Avisynth Help thread [Archive] - Page 2

Zergrinch

2009-02-08, 05:40

@daiz:
Thanks for your recommendations. I'm afraid most of those settings is greek to me, which is precisely why I was using presets to begin with :).

@checkers:
I do believe it was "said" circa 2004 or 2005 :p Those two functions took a lot of tracking down of filters and scripts. Avisynth at its current is sure unfriendly to noobs like me! At any rate, I will try one last time with meGUI, using the script settings you recommended.

I used a preset called "Unrestricted 2pass Anime_Toons HQ". I didn't know what DVXA meant, actually, and was just doing it based on the default recommendations (http://forum.doom9.org/showthread.php?t=139765) on the Doom9 forum. For your reference, the command line was:

program --pass 2 --bitrate 1000 --stats ".stats" --ref 8 --mixed-refs --bframes 16 --b-pyramid --direct auto --deblock 1:1 --subme 7 --trellis 2 --psy-rd 0.6:0 --partitions all --8x8dct --me umh --threads auto --thread-input --aq-mode 0 --progress --no-psnr --no-ssim --output "output" "input"

http://upload.jetsam.org/images//meGUI-05-preview.jpg

http://upload.jetsam.org/images//meGUI-06.jpg

Which plainly shows that the external filters aren't having any effect.

Perhaps I'm doing this the wrong way, or meGUI is doing this the wrong way. In any case, I'd like to test this with Virtualdub and see if avisynth is to blame. Could anyone possibly point me to a tutorial where AviSynth is used in conjunction with Virtualdub, as I have never done that before.

Thanks, everyone.

checkers

2009-02-08, 06:11

the problem is probably your renderer. I'm guessing that MPC is using something that works in a retarded way on your system (read: VMR*). Try changing it (http://www.cccp-project.net/wiki/index.php?title=FAQ#How_do_I_change_video_renderer s.3F).

Zergrinch

2009-02-08, 06:30

..........

That worked.

..........

I've been watching movies that look washed out and retarded all this time.

..........

Thanks a bunch, checkers! :D

andy_blah

2009-02-08, 07:07

I want to burn an animated logo into a video file. My first problem is that when I open this script in VirtualDub (for testing purposes, I actually encode my videos in H.264 with MeGUI):

video = DirectShowSource("opening.avi").ConvertToRGB32
logo = ImageSource("%04d.png", start=1, end=9, fps=2,pixel_type = "rgb32")
logo = ChangeFPS(logo, 29)
Layer(video, logo, x=600, y=0)

And I see the video opening, and the logo displays, but it only shows one image from the sequence. The images are from 0001.png to 0009.png (will increase in number as I use those 9 images just for testing).

I am converting to RGB32 because the logo requires some transparency areas, so there will be no problem converting it back to YV12 (by the way, does it affect the video size if the video is in RGB32?) in VirtualDub, but what about MeGui, where I cannot do this, should I put an ConvertToYV12() at the bottom of the script? Will this affect the transparency of the logo?

Thank-you in advance

TheFluff

2009-02-08, 09:42

I am converting to RGB32 because the logo requires some transparency areas
layer()/overlay() does not require the overlay and the base video to be the same colorspace, everything is converted to an internal intermediate format anyway, so no need to go yv12->rgb32->yv12

(by the way, does it affect the video size if the video is in RGB32?)
if by size you mean resolution then no; if you by size mean size in megabytes then yes, but only in memory, not when encoded, because no mpeg4 codec supports rgb32, it's always converted to some yuv variant (usually yv12) before encoding

should I put an ConvertToYV12() at the bottom of the script?
no, you should never convert the base video to rgb32 to begin with so it shouldn't be necessary

Will this affect the transparency of the logo?
no

andy_blah

2009-02-08, 11:17

Thank-you for the anwsers. But now, can somebody help me with my logo issue?

EDIT: Removed the .ConvertToRGB32, but now it says that image formats don't match

andy_blah

2009-02-11, 10:56

Started a thread of Doom9, and got it fixed, for anybody interested: http://forum.doom9.org/showthread.php?p=1248185

Also, was wondering if I can fix those "ghosts" in some of my videos, apparently the video was poorly encoded (not by me), and I don't have the source DVD or whatever

Here is an example:
http://i61.photobucket.com/albums/h41/andy_blah/snapshot-5.png

Also, what should I do to fix those bouncing up and down videos? (they quite appear on older anime shows)

This next shot from another video file shows that the video is a little bit fuzzy, any suggestions to improve sharpness? Also, any other suggestions? (just to improve the quality of the video)

http://i61.photobucket.com/albums/h41/andy_blah/Snapshot2.png

martino

2009-02-11, 12:26

Also, was wondering if I can fix those "ghosts" in some of my videos, apparently the video was poorly encoded (not by me), and I don't have the source DVD or whatever
Though luck, it may be near impossible to restore it when you don't have the original. You might want to take a look at some of the scripts/filters here (http://avisynth.org/mediawiki/External_filters#Fieldblending_and_Frameblending_r emoval) and do some trial and error to see what you can get out of it.

