Good quality? What changes could be made?

[darkdream4]

I was wondering if the way Ive been encoding my backup is decent enough quality in your opinion. Im a bit picky with wanting the best quality for my backup and was wondering if anyone would care to tell me ways to improve this or make it better if its not already good enough.

Heres a small 38 second sample clip.
http://www.megaupload.com/?d=FGA6YXXB

If you notice anything you think I should work on please gimme some feedback on it so I can improve at it. Mediainfo should give you the answers to most of your questions on how it was rendered but if not, feel free to ask what I did with it and what I used.

""""""BECAUSE OF SO MANY REQUESTS FOR THIS AND CLAIMING I SHOULD SHOW THE SOURCE"""""

Ive added the same 38 second sample's VOB Source so you can see that as well.

http://www.megaupload.com/?d=3CDKXGCQ

[poisondeathray]

Most people would use 640x480 , not 630x480 , for a 4/3 AR - so your aspect ratio is a bit off . So if you cropped, most would resize to 640x480

Some standalone players will have problems with non-mod16 or mod8 resolutions (not evenly divisible by 16 or 8 ) , and display green or corrupted borders. Also using mod16 dimensions will maximize compression efficiency , because most codecs will pad internally to mod16 anyway

There were some frame repeats and a few blends, not sure if it was from bad IVTC, or the source is screwed up. Anime is often not perfect

One thing that will bother picky anime folk are the minor edge artifacts around lines. If it was in the source, then some might choose filter it. If it was caused from the xvid compression, some might choose to use higher bitrate, or use different compression like h.264

If you think it's good enough , that's all that is important. A lot of it is personal preference. Some like really clean anime, some prefer to have some noise.

[darkdream4]

Most people would use 640x480 , not 630x480 , for a 4/3 AR - so your aspect ratio is a bit off . So if you cropped, most would resize to 640x480

Some standalone players will have problems with non-mod16 or mod8 resolutions (not evenly divisible by 16 or 8 ) , and display green or corrupted borders. Also using mod16 dimensions will maximize compression efficiency , because most codecs will pad internally to mod16 anyway

There were some frame repeats and a few blends, not sure if it was from bad IVTC, or the source is screwed up. Anime is often not perfect

One thing that will bother picky anime folk are the minor edge artifacts around lines. If it was in the source, then some might choose filter it. If it was caused from the xvid compression, some might choose to use higher bitrate, or use different compression like h.264

If you think it's good enough , that's all that is important. A lot of it is personal preference. Some like really clean anime, some prefer to have some noise.

I only used 630 x 480 because I cropped 5 pixels off of each side of the video to remove black lines and I didnt want it to be stretched any to fit 640 x 480 that way it stayed normal, I wanted to keep it regular like I saw it in powerdvd. It was 640 x 480 with the black lines in powerdvd. I thought if you take off the black lines keeping that off the resolution size wouldnt hurt any. Is that really a big issue? I was encoding this for PC playback, I didnt mean for it to be burned to DVD's

Frame repeats and blends? Id like to know how you could tell that and or how its fixed lol. I didnt really alter the source too much. I actually think the dvd source was a bit screwy from this anime. It actually ended up looking almost better than the source did to me. The only filter I remember using other than the resize and cropping ones are the deinterlace only when necessary one. Could that have caused the frame repeats and blends? Without it I was seeing lines through the video when it play back, they looked like blinds on the video and were annoying. (I made a post about it here https://forum.videohelp.com/topic377726.html)I couldnt find out how to get rid of em except by this filter. Deinterlace blurred the video some and IVTC didnt fix them all so I didnt know how to handle it other than the one video filter.

What exactly do you mean by edge artifacts around lines?

Also I was going to use the h.264 compression but I couldnt find any good software to do it with. Except perhaps MeGUI but it seemed a tiny complicated to me.

What would you suggest I use to make these fixes?

I can give you a sample of the source where the video is done with direct stream copy that way you can view the source if needed.

[Bwizzy]

you have to tell me what you used to encode that sample video. I have a lot of media I'm trying to encode to files for my home server (mainly 32 disc of DBZ). Gimme all the settings also.

[darkdream4]

you have to tell me what you used to encode that sample video. I have a lot of media I'm trying to encode to files for my home server (mainly 32 disc of DBZ). Gimme all the settings also.

