1000 Kbps is really the highest to go b4 xvid and divx. Any higher divx will produce more DVD like encode.
However, you mention no blurring and smothing occurs. So far i have never met that on RV9. Since the highest bitrate i have try is 900 Kbps. So i may try even higher to see what happens...... ( Actually i have try 1200 but i see no different to 900 ones... )

I said it is a streaming format because it is lacking for some other contanier format has have. Chaptering......

Hopefully it would be possible later to put RV9 in MCF.

ooops!

30th April 2003, 12:32

Yep, unfortunately you can't really tell how good a video encode is by providing a Jpeg still. This is why I always offer to send a CD.

My test encode of Titanic using WM9 at 444kbps video / 64kbps audio (I know, I know, but it's the longest movie I've got at 187mins PAL!) looked poor. So WM9 at this setting is not the way to go at present.

I installed the 'RealOne' basic player and 'Helix Producer Basic 9' this morning and when I tried to import a 48MB Mpeg2 (video only) test file. All it did was play the file and crashed the 'Helix' software.

As I mentioned, I've not been a big fan of Real products in the past, but alot of people are saying good things about their new codec. So it's only fair I should give it a go!

I don't know my way around the software yet. So any hints would be great. Every time I tried to change a setting (like the frame rate) it said I had to upgrade!

It would seem that most of you guy's are all living in the world of NTSC (720x480 @ 30FPS). I'm in England where we use PAL (720x576 @ 25FPS).

kilg0r3

30th April 2003, 12:42

@Ooops

Try one of the frontends AutoRV9 or x2real. No need to learn the commandlines.

Sirber

30th April 2003, 14:25

@Ooops

AutoRV9 is definitly the best to encode into RV9.

@iwod

I encode y animes (640x480) at 350kbps, including a 32kbps VOICE sound track. Quality is not so bad, it depends on the source.

Vampire Princess Miyu: 10/10
Evangelion: 9/10
Gundam Wing: 9/10
Hyper Police: 7/10

For my LQ (320x240), I encode at 300kbps, including a 32kbps VOICE sound track.

Futurama: 10/10

Don't forget to set Startup Latency at 25 sec.

ooops!

30th April 2003, 22:22

Thanks you guys, I'll hunt 'AutoRV9' down.

I did manage to alter the settings 'Helix Producer Basic 9' by copying and changing the profile of one of the standard presets in the audiences folder.

Since then however, I tracked down a hookie copy of 'Helix Producer Plus' and a KeyGen! But it does not like an Mpeg2 (vob) source file.

When I test a new video codec, I only encode the video. And always use the same 'source' video file (which happens to be Star Wars 2 Chap42 at the moment). I always set the codecs output to the same frame rate (25fps) and the same pixel size (720x576), as the source file.

I do this so each new codec is treated exactly the same. I never encode the audio (when testing) because this can throw in too many variables with the finished file size. After all, we are trying to see how good the video is at a proposed target file size.

When using this method, you quickly find that every manufacturers codec, produces a totally different file size, despite being set to the same target bit rate (615kbps 2pass VBR on this occasion). So in order to obtain a proper comparison I take the value of the largest file size produced (which is always the WM9 2passVBR file) and raise the bitrates of the other codecs slightly, so I end up with same output size for each codec tested.

Still, with me………… or have you all fallen asleep by now?

So onto the results. All of which has been viewed and scrutinised by 6 independent people after having been displayed on a 42” 16:9 plasma display.

After all this testing, the…………

RealVideo9 codec is very good. It encodes the fast moving elements of the video very well. However we all felt, it didn’t encode the background elements quite as well. Which gave the finished video a kind of washed out look!

DivX (v5.0.5) stood up very well. It too encoded the fast moving elements very well but and encoded the background elements better than the RealVideo9 codec. But the overall image appeared to be more detailed but more pixeled.

WinMedia9…………….. It’s not perfect yet, but everybody preferred it!