Also, what should I do to fix those bouncing up and down videos? (they quite appear on older anime shows)
Stab() (http://avisynth.org/mediawiki/Stab) does a decent job.

This next shot from another video file shows that the video is a little bit fuzzy, any suggestions to improve sharpness?
It looks sharp enough to me, actually even too sharp, but that is often subjective. If you really want to sharpen take a look at the WarpSharp package, and LimitedSharpenFaster which people seem to praise quite a lot.

andy_blah

2009-02-12, 10:20

Thank you for the reply

You might want to take a look at some of the scripts/filters here (http://avisynth.org/mediawiki/External_filters#Fieldblending_and_Frameblending_r emoval) and do some trial and error to see what you can get out of it.

Tried some of those, and none seem to do anything to it, some don't even work properly (for example on one plugin, introduced the default arguments written by the author and Avisynth kept on complaining that the arguments are invalid @.@

Stab() (http://avisynth.org/mediawiki/Stab) does a decent job.
Thank you for suggesting that one, it does the job very well, it's only drawback is that I have to crop ^_^

It looks sharp enough to me, actually even too sharp, but that is often subjective. If you really want to sharpen take a look at the WarpSharp package, and LimitedSharpenFaster which people seem to praise quite a lot.

It seems that sharpening it didn't help solving the problem. If you would take a closer look to the two girls, at their sweat drops more precisely, you can see that the image is quite fuzzy in that area, and in other places it is too sharp, also, the colours seem to be wrong, any suggestions?

Scab

2009-02-12, 10:48

It seems that sharpening it didn't help solving the problem. If you would take a closer look to the two girls, at their sweat drops more precisely, you can see that the image is quite fuzzy in that area, and in other places it is too sharp, also, the colours seem to be wrong, any suggestions?

I suppose you could kill the halos with blinddehalo3/awarpsharpdering/similar, then resharpen it evenly with your favorite sharpener.

andy_blah

2009-02-12, 18:47

I suppose you could kill the halos with blinddehalo3/awarpsharpdering/similar, then resharpen it evenly with your favorite sharpener.

Seems that it doesn't help, and add a strange effect to the video
http://i61.photobucket.com/albums/h41/andy_blah/shot3.png

My script:

LoadPlugin("D:\Program Files\Avisynth 2.5\Plugins\msharpen.dll")
DirectShowSource("ep84.avi")
Stab()
blinddehalo3()
MSharpen()

martino

2009-02-12, 20:15

SHiN-gx

2009-02-13, 07:32

Please pardon my English and I'm not yet familiar with the jargon you guys are using.

I'm experiencing these "horizontal lines" artifact during fade in and fade out of scenes. How do I solve this problem?

http://img141.imageshack.us/img141/2274/sample1bl8.png

http://img26.imageshack.us/img26/9875/sample2ww1.png

http://img15.imageshack.us/img15/2696/sample3tv7.png

And this is my avs script recipe:

DGDecode_mpeg2source("H:\My;You vol 1\VIDEO_TS\VTS_01_1.d2v", info=3)
ColorMatrix(hints=true, interlaced=true, threads=0)
tfm(order=1).tdecimate(mode=1)
crop( 4, 0, -4, 0)

LanczosResize(720,480) # Lanczos (Sharp)
#denoise

By the way, I used megui and followed what "ANALYZE" suggested. Am I missing something?

On an unrelated note, how do I encode a hybrid DVD source (.VOB) into a 23fps VFR MKV? <-- for the mean time, I'll check this thread in case this question has been answered.

Thank you.

andy_blah

2009-02-13, 08:59

Well, default settings on BDH3 will not look very nice. In fact, no halo removal script will give you a sharp result on default settings... or at least it's to be expected. Try PPmode=-3, weak but sharp. Also anything that uses aWarpSharp may be of help, so what Scab mentioned, aWarpSharpDering. Searching for "halo" or "ringing" on Doom9 should give many interesting results.

Also FYI, MSharpen tends to increase haloing and ringing much more than most other sharpeners. You may be better off by using something else.