I practically used and did almost everything you will see and read on this. http://www.megaupload.com/?d=IPO4RF1W

All I did was change a few settings for the better in VirtualDubMod and added one more filter that was not mentioned in there because it fixed the issue I mentioned above and made the output better.

I love DBZ and have many Discs of it if you ever want to use them.

[Bwizzy]

Man.... That's a lot of stuff you need to do it like you do. But the proof is in the pudding. do you use this method for all your anime on DVD? if so, are you able to do multiple audio streams and soft subs? I also noticed you're using XP, do you know if any of this works on Windows 7 64bit?
Do you use all those apps that were in that guide?

[manono]

I only used 630 x 480 because I cropped 5 pixels off of each side of the video and I didnt want it to be stretched any to fit 640 x 480 that way it stayed normal, I wanted to keep it regular. Is that really a big issue?

Well, aren't you the one that said you're picky about quality? And if so, you may as well learn to do it right.
1. Crop the widths by an even numbers of pixels - no odd numbers like your 5 on each side. If it means cropping into the active video, then so be it.
2. A 640x480 AVI resize is based on a 704x480 DVD source, not 720x480, so ideally you'd crop 8 pixels from each side before resizing to 640x480. poisondeathray is correct in saying your aspect ratio is off. 640x480 will be closer to correct, even if cropping 6 from each side.
3. I'm a firm believer in Mod16 resolutions (height and width divisible by 16), but you shouldn't go below Mod8. Your 630 is Mod2. It's unacceptable as a width. The greatest encoder efficiency and the greatest playback compatibility comes with resizing to Mod16 resolutions.

What exactly do you mean by edge artifacts around lines?

Mosquito noise - all the little dots outside the lines. They might already be in the source, hard to know without a sample from the DVD. If you're really serious about your video encoding hobby, you'll learn to frameserve using an AviSynth script, and not rely on VDub for such things as IVTC and other filtering.

[jagabo]

What exactly do you mean by edge artifacts around lines?

Mosquito noise - all the little dots outside the lines. They might already be in the source, hard to know without a sample from the DVD.

You have to step Xvid down to quantizer = 1 to prevent DCT ringing artifacts. At that point there's so little compression there's no point in using Xvid.

You can enable Xvid's deblocking and deringing features for playback on your computer. That will reduce the artifacts when playing the video. But you don't have that control on other peoples players, or standalone players.

[manono]

You have to step Xvid down to quantizer = 1 to prevent DCT ringing artifacts. At that point there's so little compression there's no point in using Xvid.

Maybe, maybe not. A little judicious filtering can help a lot:

http://forum.doom9.org/showthread.php?t=134078

Since he's the main x264 dev, he's using x264, but the same results can be obtained using XviD, although a better matrix than the 2 main ones may be called for. Yes, the size might jump, but there's no reason an XviD encode can't look as good (or better) as the source DVD.

[jagabo]

You have to step Xvid down to quantizer = 1 to prevent DCT ringing artifacts. At that point there's so little compression there's no point in using Xvid.

Maybe, maybe not. A little judicious filtering can help a lot:

http://forum.doom9.org/showthread.php?t=134078

Since he's the main x264 dev, he's using x264, but the same results can be obtained using XviD

Sure, he can reduce the DCT ringing in his source, but Xvid compression will recreate them. The sharper the source, the worse the problem will be.

A pristine source (all images are 8x point up-sized, cropped from a larger frame, all I frames):



encoded with Xvid (Q=3, Q=2, Q=1), played back with Xvid's deringing feature disabled:



With the decoder's DeRinging feature enabled:

[darkdream4]

Man.... That's a lot of stuff you need to do it like you do. But the proof is in the pudding. do you use this method for all your anime on DVD? if so, are you able to do multiple audio streams and soft subs? I also noticed you're using XP, do you know if any of this works on Windows 7 64bit?
Do you use all those apps that were in that guide?

Yes I used every piece of software and step in that and it does need to be followed just like that. The only step you might be able to skip is the fixing the AC3 delay because I havent had one yet with the files. The settings noted in it could be changed like the zone to 1.0 weight instead of 0 for example. 0 Just drags the quality down. I didnt want a one click do everything for me application. I prefer the longer steps and more use of different softwares so I get more customization and quality. If you do Id suggest AutoGK if you can get it to work. I was having problems with it giving me my video file.