You’ve got to remember that when you encode video at low bitrates using a 720x576 (or 480) pixel size, you are really pushing the codec to its limits. I suspect good results could still be obtained if output pixel size was reduced to 352x288 PAL or 320x240 NTSC i.e. half the source definition.

I'll just stand by now for the shit to hit the fan....... ha!

kilg0r3

30th April 2003, 23:10

@Ooops

1.
What do you use for encoding with WM9?

2.
For RV9. Try the latest AutoRV9. It supports YV12 input => no color conversion. It also contains the latest milestone release of producer.
Get it
here (http://www.mynetcologne.de/~nc-allgeife8/downloads/autorv9-avs25-repack-v2.rar)

3.
And, I'd really like to see how xvid would score in your test. If you like, try the latest koepi build.

with these settings

Advanced options/General:
-------------------------
chroma motion yes
lumimasking yes
qpel no
gmc no
quantization type: mpeg
vhq 1
max B-frames 2
BF Quantratio 150
BF QuantOffset 100
BF Threshold 255

leave everything else at default

Decoder Options
---------------
luma deblocking yes
chroma deblocking yes

if you get many blocks try increasing the bframe count by 1 or even 2

have fun!

ooops!

1st May 2003, 00:36

Thanks kilg0r3 for your reply. Message recieved and understood!

By all accounts 'Helix Producer Plus' is not the best RV9 encoder out there. So I'll compile some test encodes using AutoRV9.

Sirber

1st May 2003, 01:11

Helix Producer uses producer. AutoRV9 too. It's just that AutoRV9 is optimized for DVD Rips :)

I use Helix for animes (MPEG/AVI --> RMVB) and AutoRV9 for DVD (D2V --> RMVB)

ooops!

1st May 2003, 17:26

Yes, after I installed AutoRV9 I soon found out that it uses 'Helix Producer'. So is it safe to assume that all the Real Video encoders out there use the same encoder engine?

Can I also take it that the only way you can get Helix to input an Mpeg2 video stream is if the read/see a .d2v file extension?

I'm not sure of this, but I from what I remember most of the applications that produce .d2v files, such as DVD2AVI 1.76, 1.77 etc utilize a VFAPI reader.

In my experience when using video applications that use a VFAPI reader, they sometimes have a nasty habit of affecting the performance of other video applications and players you may have installed on your PC. Its effects aren’t immediately noticeable but when you are tying to encode at low bit rates, it can be!

I have to confess that I had to convert the Mpeg2 16:9 (720x576) data stream to an Mpeg1 16:9 (720x576 high resolution 9800kbps VBR) data stream so that Helix could read it. If you have some spare time on your hands, I’d recommend you having a go at converting an Mpeg2 file to a HR Mpeg1 file. There really is no perceivable loss in quality between the two. We use HR Mpeg1 most of the time now at work, when we create Macromedia interactive CD’s for clients.

Kilg0r3 asks what WM9 encoder I use… Well it’s just Microsoft’s standard encoder. There really is no quick way of encoding WM9. Even if you’ve got Ulead MSPro 7 and/or Adobe Premier 6.5 (with the Microsoft beta plugin) as we have at work.

Which version of the koepi XviD codec do you use? The latest version is 04102002-1.

iwod

1st May 2003, 17:39

Yes, after I installed AutoRV9 I soon found out that it uses 'Helix Producer'. So is it safe to assume that all the Real Video encoders out there use the same encoder engine?

Well....yes... they all uses the SAME encoder engine. But you can Choose different version of the SAME encoder engine. :D

The defualt encoder engine inside HelixProducer Basic is 9.01 ( if i remember correctly ) And the lastest avalible on Helixcommunity is 9.2 milestone 2 ( i need to ask something to karl about this lastest version.... )

So in Auto RV9 you could choose to use different version of encoding engine. ( I think there are a few things to be edited before you can use AutoRV9 with 9.2 just search around the forum. )

Can I also take it that the only way you can get Helix to input an Mpeg2 video stream is if the read/see a .d2v file extension?
What's wrong with that?? As far as i know, yes... do it with autorv9 or x2real.I have to confess that I had to convert the Mpeg2 16:9 (720x576) data stream to an Mpeg1 16:9 (720x576 high resolution 9800kbps VBR) data stream so that Helix could read it. If you have some spare time on your hands, I’d recommend you having a go at converting an Mpeg2 file to a HR Mpeg1 file. There really is no perceivable loss in quality between the two. We use HR Mpeg1 most of the time now at work, when we create Macromedia interactive CD’s for clients.