Tried what you suggested, and fiddled with some filters in AviSynth (and my plugin folder gets quite big...) I got this:
http://i61.photobucket.com/albums/h41/andy_blah/shot4.png
The quality got better, tho it is still a little bit blurry. I still have the following problems:

1.) Small details (mostly those that have low contrast towards their surrounding objects, like those sweat drops on the two ladies) are very blurry.
2.) Can't seem to remove those ghosts seen in my post here (http://forums.animesuki.com/showthread.php?p=2211033#post2211033) , maybe somebody else can guide me, what settings to use (I suck at fiddling with filter options...)
3.) The image is a little bit blurry
4.) The colours seem to be wrong, in some place I have a too high saturation

Here is my current script:

LoadPlugin("D:\Program Files\Avisynth 2.5\Plugins\undot.dll")
LoadPlugin("D:\Program Files\Avisynth 2.5\Plugins\bifrost.dll")
LoadPlugin("D:\Program Files\Avisynth 2.5\Plugins\aWarpSharp.dll")
DirectShowSource("ep87.avi")
dupped()
Stab()
BiFrost()
undot()
DeHalo_Alpha(darkstr=.4)
aWarpSharp()

martino

2009-02-13, 10:02

I'm experiencing these "horizontal lines" artifact during fade in and fade out of scenes. How do I solve this problem?
What you need is smartfade() (http://forum.doom9.org/showpost.php?p=1154887&postcount=12). Apply it after deinterlacing and before decimating... I think.

On an unrelated note, how do I encode a hybrid DVD source (.VOB) into a 23fps VFR MKV? <-- for the mean time, I'll check this thread in case this question has been answered.
You might want to take a look at YATTA, although if you feel this is too advanced for you TIVTC does have a hybrid switch IIRC (however the results produced are usually not very good). You'll have to read through the manuals, I don't remember using it myself so I won't be able to tell you out of memory.

Also IIRC M;Y isn't VFR... although it's been a while since I played around with that source... as well as being a bitch to IVTC. ;)

@andy, I believe you're using filters which your source doesn't need, as well as overfiltering. I might be able to have a play with it later, but now I should get back to bed to tackle my fever.

martino

2009-02-13, 20:06

1.) Small details (mostly those that have low contrast towards their surrounding objects, like those sweat drops on the two ladies) are very blurry.
That could be aWarpSharp's and/or DeHalo_Alpha's fault. I never used the latter too much, but given the former's warping many small details will be removed. Look further below.
2.) Can't seem to remove those ghosts seen in my post here (http://forums.animesuki.com/showthread.php?p=2211033#post2211033) , maybe somebody else can guide me, what settings to use (I suck at fiddling with filter options...)
You have the source, so you will be able to get the best results, unless you upload a sample so that others can have a play with it. But as I mentioned before, you don't have the original, and without it problems like yours often can't be corrected.
3.) The image is a little bit blurry
Sharpen then.
4.) The colours seem to be wrong, in some place I have a too high saturation
Tweak() and Levels() are your friends, as well as Trim() if they are inconsistent.

#source
Stab()
BlindDeHalo3(PPMode=-3)
EdgeCleaner(fix=false) #or aWarpSharpDering
#crop here and do the rest

Disclaimer: I dunno what it might look like since I did it blind, but it's a general formula that I often start off with, have a play and see what you get.

1) Bifrost is for rainbow removal. If your source doesn't have rainbows, don't use it.
2) Dupped is pretty much useless. It just makes a mess. You'd be better of using Dup, or neither at all.

SHiN-gx

2009-02-13, 21:40

What you need is smartfade() (http://forum.doom9.org/showpost.php?p=1154887&postcount=12). Apply it after deinterlacing and before decimating... I think.
Wow! It worked like a charm! The only problem though is that the fade in scene looks messy and I see some "rainbowing".

http://img14.imageshack.us/img14/4585/sample1nf2.png

http://img134.imageshack.us/img134/333/sample2sb7.png

Maybe the settings was too strong?

You might want to take a look at YATTA, although if you feel this is too advanced for you TIVTC does have a hybrid switch IIRC (however the results produced are usually not very good). You'll have to read through the manuals, I don't remember using it myself so I won't be able to tell you out of memory.
Ugh... In my current level, I think I'm not ready yet for YATTA so I'll take my chance on TIVTC. I could save myself the trouble by separately encoding the OP and ED but I don't like that kind of approach. =.=

Emess

2009-02-16, 01:55

Tweak() and Levels() are your friends, as well as Trim() if they are inconsistent.ApplyRange() (http://avisynth.org/mediawiki/ApplyRange)

E~

martino

2009-02-16, 07:53

Wow! It worked like a charm! The only problem though is that the fade in scene looks messy and I see some "rainbowing".
Source screenshot please.