Yes you can use multiple audio streams and soft subs, youll just need to extract the subtitles with a method not mentioned in there. Meaning another piece of software thats not on that list for ya. I think Ill edit that pdf file to explain adding more than one stream and optional subtitles.

You sound like you just wanted to know how I did it without much exp yourself lol. Im not exactly sure it works on windows 7 but I dont know why they wouldnt. "I couldnt get it to work on windows Vista when it came to using proxyoff registry entry that comes with VirtualDub so you can use the VFAPICodec program" That was keeping me from doing it on windows Vista but windows 7 is different and I have no exp with that operating system. So go ahead and try them out and see for yourself. I dont see why they wouldnt, the sites you download em from should let you know what their compatible with.

[darkdream4]

I only used 630 x 480 because I cropped 5 pixels off of each side of the video and I didnt want it to be stretched any to fit 640 x 480 that way it stayed normal, I wanted to keep it regular. Is that really a big issue?

Well, aren't you the one that said you're picky about quality? And if so, you may as well learn to do it right.
1. Crop the widths by an even numbers of pixels - no odd numbers like your 5 on each side. If it means cropping into the active video, then so be it.
2. A 640x480 AVI resize is based on a 704x480 DVD source, not 720x480, so ideally you'd crop 8 pixels from each side before resizing to 640x480. poisondeathray is correct in saying your aspect ratio is off. 640x480 will be closer to correct, even if cropping 6 from each side.
3. I'm a firm believer in Mod16 resolutions (height and width divisible by 16), but you shouldn't go below Mod8. Your 630 is Mod2. It's unacceptable as a width. The greatest encoder efficiency and the greatest playback compatibility comes with resizing to Mod16 resolutions.

What exactly do you mean by edge artifacts around lines?

Mosquito noise - all the little dots outside the lines. They might already be in the source, hard to know without a sample from the DVD. If you're really serious about your video encoding hobby, you'll learn to frameserve using an AviSynth script, and not rely on VDub for such things as IVTC and other filtering.

Well that sure explains alot to me. I didnt know making it 640 x 480 and then cropping off the black lines was a bad thing. I just noticed the image stretched a little more and when comparing it to powerdvd screenshot it was obvious it was a little stretched but if thats better you say, Ill believe since I know you pretty well from Doom9.

Those dots were in the source alright, I think just a little "less noticeable" but still in it sadly. I really wanted to figure out how to use that AVISynth scripter but I couldnt figure out how to use it. Perhaps Ill read up on just AVISynth and see if I can figure it out. I keep hearing its done in notepad from ppl on Doom9 but I just couldnt seem to get it going lol.

Thanks for that info, it helps.

[darkdream4]

What exactly do you mean by edge artifacts around lines?

Mosquito noise - all the little dots outside the lines. They might already be in the source, hard to know without a sample from the DVD.

You have to step Xvid down to quantizer = 1 to prevent DCT ringing artifacts. At that point there's so little compression there's no point in using Xvid.

You can enable Xvid's deblocking and deringing features for playback on your computer. That will reduce the artifacts when playing the video. But you don't have that control on other peoples players, or standalone players.

Well I did use quantizer 1 because I was picky, but 2 wont hurt either. All 1 really does from my reading on Doom9 is better quality thats hardly even noticable to the human eye and a bigger file size which really isnt needed to be bloated out that big.

[poisondeathray]

Well I did use quantizer 1 because I was picky, but 2 wont hurt either. All 1 really does from my reading on Doom9 is better quality thats hardly even noticable to the human eye and a bigger file size which really isnt needed to be bloated out that big.

IF you used Q=1, that suggests most of the noise was from the source. Consider pre-filtering it, denoising it will reduce the bitrate requirements (noise "eats up" bitrate). And consider using h.264 (e.g. x264 encoder) which is much more efficient than xvid for compression.

[darkdream4]

Well I did use quantizer 1 because I was picky, but 2 wont hurt either. All 1 really does from my reading on Doom9 is better quality thats hardly even noticable to the human eye and a bigger file size which really isnt needed to be bloated out that big.