If you don't mind quality.... since transcoding always degrade quality in some way.

Assault

1st May 2003, 18:47

@ ooops!

Which version of the koepi XviD codec do you use? The latest version is 04102002-1.

You should use Koepi's latest unstable version XviD-05042003-1. I agree with Kilg0r3 on the settings except that I wouldn't use lumimasking because I often saw it producing strange artefacts.

Regards
Assault

kilg0r3

1st May 2003, 19:33

@ooops!

The Autorv9 version on my site includes the latest milestone release 9.2 and is a modified/'hacked' version done by ramirez. I have not tested it much, but it seems to work fine.

@Assault
The lumimasking code in koepi's build referencedivx's, which is better than the standard code. It s also slower of course.

Sirber

1st May 2003, 19:40

What's the quality of WM9 for DVD-rips?

ooops!

1st May 2003, 21:38

For the purposes of this test, my 'Video Only' WM9 encode was set to 615kbps 2pass VBR.

However, if you set the 'Audio' to 64kbps as standard. With a 'Video' setting of.....
615Kbps will encode at 5.00MB per min (or 136-140mins on a 700MB CD)
705Kbps will encode at 5.66MB per min (or 120-123mins on a 700MB CD)
955Kbps will encode at 7.55MB per min (or 90-94 mins on a 700MB CD)
1475kbps will encode at 11.3MB per min (or 60-62 mins on a 700MB CD)

The above is assuming you are encoding using a pixel size of 720x576 at 25fps. Works for me first time, every time. But takes a bloody long time!

Assault

1st May 2003, 22:40

@ kilg0r3

The lumimasking code in koepi's build referencedivx's, which is better than the standard code. It s also slower of course.

Yes, I know that. ;) But it didn't convince me either. I made some tests with Vanilly Sky some time ago and it didn't improve neither quality nor compressibility (I know that Koepi mentioned that refdivx's lumi masking is something like a bitrate redistributor). Furthermore I think ooops! shouldn't use a feature like lumi masking for his codec comparison which doesn't improve quality much but could produce artefacts.

Regards
Assault

kilg0r3

1st May 2003, 22:49

ok :)

ooops!

1st May 2003, 22:51

Wow guys don't start falling out........ I'm off to the Pub. See you!

Assault

1st May 2003, 23:07

This wasn't meant as an offence. I only wanted to make clear why I wouldn't use it. :) :) :)

kilg0r3

1st May 2003, 23:18

Originally posted by Assault
This wasn't meant as an offence. I only wanted to make clear why I wouldn't use it. :) :) :)
Didn't take it to be one. :)

@all
what's curretnly the easiest way to get some info on how to encode DVD->WMV9, including how to split onto 2 CDs, mux two Audio tracks plus subtitles?

Ramirez

3rd May 2003, 01:43

@IWOD
Your observations concerning RV9 are correct;that RV9 rip that I mentioned earlier only looks sharp and detailed at the first look, as a moment I'm making a comparison between RV9 XVID WMV9 so in progress of creation of that comparison I compared almost 250 frames head to head (same bitrate/res) of RV9/XVID/WMV9..Man I was devastated, in many cases RV9 literally destroys quite a big amount of the original frame.

my theory is that's how RV9 achieves its impressive compressions levels,I think that depending on the complexity of the particular scene RV9 designed to destroy every none "vital" parts of the original frame in order to save some bits, so even if we'll ever be able to control RV9's P.P levels we won't see these details back, they're gone,wiped out from existence...(like for example there is a heavy rain in one of the frames in my comparison, same frame taken from RV9 and.. there is no RAIN! it's a miracle!! Now RV9 controls weather:-() Yes I can back it up, my comparison is 70% done (tons of work) I made it in both 600/1000kbps all frames will be easily viewable/switchable in both 600/1000kbps mode. I'll make it available very soon.