SHiN-gx

2009-02-16, 10:13

Source screenshot please.

Here are the source screenshots of the same scenes:

http://img150.imageshack.us/img150/6263/sample1aoz0.png

http://img5.imageshack.us/img5/4569/sample2ajw4.png

The second screenshot is less grainy than the screenshot with the smartfade()

Btw, this is where I placed smartfade()
DGDecode_mpeg2source("VTS_04_1v2.d2v", info=3)
ColorMatrix(hints=true, interlaced=true, threads=0)
tfm(order=1).SmartFade(dgm=true).tdecimate(mode=1)
crop( 0, 0, -4, 0)

LanczosResize(720,480) # Lanczos (Sharp)
#denoise

martino

2009-02-16, 10:52

You could perhaps try the other scripts posted in that thread, firstly the one with Vinverse and without nnedi (as AA). It's most likely one of those two causing the problems. If it's Vinverse's fault then you might have to end up using the original 0.1 version, which uses some rather brute methods to remove the combing (or the other script posted by Dragon -- it probably works better).

Heibi

2009-02-16, 13:43

Okay, trying to get into using AVS to do our encode and get away from the evil of using h264 to encode into .avi.

Got to the point where I need to encode - Used h264 and encoded into an mkv with 2200bit rate set for LOGH episode 13 to test my methods.
Good part - Picture quality was great
Bad parts - File size was twice the size of my normal methods 658 megs and the pan downs and sideways movement of the picture was jerky.
Programs used - Virtuadubmod 1.5.10.2 with h264.
Of course I may be doing something completely wrong.

I compared an encode I did prior using my old method and the picture movement was smooth and the quality was about the same.

Any settings I should be using in h264 for the encode? I've fallen behind on my DVD encodes because of this quandry.

Mango-chan

2009-02-16, 15:00

Heibi

2009-02-16, 15:41

>>virtualdubmod

And, this means....what? Don't be cryptic or sarcastic please.

letsza

2009-02-16, 15:48

hi i use megui to encode my Fansub
but my problem is that i must always encode my files from afx[lossless layer] to YV12
and the picture always get dark
Script:
DirectShowSource("D:\displaypic_yoichi06.avi",fps=23.976, convertFPS=true)
ConvertToYV12()
Raw Avi from AFX:
http://img222.imageshack.us/img222/4554/rawmz4.jpg
Encode with megui x264 1300bitrate:
http://img186.imageshack.us/img186/734/encodedhs6.jpg

and a second question what filters should i use for this raw [commands in avs would be good]
http://img222.imageshack.us/img222/6253/yoichima5.jpg

^^

andy_blah

2009-02-16, 15:54

Okay, trying to get into using AVS to do our encode and get away from the evil of using h264 to encode into .avi.

:rolleyes:

Bad parts - File size was twice the size of my normal methods 658 megs and the pan downs and sideways movement of the picture was jerky.

You could expect that at that bitrate, you should probably use AnimeIVTC() if your source is a DVD in Avisynth

Any settings I should be using in h264 for the encode? I've fallen behind on my DVD encodes because of this quandry

Dup() or Dupped() in Avisynth and one of the h.264's profiles in MeGui should work out nicely with a bitrate of 500-700

martino

2009-02-16, 17:32

And, this means....what? Don't be cryptic or sarcastic please.
VirtualDubMod hasn't been updated for ages now, and in addition it is based on an OLD version of VirtualDub, which is currently STILL under development. There's really no reason why use Mod over plain VDub.

Heibi

2009-02-16, 20:39

VirtualDubMod hasn't been updated for ages now, and in addition it is based on an OLD version of VirtualDub, which is currently STILL under development. There's really no reason why use Mod over plain VDub.

The person I was getting help from hasn't been in contact with me (possible computer problems) regarding the encoding step, so I was using the old methods combined with the newer ones. The version of vdubmod was what he was using for a particular step. He probably wasn't using it for the encode.(I think)

As for using an older vdubmod - AVS wasn't working well with the newer version of vdub very well, plus you could only save the file as an avi. Therefore, using h264 for the encode in vdub made you use avi as the container.

checkers

2009-02-16, 21:35

When you save "mkv" with VDM, it's saving h264-vfw streams, which is functionally identical in all its failings to h264-in-avi. Please, save some kittens and help deprecate x264-vfw once and for all.

Heibi

2009-02-16, 23:26

When you save "mkv" with VDM, it's saving h264-vfw streams, which is functionally identical in all its failings to h264-in-avi. Please, save some kittens and help deprecate x264-vfw once and for all.