IF you used Q=1, that suggests most of the noise was from the source. Consider pre-filtering it, denoising it will reduce the bitrate requirements (noise "eats up" bitrate). And consider using h.264 (e.g. x264 encoder) which is much more efficient than xvid for compression.

Well I now have the source sample above if you wish to check it out.

[Bwizzy]

Man.... That's a lot of stuff you need to do it like you do. But the proof is in the pudding. do you use this method for all your anime on DVD? if so, are you able to do multiple audio streams and soft subs? I also noticed you're using XP, do you know if any of this works on Windows 7 64bit?
Do you use all those apps that were in that guide?

Yes I used every piece of software and step in that and it does need to be followed just like that. The only step you might be able to skip is the fixing the AC3 delay because I havent had one yet with the files. The settings noted in it could be changed like the zone to 1.0 weight instead of 0 for example. 0 Just drags the quality down. I didnt want a one click do everything for me application. I prefer the longer steps and more use of different softwares so I get more customization and quality. If you do Id suggest AutoGK if you can get it to work. I was having problems with it giving me my video file.

Yes you can use multiple audio streams and soft subs, youll just need to extract the subtitles with a method not mentioned in there. Meaning another piece of software thats not on that list for ya. I think Ill edit that pdf file to explain adding more than one stream and optional subtitles.

You sound like you just wanted to know how I did it without much exp yourself lol. Im not exactly sure it works on windows 7 but I don't know why they wouldnt. "I couldnt get it to work on windows Vista when it came to using proxyoff registry entry that comes with VirtualDub so you can use the VFAPICodec program" That was keeping me from doing it on windows Vista but windows 7 is different and I have no exp with that operating system. So go ahead and try them out and see for yourself. I dont see why they wouldnt, the sites you download em from should let you know what their compatible with.

You are absolutely right. I've been experimenting with one click encoders for a little while now, with bad results. I'm way better at encoding BD movies than I am at DVD anime. And anime is my bread and butter whenever I wanna watch TV or something. I have a lot of anime on DVD I need to encode, so there's no time like the present to learn. Windows 7 has an XP mode I still have yet to try, so I guess I'll be experimenting with your technique here. If you change the .pdf guide i have, be sure to PM me or post it on this thread. Thanks

[poisondeathray]

I looked at your source, and the blends in your encode are from your poor IVTC or deinterlacing method

Here is an example of what I mean, from your encode. This isn't present when you use avisynth to IVTC (using TFM().Tdecimate() )



The minor line noise (or mosquito noise) is present in the source. But it's made worse by xvid compression, or whatever settings you used. Most people would filter it slightly before encoding. If you can't see it, pay attention to the scenes where the character is running, and look at the grey patches near his shoulders, and neck. It will look similar to those examples jagabo posted earlier. Most anime fans detest this type of noise, but again, this is personal opinion

In this zip file, are 2 .pngs, 1 of the source, and 1 filtered. Flip back and forth, and pay attention to the grey areas mentioned earlier, the right armpit. These are uncropped 1:1, and not resized, and before any compression.


257.zip

[darkdream4]

Your filtered looks great compared to the other one. I wish I knew how ya did it. I also detest those things I noticed that your filtered gotten rid of. I was trying to get rid of the noise myself but couldnt get it all. Yours looked pretty amazing to me.

Thanks for all the info and help. Apparently trusting the VirtualDub or VirtualDubMod app to deinterlace or IVTC is a big mistake. I guess Ill have to learn scripting.

[poisondeathray]

Filtering is only 1 part of the equation, the other part is encoding settings and behaviour

I only know the basics, and some simple filters. The complex hardcore guys and their scripts can be found on doom9, and I'm sure they could do a way better job

This was my script, with the crop and resize added

MPEG2Source("VTS_01_1.d2v")
TFM().TDecimate()
Crop(6,0,-4,0)
LanczosResize(640,480)
Deen()
LSFMod(strength=50)

Once you learn the avisynth basics, here is a helpful page to start learning about filters; it's actually geared towards anime. You mouse over some of the pictures and see the differences
http://www.aquilinestudios.org/avsfilters/

[darkdream4]

Well so far your about the only one to answer my questions thoroughly and actually help out "alot" compared to some other people who have an act like their gods complex and do nothing but say "read this" and talk big without much more help than that. But I know how it can be trying to help someone whom you think doesnt have a clue sometimes.