Sirber

3rd May 2003, 02:18

@Ramirez

You're right. RV9 controls the weather. RV9 removes all what it thinks it's noise, like noise, noise, rain, noise, backgrounds, and at low bitrates, details. If you pre-process it correctly, you can have nice results (for low bitrates).

ooops!

3rd May 2003, 14:32

> kilg0r3
A strange thing happened after I installed the Xvid codec. I didn’t see it straight away but a co-worker did, immediately. After installation of the Xvid codec, playback of previously encoded DivX files appeared to become more blocky.

This was noticed when the files were viewed via the WinMedia9 and DivX 2.1 ‘software’ players. Strangely too, QuickTime .Mp4 ‘video only’ encoded files were similarly affected but not as much.

I really am not kidding about this, because at work we are able to play copies of the same file on two (or more) identical spec PC’s and output the images to a plasma monitor via ‘split screen’.

When the Xvid codec was removed and a system restore initiated the output returned to normal!

Output via the ‘hardware’ Sigma Xcard was not affected. (However I do seem to remember a similar thing happening when I installed the Sigma Designs Mpeg4 codec some time ago.)

> All
Has anybody else noticed this or know of anyone experiencing this anomaly?

Also, I wonder if some of you guys could confirm my findings by performing your own tests using the same ‘test file’ as me (assuming you have the DVD ofcourse). And then post your results. As I think this will be the only way we can all see which codec looks the best.

I’m using ‘Star Wars 2 - Chapter 42 - Yoda’s Cavalry’ at 1min 17sec / Video Only - MPEG-2 - 720x576 (PAL) - 16~9 - Letterboxed.M2V / File size - 49,448KB (48.2MB).

All your output files must have an output size of '720x416' (PAL in my case), for the test to be fair and accurate.

I know some of you are in NTSC land, so the file will be different. But it might be interesting all the same.

My kbps settings and file size totals are as follows: -
01: WM9 2pass @ 615kbps VBR = 5,964KB
02: DivX5 5pass @ 629kbps VBR = 5,960KB
03: RM9 2pass @ 625kbps VBR = 5,900KB
04: QT MP4 1pass @ 664kbps CBR = 5,955KB 'absolute pants'!

Happy encoding............

ChristianHJW

3rd May 2003, 14:36

Has anybody time to test WMV9 VCM in matroska with Vorbis please ? Should work by now ...

ooops!

3rd May 2003, 15:32

> kilg0r3 / Assult

I'm trying the Xvid codec, using the settings you guys gave me but all I get is a blocky mess!

I wonder if I'm not using the correct 2pass method. Can either of you provide me with the info I need to do it manually (i.e. without the use of an all singing and dancing 2 pass application).

Bye the way have any of you guys tried MPEGmediator v1.5. I came across it by chance the other day and it's the fastest encoder I've used.

Sure, it doesn't have all the bells and whistles of other applications, like auto 2pass, but you can easily change the ratio of the source image, perform crops etc under the 'Filters' tab and create 2Ch wav files from an ac3.

Just make sure in the 'Output' tab you select 'OpenDML AVI Output' so you can get to the Mpeg4 codec.

And when its encoding if you dont display the frames, it's even faster!

Sirber

3rd May 2003, 15:52

MPEG Quants give less blocks, but sometimes moskitos noise.

Assault

4th May 2003, 03:26

@ ooops!