That's why I'm asking for help here.

ashye

2009-02-17, 00:14

Easiest way out would probably be to use the x264 CL or meGUI to encode to raw .264 or mp4 and mux it with MKVMerge to .mkv

At 2200 bitrate, you get about 700 mb for 45 mins. If you want a smaller filesize, reduce the bitrate.

andy_blah

2009-02-17, 07:58

martino

2009-02-17, 08:05

No. MeGUI is just a GUI for CLI apps, doesn't use VfW at all.

<rant>That's why people should at least mess with the CLI and not depend on any silly GUIs to find out what the heck is actually going on.</rant>

TheFluff

2009-02-17, 20:55

Dup() or Dupped() in Avisynth

don't use dup

guest0815

2009-02-18, 01:20

should at least mess with the CLI and not depend on any silly GUIs to find out what the heck is actually going on.</rant>

I encountered a problem when using the clis.
Using directshowsource or dss2 makes the dos-window pop up showing the commandline and doing nothing (but blocking the files for a few seconds until returning to the prompt) but they all do work when using MeGui. The same batchfiles and videos work as expected with the cli when using ffvideosource or avisource to load the videos.
It's the same for x264.exe and xvid_encraw.exe, the latter also shows "trying to retrieve width bla" and "avisynth detected" before stopping. The directshowfilters are ffdshow and xvid. A mencoder ffvhuff commandline doesn't care how the videos are opened in avisynth.

for example

dss2("Raw.mp4").trim(0,100) #doesn't encode via cli but megui
ffvideosource("Raw.mp4").trim(0,100) #encodes both ways

x264.exe --crf 22 --ref 8 --mixed-refs --no-fast-pskip --bframes 16 --b-pyramid --weightb --direct auto --subme 8 --trellis 1 --partitions all --8x8dct --me umh --merange 24 --threads auto --thread-input --aq-strength 0.5 --sar 1:1 --progress --no-psnr --no-ssim --output "test.mp4" "test.avs"

there's a 2pass (havent tried cq yet but I doubt that's the problem) xvid that also doesn't take dss/dss2:

xvid_encraw.exe -i "test.avs" -type 2 -o test.avi -pass1 xvid.log -bquant_ratio 100 -bquant_offset 100 -max_key_interval 250 -noclosed_gop -nochromame -nopacked -par 1 -progress 20 -quality 5 -qtype 0 -threads 3 -turbo -vhqmode 1 -imin 2 -imax 6 -bmin 2 -bmax 6 -pmin 2 -pmax 6 -zones 0,q,2,-0OC

xvid_encraw.exe -i "test.avs" -type 2 -o test.avi -pass2 xvid.log -bitrate 890 -bquant_ratio 100 -bquant_offset 100 -bvhq -chigh 20 -clow 7 -max_key_interval 250 -noclosed_gop -nopacked -par 1 -progress 20 -quality 6 -qpel -qtype 0 -threads 3 -vhqmode 4 -imin 2 -imax 6 -bmin 2 -bmax 6 -pmin 2 -pmax 6 -zones 0,w,1,-0OC

And the mencoder commandline that works with all of them:
mencoder.exe test.avs -o test.avi -of avi -nosound -ovc lavc -lavcopts vcodec=ffvhuff:vstrict=-2:pred=2:context=1

50119

2009-02-19, 06:33

don't use dup

why? :confused: I haven't seen any harm using it with threshold < 0.5 :confused:

TheFluff

2009-02-19, 06:43

why? :confused: I haven't seen any harm using it with threshold < 0.5 :confused:
then you haven't looked hard enough

(hint: in most material you want at least SOME kind of dithering/grain to make it not block like shit in flat areas, particularly dark ones, and dup really screws with that, as well as with small mouth movements and stuff like that. the gains are insignificant anyway so just don't bother)

checkers

2009-02-19, 23:20

x264 is efficient enough that it provides very little compression benefit nowadays. And I've also seen problems occur with settings of even 0.3 too, so it's Just Not Worth It.

martino

2009-02-19, 23:43

then you haven't looked hard enough

(hint: in most material you want at least SOME kind of dithering/grain to make it not block like shit in flat areas, particularly dark ones, and dup really screws with that, as well as with small mouth movements and stuff like that. the gains are insignificant anyway so just don't bother)
Uhm... Dup won't remove that grain (I know, you probably didn't mean it since you very well know that, just used bad phrasing I guess), it'll only might make it appear more stable on a still scene. I personally much rather like to see "static" grain rather than some shit that jumps around like crazy even when nothing happens in a scene. I think that for an effect like this is worth using it...