I gotta give you two thumbs up for that one and I appreciate it alot as well!! So thanks for both the feedback and the details, I learned plenty today from both forums that I use.

I dont like to give up easily on learning something Im curious about. if I have the freetime Ill continue till I get it down or cant stay awake to keep going and then just resume when I wake up lol. But if I get this down and eventually become a hardcore scripter as you put it, Ill be more than glad to make you some as a thanks.

I only have one more quick question. If I wanted to use that script, how would I be able to use it? I have AVISynth installed but its obviously not a program. Apparently I write the scripts in notepad. Once that parts done, how do I use the script? Thats pretty much the only other question I can think of to get me started.

[poisondeathray]

Well , honestly a lot of this stuff has been covered already, again and again in various threads, and some people get sick of repeating it over and over.

You can learn a lot just by searching and looking in the forums. I know the search engine sucks (especially at D9), but even if you pick up tidbits here and there you will eventually learn enough to get by. Just make sure you research and search thoroughly before you ask redundant questions (especially at D9!). Topics tend to be a bit more advanced there compared to here.

I hated avisynth for years, but now that I know how to use it a little bit, I can't do anything without it. It's easily the most versatile tool for many many video & audio functions. I can't emphasize enough to start learning the basics!

Good luck in your learning

[darkdream4]

If you're really serious about your video encoding hobby, you'll learn to frameserve using an AviSynth script, and not rely on VDub for such things as IVTC and other filtering.

Well Manono after all the reading Ive done and help Ive recieved thats exactly what my next plan is. To learn AVISynth to the point I feel as if I created it myself. Apparently thats about the only thing keeping me from my goal at the moment.

[jagabo]

http://avisynth.org/mediawiki/Main_Page#New_to_AviSynth_-_start_here
http://avisynth.org/mediawiki/First_script

[darkdream4]

This was my script, with the crop and resize added

MPEG2Source("VTS_01_1.d2v")
TFM().TDecimate()
Crop(6,0,-4,0)
LanczosResize(640,480)
Deen()
LSFMod(strength=50)

Ive been reading about this pretty damn hard and I at least understand your script now so that must be a good thing lol. I think this AVISynth is pretty amazing so far. I can see why ppl talked so much crap about VirtualDub when it came to comparing the 2. Theres only one or 2 things I havent heard of though. The Deen, TFM and LSFMod you used. In fact I couldnt even get em to do anything when I tried em out. I just got script errors in WMP. Can I ask what each of those are for? Ill probly find out soon enough but it wont hurt me to ask.

[poisondeathray]

Ive been reading about this pretty damn hard and I at least understand your script now so that must be a good thing lol. Theres only one or 2 things I havent heard of though. The Deen, TFM and LSFMod. Can I ask what each of those are for? Ill probly find out soon enough but it wont hurt me to ask.

Come on, you're not trying very hard... If you asked this at D9, you would be shot and killed.

All the filters have documentation. Deen is for denoising, TFM comes with TIVTC and is a field matcher, LSFMod is a sharpener. They all have individual settings, so read the documentation

http://www.aquilinestudios.org/avsfilters/spatial.html#deen
http://avisynth.org.ru/docs/english/externalfilters/tivtc.htm
http://forum.doom9.org/archive/index.php/t-142706.html

[darkdream4]

Ive been reading about this pretty damn hard and I at least understand your script now so that must be a good thing lol. Theres only one or 2 things I havent heard of though. The Deen, TFM and LSFMod. Can I ask what each of those are for? Ill probly find out soon enough but it wont hurt me to ask.

Come on, you're not trying very hard... If you asked this at D9, you would be shot and killed.

All the filters have documentation. Deen is for denoising, TFM comes with TIVTC and is a field matcher, LSFMod is a sharpener. They all have individual settings, so read the documentation

http://www.aquilinestudios.org/avsfilters/spatial.html#deen
http://avisynth.org.ru/docs/english/externalfilters/tivtc.htm
http://forum.doom9.org/archive/index.php/t-142706.html

I might be shot and killed for it lol. I only got past the first 4 things on the avisynth main page.
Your first script - A guide for total newbies.
Getting started - An introductory guide for getting started with AviSynth.
Filter introduction - An introductory guide to most common AviSynth filters.
Script examples - Assorted examples for putting everything together in the real world.

a couple hours of reading. Thanks Ill go read em.