Try it with VirtualDubMod. You have to select 2pass - 1st pass for the first pass and 2pass - 2nd pass int. for the second. Furthermore you should use avisynth 2.5 to avoid colorspace conversions (it's faster). ;)

Regards
Assault

kilg0r3

4th May 2003, 10:22

OOps!
Kilg0r3 asks what WM9 encoder I use… Well it’s just Microsoft’s standard encoder. There really is no quick way of encoding WM9 ... What do you mean by quick; encoding speed or easy to understand, set up?
Aussault: You should use Koepi's latest unstable version XviD-05042003-1.
Yes definitely.
After installation of the Xvid codec, playback of previously encoded DivX files appeared to become more blocky.
I really don't know what could cause this :confused:
I'm trying the Xvid codec, using the settings you guys gave me but all I get is a blocky mess!
Assuming you are using the latest _unstable_ koepi build, I must say that it is quite possible that xvid cannot cope very well with bitrates as low as yours. At the resolutions given above, I assume, xvid needs at least 900kbit to perform well. !!Just a speculation!!
Can either of you provide me with the info I need to do it manually (i.e. without the use of an all singing and dancing 2 pass application).
I think the best sources for this are currently doom9's guide plus emp3r0r's document (search in the forum). In this order.

Btw, there has been a very detailed test in the german computer magazine ct resulting in the following ranking.
1. WMV9 (God, why do you allow this?)
2. RV9
3. XVID
4. DIVX
The test was conducted with a variety of different resolutions and two bitrates.

ooops!

4th May 2003, 13:01

> Kilg0r3

With regard to the WM9 encoder and front end provided by Microsoft, it takes along time to encode and create a 2pass WM9 file.

It can also be a bit difficult to set up the encoder correctly in the first place, if you are new to it.

However once you get your head around it and create your own presets the results are good.

But if you are inputting large files it's vital that you encode the audio and video separately and use the 'media stream editor' to mux them back together.

Thanks for your info about XviD 2pass at low bit rates. I tried encoding my favourite 1min 17sec 'test file' using 1pass CBR 630kbps, 1pass quality and 1pass quantizer modes. The resulting files were at least watchable.

When I tried creating 2pass file, I was hoping I would obtain a better quality image. But this is when I got the 'blocky mess'.

So this is what lead me to believe that it's me that's doing something wrong an not the encoders fault.

Any feedback would be great. Thanks

haibane

4th May 2003, 22:11

Originally posted by ChristianHJW
Has anybody time to test WMV9 VCM in matroska with Vorbis please ? Should work by now ...

I v0.3.1, with WMV9, it produced result like this:
http://skywalker1.mysitespace.com/Snap1.jpg

however, the the vorbis can still be decode ok......

ChristianHJW

4th May 2003, 22:58

Originally posted by haibane
I v0.3.1, with WMV9, it produced result like this:
http://skywalker1.mysitespace.com/Snap1.jpg

however, the the vorbis can still be decode ok...... .. we are looking into this, we feel it micht be a problem with 1st frame not being reported as a keyframe from our parser filter, and then the filter hangs.

Somebody with more time than me should test with a keyframe interval of 1, if this could help ( to have a lot of keyframes in the beginning ... ) ...

haibane

5th May 2003, 00:18

Originally posted by ChristianHJW
.. we are looking into this, we feel it micht be a problem with 1st frame not being reported as a keyframe from our parser filter, and then the filter hangs.

Somebody with more time than me should test with a keyframe interval of 1, if this could help ( to have a lot of keyframes in the beginning ... ) ...

I tried a encode with all keyframes, I can't even start the play, it is locked in (paused/stopped mode), I was using MPC and WMP7.

amirm

5th May 2003, 06:40

I have seen a lot of requests for turning off post filters on WMV9. Here is the way to do it with a reg key setting:

\HKEY_CURRENT_USER\Software\Microsoft\Scrunch\Force Post Process Mode

This is a DWORD value. The valid value is from 0 to 4, 0 being turning off post filter and 4 being the strongest. If the value is outside this range, the decoder will use the automatic mode to decide the post filter strength.

Note that the faster your machine (up to a limit), the stronger the (automatic) postfilter is which may explain why people are seeing different results. Let me know what you think.

Amir