@checkers
I'd be well interested in knowing what that was... I encountered problems with using 0.3 just a single time, and that was a pan on a static scene where it moved just by one single line (stupid mahou shoujo craziness). Using 0.2 helped, but then 0.2 is so low it barely affected the other, non-problematic areas.

Dark Shikari

2009-02-19, 23:52

Uhm... Dup won't remove that grain (I know, you probably didn't mean it since you very well know that, just used bad phrasing I guess), it'll only might make it appear more stable on a still scene. I personally much rather like to see "static" grain rather than some shit that jumps around like crazy even when nothing happens in a scene. I think that for an effect like this is worth using it...IMO dfttest-like denoising + addgrain(constant=true) is better than dup for that.

TGBatman

2009-02-20, 12:11

Hi again.

I'm having some weird trouble with Evangelion Rebuild 1.0 (You're Not Alone).

The thing is, the video is reported by DGIndex as Film 99.50%, Frame structure Frame.

If I use Force Film, paning scenes and the video overall stays as it should, no interlacing, combing nor dups at most scenes. BUT...

In certain scenechanges, I get some nice combing, I was not expecting it:

http://img26.imageshack.us/img26/3871/fuckencombingir1.th.png (http://img26.imageshack.us/my.php?image=fuckencombingir1.png)

So, I wonder if I should just take note of all these scene changes, and freezeframe every combed frame? Or is there a better solution? I don't think this source needs IVTC, as Forced Film does what it should, except for this little problem.

This is my script, pretty simple:
crop(4,8,-4,-6).spline36resize(720,480)

TheFluff

2009-02-20, 20:12

if it's the R2J DVD, for some reason it isn't mastered properly. you are correct that it is indeed progressive and it really should be 100% FILM, but for some bizarre reason there seem to be missing fields at some scenechanges (they probably cut it improperly). this is why you usually don't use force film for anything less than 100% FILM, or if you do you really need to QC it carefully because you really can't trust studios to do it right.

tl;dr you need to either freezeframe every single affected scenechange (lots of effort), use YATTA (or use fieldhint() directly) to manually force a n or p match at the affected scenechanges (same effort as freezeframing pretty much) or just set dgindex to honor pulldown flags and do ivtc as usual (a lot less effort for a probably equivalent result).

TGBatman

2009-02-20, 21:06

I see, so it is badly mastered. So, doing IVTC with AnimeIVTC would suffice? I really want to avoid YATTA for an almost 2 hours movie :p.

TheFluff

2009-02-22, 09:46

animeivtc is overkill in that case, just do tfm().tdecimate(mode=1)

TGBatman

2009-02-22, 12:27

All right, thank you TheFluff. I'll use that, I hope it comes out without terrible combings.

EDIT: Worked like a charm :)

Miraino

2009-03-04, 09:29

What better way to learn than to ask. So here goes.

What is AR encoding?
If anyone has a guide or a process for this can you share, please. :)

Thought here would be the best place to ask, if not, please point me to the right place.

TheFluff

2009-03-04, 12:54

AR usually stands for Aspect Ratio, beyond that I have no clue what it might mean

martino

2009-03-04, 13:08

Maybe how the AR gets encoded in the video stream?

ie x264 --sar or something like that.

Miraino

2009-03-04, 16:53

Video: MPEG4 Video (H264) 704x480 (4:3) 23.98fps [Video]
Video Size: 704 x 528

I'm talking about how it's encoded at a specific resolution, but then the aspect ratio ends up switching the resolution. Just like above. It says 704x480 but it plays at a 704x528 resolution.

Does that make sense? If so, is there an encoding guide somewhere?

TheFluff

2009-03-04, 18:20

it's called anamorphic video and is extremely common, just google and you'll find a ton of decent explanations

Kristen

2009-03-05, 00:08

I'm not sure if this is the right place to ask, but it was the closest I could find.

How exactly can you get an AT-X transport stream audio in synch with the video? I analyzed it, and the decrypted AAC alone is 1 minute longer (26:30) than the one just playing back the stream (25:30). Is there something special about it that makes it do that? Or is there a guide someone on how to handle it? I tried researching, but I couldn't find anything about it anywhere...

neothe0ne

2009-03-05, 01:47

You ran the TS through DGIndex and its length doesn't match the AAC audio?

(I've never really dealt with TS sources but that doesn't sound right)

jfs

2009-03-05, 02:20

I'm guessing that the audio stream simply has too much data and you'll have to cut it at one of the ends or both.

Kristen

2009-03-05, 08:35

You ran the TS through DGIndex and its length doesn't match the AAC audio?