Also Ive went ahead and gave the most basic script a try for practice and I was wondering. could you take a look at this sample and tell me why the audio is doing what it does. youll notice at the very beginning for the first 2 seconds or so it plays audio from half way through the clip and then starts to play normally.

http://www.mediafire.com/file/yjzzogfmojg/dgdecodefirsttry2.avi

heres the script I used on it.

LoadPlugin("M:\Desktop M\My Documents\My Videos\Encoding Stuff\VOB To AVI stuff\dgmpgdec157\DGDecode.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\ac3source.dll")
MPEG2Source("C:\Documents and Settings\Aiden\Desktop\VIDEO_TS\VTS_01_1.d2v")
AC3Source("C:\Documents and Settings\Aiden\Desktop\VIDEO_TS\VTS_01_1 T80 2_0ch 192Kbps DELAY 0ms.ac3")

[poisondeathray]

load your audio with nicac3source() instead (nicaudio.dll)

don't judge your avs audio sync by vdub, it's not meant as a player. Use MPC or a video player

when you have filters, it will slow playback (and encoding) because they are cpu intensive. If you want
to check sync, eliminate the all filters except the basic ones , because the filters maybe the bottleneck and it will not be an indication of sync

MPEG2Source("VTS_01_1.d2v")
TFM().TDecimate()
vid=last

aud=NicAC3Source("VTS_01_1 T80 2_0ch 224Kbps DELAY -85ms.ac3").DelayAudio(-0.085)

AudioDub(vid,aud)

If sync is ok (and it is for me), then add them back

MPEG2Source("VTS_01_1.d2v")
TFM().TDecimate()
Crop(6,0,-4,0)
LanczosResize(640,480)
Deen()
LSFMod(strength=50)
vid=last

aud=NicAC3Source("VTS_01_1 T80 2_0ch 224Kbps DELAY -85ms.ac3").DelayAudio(-0.085)

AudioDub(vid,aud)

most .dll's that are in the avisynth/plugins folder will autoload (no need to load them in the script manually,
but some still do need LoadPlugin), and .avsi scripts in the plugins folder will autoload as well. For
example, limitedsharpenfaster, and lsfmod both come as .avsi and can autoload. Read the documenation
on loading scripts for more info

A great tool to learn using scripts is AvsP. You can preview different scripts in different tabs, and use F5 to see the video preview, and make changes quickly. More advanced features include slider control, macros, etc.. Read the documentation for full list of features

Always preview your script (video), and playback the basic avs script in a media player to check audio sync, before encoding, or you're going to be wasting a lot of time

[darkdream4]

load your audio with nicac3source() instead (nicaudio.dll)

don't judge your avs audio sync by vdub, it's not meant as a player. Use MPC or a video player

most .dll's that are in the avisynth/plugins folder will autoload (no need to load them in the script manually,
but some still do need LoadPlugin), and .avsi scripts in the plugins folder will autoload as well. For
example, limitedsharpenfaster, and lsfmod both come as .avsi and can autoload. Read the documenation
on loading scripts for more info

A great tool to learn using scripts is AvsP. You can preview different scripts in different tabs, and use F5 to see the video preview, and make changes quickly. More advanced features include slider control, macros, etc.. Read the documentation for full list of features

Thanks for the links, they sure helped alot. I didnt find those on the main wiki for avisynth. http://avisynth.org/mediawiki/Main_Page Apparently I need the plugins too because I only have the ones that came with the download except for the ac3audio.dll and mpeg2, dgdecoder ones.

I didnt use Virtualdub to see the audio sync, I just used VirtualDub to create the avi file with that script and when I play it back in windows media player I hear that audio problem. but if I load the ac3 separately as a stream in virtualdub and not part of the script I use in virtualdub it seems to work perfectly in the created avi file.

oh I didnt know they autoload, I was just doing it how the site doom9 gave taught me to. http://neuron2.net/dgmpgdec/QuickStart.html some of it wasnt working for me when I tried it. When it got to the adding audio part thats when I was getting script errors.
Ill give it a shot without it and see if it works like you say.

why nicac3cource.dll instead of nicaudio.dll?

Ill go check out that tool now, it sounds like it must be a pretty good one. Thanks!!