(I've never really dealt with TS sources but that doesn't sound right)

Yep, that's exactly what I did.

jfs: The audio playback starts at 0:02, and then ends at 25:25 when I play it back. The .ts starts at 0:02, and ends at 24:55. By ends, I mean that the same sound heard on playback of the .ts at 24:55 is heard at 25:25. So it's not adding 30 seconds of junk data, surprisingly.

jfs

2009-03-05, 16:42

That doesn't look like framerate problems. I was going to suggest audio sampelrate, but it wouldn't match with any common samplerates either (eg. 44100 vs. 48000). It's about 2% slowdown... well check the audio samplerate, but I don't have any further suggestions.

Kristen

2009-03-05, 17:39

From MediaInfo on the .ts itself:

General
ID : 7100
Complete name : C:\Downloads\ATX09.m2ts
Format : BDAV
Format/Info : BluRay Video
File size : 1.25 GiB
Duration : 25mn 3s
Overall bit rate : 7 154 Kbps
Maximum Overall bit rate : 26.6 Mbps

Video
ID : 4099 (0x1003)
Menu ID : 333 (0x14D)
Format : MPEG Video
Format version : Version 2
Format profile : Main@Main
Format settings, Matrix : Default
Duration : 25mn 1s
Bit rate mode : Constant
Bit rate : 6 611 Kbps
Nominal bit rate : 8 000 Kbps
Width : 720 pixels
Height : 480 pixels
Display aspect ratio : 4/3
Frame rate : 29.970 fps
Standard : NTSC
Colorimetry : 4:2:0
Scan type : Interlaced
Scan order : Top Field First
Bits/(Pixel*Frame) : 0.772

Audio
ID : 4355 (0x1103)
Menu ID : 333 (0x14D)
Format : AAC
Format/Info : Advanced Audio Codec
Format version : Version 2
Format profile : LC
Muxing mode : ADTS
Duration : 25mn 3s
Bit rate mode : Constant
Bit rate : 251 Kbps
Channel(s) : 2 channels
Sampling rate : 48.0 KHz
Resolution : 16 bits
Video delay : -277ms

When extracted:

General
Complete name : C:\Downloads\ATX09 AAC PID 1103 DELAY 93ms.aac
Format : ADTS
Format/Info : Audio Data Transport Stream
File size : 45.9 MiB
Duration : 25mn 30s
Overall bit rate : 251 Kbps

Audio
Format : AAC
Format/Info : Advanced Audio Codec
Format version : Version 2
Format profile : LC
Muxing mode : ADTS
Bit rate mode : Constant
Bit rate : 251 Kbps
Channel(s) : 2 channels
Sampling rate : 48.0 KHz
Resolution : 16 bits

Aporia

2009-04-05, 21:46

Can an avisynth script read video directly off a DVD?

Meltingice

2009-04-05, 21:56

No, you need to rip it first with DVDDecrypter.

jfs

2009-04-05, 22:03

Aporia

2009-04-05, 22:12

Yes and no... if the DVD doesn't have CSS coding etc. you can index the VOB files on it directly with DVD2AVI or DGIndex, but you probably don't want to do that for performance reasons.
If it's because you don't have harddrive space to copy the files over to, you have a much bigger problem to solve first.

Nope, disk space isn't the issue. I want to be able to give the script to other people who have the DVD and let them run it without having to rip it. How serious are those performance reasons you speak of?

guest0815

2009-04-06, 01:04

Can a 1440x810 source somewhow be successfully deblocked with some cropping or resizing to a mod8 resolution of the same aspect ratio like 1280x720 or is there no hope for deblocking in avisynth :( ? Deblock() is doing *something* but every deblocker says to not crop or resize beforehand and I can't tell if it's smoothing in all the wrong places without knowing how a mod8 source would have looked like, so...

thanks

jfs

2009-04-06, 09:25

Nope, disk space isn't the issue. I want to be able to give the script to other people who have the DVD and let them run it without having to rip it. How serious are those performance reasons you speak of?

I don't know, haven't tried, only guessing.
But every way of accessing DVD video through Avisynth (that I know of) requires the source to be indexed beforehand using a separate program, meaning you can never get a fully one-click script. Also, the index files contain the full path and filename of the source VOB files, meaning that if someone who has their DVD drive with a different drive letter than you, they'd have to fix up the index file. (And needless to say, an index file is made for one specific DVD.)