[poisondeathray]

nicaudio.dll is the plugin

The notation is nicac3source() for ac3sources, there is also NicDTSSource(), NicMPG123Source() etc, etc.. all from the same nicaudio.dll. It just loads different types of audio

AC3Source() is an old plugin and doesn't work well IIRC

Some people like using the LoadPlugin() for everything, as it is their way of organizing things. The more autoloading plugins and avsi scripts you have, the slower avisynth becomes for loading. You decide for yourself how you want to organize stuff

I added a line above, to make sure you preview your scripts before encoding. This is solid advice, or you're going to be wasting a lot of time encoding

[manono]

Sure, he can reduce the DCT ringing in his source, but Xvid compression will recreate them. The sharper the source, the worse the problem will be.

I have no intention of going down to the pixel level for this, but I have zoomed in a bit. I've enclosed 6 pics in a ZIP file, after reencoding the sample 5 times (one of the pics is of the uncompressed source). My contention is that the 2 main matrices used for XviD are designed for heavy compression - for compressing movies down to a CD or 2. But with better matrices (and a larger size), XviD can do a decent job without creating much noise around edges. Anime is a good test, because of all the hard edges it has.

1.Uncompressed-Unencoded.png It is this script opened in VDubMod and a PNG taken, then the PNG is zoomed in. The original script:

TFM(D2V=("E:\Test\test.d2v"),PP=0)
TDecimate()
Crop(6,0,-4,0)
LanczosResize(640,480)

The zooming script for the PNG:

Crop(166,184,-200,-90)
LanczosResize(640,480)

That's the base. The original VOB sample was 34.8 MB in size.

2. Fox-Quant3-Unfiltered.png That same script encoded for XviD, a pic taken and zoomed in. I'm using the same frame that poisondeathray used. The matrix used is sometimes called Fox Home Theater and is commonly used on DVDs. The reencoded size is 18.7 MB

3. Fox-Quant3-LightFilter.avi Same matrix and script with the addition of some light filtering, similar to what PDR suggested:

Deen()
Dull=Last
Sharp=Dull.LimitedSharpenFaster(ss_x=1.25,ss_y=1.2 5,Smode=4,strength=350,soft=30)
Soothe(Sharp,Dull,25)

This is, in my opinion, the best of the bunch, and better than the source as some of the noise has been removed, without the details taking much of a hit. Size is 17.5 MB.

4. MPEG-Quant3-Unfiltered.png Everything the same as pic #2 except for using the common MPEG Matrix. Size is 8.46 MB.

5. MPEG-Quant3-LightFilter.png Everything the same as pic #3 except for the use of the MPEG matrix. Size is 8.72 MB. Both pics 4 and 5 have a greatly increased amount of noise which stands to reason when reencoding to a lossy format using tremendous compression.

6. Fox-Quant3-HeavytFilter.png Filtered it differently, following the script (pretty much) in the link I posted earlier:

Deblock()
dfttest()
FastLineDarken()
deen("a3d",4,8,9)
Spline36Resize(last.width*2,last.height*2)
aWarpSharp(depth=12,blurlevel=4,thresh=0.2,cm=1)
FastLineDarken()
aWarpSharp(depth=6,blurlevel=4,thresh=0.7,cm=1)
FastLineDarken()
DeHalo_Alpha()
fft3dFilter(bt=3,sigma=8,sharpen=1.3)
Spline36Resize(last.width/2,last.height/2)

Slow as hell. It pretty much wipes out the noise, but most of the texture detail on walls and elsewhere as well. Size 13.6 MB.

Conclusion? A Custom Quant matrix won't create the noise that the 2 common matrices will. The size increases, and rather than getting a file roughly 25% the size of the source you get one roughly 50% the size. The file size savings is still considerable. Some standalones won''t play XviDs that use Custom Quant matrices. Those with ESS chipsets won't. I believe that those with Mediatek chipsets will. It's been several years since I tested that, though. HTPCs and the similar boxes that feed into TV sets shouldn't have any problems at all. I'm not sure, jagabo, if you're saying that XviD by its very nature creates noise or if using the common settings and matrices will result in noise being created. I'm saying that less compression, but without having to go down to quant 1, can result in a video that for all intents and purposes is transparent to the source.

zoomed.zip