jofori89

2009-04-07, 09:05

DirectShowSource("D:\My documents\Anime\TV Series\_UnComplete\Rozen Maiden\Rozen Maiden Episode 09 (Dual).mkv",audio=true,fps=23.976,convertfps=true)
Textsub("D:\My documents\Anime\TV Series\_UnComplete\Rozen Maiden\[Vi-Ani] Rozen Maden (subtitles Eng & Vie chapters and fonts)\Vietnamese Subtittles\[Vi-Ani] Rozen Maiden Episode 09.ass")
LanczosResize(880,480) # Lanczos (Sharp)

logo = imagesource("D:\My documents\Anime\TV Series\_UnComplete\Rozen Maiden\Credit pics\Credit pic 01.png", pixel_type="RGB32", end=120)
logo = audiodub(logo, blankclip(last, length=120))

logo ++ last

There is some error, I can't insert picture before video start. When i play avs file by MPC, has error info "Splice: Video formats don't match, line 09"

How to can i fix this error?

martino

2009-04-07, 09:08

I think you're trying to put together an RGB32 and YV12 clip. You should add ConverToYV12() for the logo... also set the framerate of the logo to be the same as of the main clip with AssumeFPS(last) or using ImageSource if it has an argument for it. Otherwise, are they of the same resolution?

jofori89

2009-04-07, 09:25

DirectShowSource("D:\My documents\Anime\TV Series\_UnComplete\Rozen Maiden\Rozen Maiden Episode 09 (Dual).mkv",audio=true,fps=23.976,convertfps=true)
Textsub("D:\My documents\Anime\TV Series\_UnComplete\Rozen Maiden\[Vi-Ani] Rozen Maden (subtitles Eng & Vie chapters and fonts)\Vietnamese Subtittles\[Vi-Ani] Rozen Maiden Episode 09.ass")
LanczosResize(880,480) # Lanczos (Sharp)

logo = imagesource("D:\My documents\Anime\TV Series\_UnComplete\Rozen Maiden\Credit pics\Credit pic 01.png", pixel_type="RGB32", end=120)
logo = audiodub(logo, blankclip(last, length=120))

insertsign(last, logo, 0, 120)
#logo ++ last
Like this it run.
They are of the same resolution (880x480)
Avisynth says there is error in this line logo ++ last

logo = imagesource("D:\My documents\Anime\TV Series\_UnComplete\Rozen Maiden\Credit pics\Credit pic 01.png", pixel_type="RGB32", end=120).converttoyv12().assumefps(last)
Like this, logo ran but insertsign doesn't run

martino

2009-04-07, 09:34

Uhmmm... what about my suggestion? Did it work or not?

for the lazy
DirectShowSource("D:\My documents\Anime\TV Series\_UnComplete\Rozen Maiden\Rozen Maiden Episode 09 (Dual).mkv",audio=true,fps=23.976,convertfps=true)
Textsub("D:\My documents\Anime\TV Series\_UnComplete\Rozen Maiden\[Vi-Ani] Rozen Maden (subtitles Eng & Vie chapters and fonts)\Vietnamese Subtittles\[Vi-Ani] Rozen Maiden Episode 09.ass")
LanczosResize(880,480) # Lanczos (Sharp)

logo = imagesource("D:\My documents\Anime\TV Series\_UnComplete\Rozen Maiden\Credit pics\Credit pic 01.png", pixel_type="RGB32", end=120, fps=23.976).ConvertToYV12()
logo = audiodub(logo, blankclip(last, length=120))

logo ++ last
EDIT: insertsign() takes a picture and not a clip.

jofori89

2009-04-08, 03:53

How to can i define location to overlay picture on screen? (Same pos(x,y) ?)

TheFluff

2009-04-08, 03:54

How to can i define location to overlay picture on screen? (Same pos(x,y) ?)

Overlay(clip, clip overlay, int ''x'', int ''y'', clip ''mask'', float ''opacity'', string ''mode'', bool ''greymask'', string ''output'', bool ''ignore_conditional'', bool ''pc_range'')
int ''x'', int ''y''

rtfm

jofori89

2009-04-08, 10:08

DirectShowSource("D:\My documents\Anime\TV Series\_UnComplete\Rozen Maiden\Rozen Maiden Episode 09 (Dual).mkv",audio=true,fps=23.976,convertfps=true)
Textsub("D:\My documents\Anime\TV Series\_UnComplete\Rozen Maiden\[Vi-Ani] Rozen Maden (subtitles Eng & Vie chapters and fonts)\Vietnamese Subtittles\[Vi-Ani] Rozen Maiden Episode 09.ass")
LanczosResize(880,480) # Lanczos (Sharp)

logo = imagesource("D:\My documents\Anime\TV Series\_UnComplete\Rozen Maiden\Credit pics\Credit pic 01.png", pixel_type="RGB32", end=120, fps=23.976).ConvertToYV12()
overlay(last, logo, int "320", int "200", clip ''mask'', float ''opacity'', string ''mode'', bool ''greymask'', string ''output'', bool ''ignore_conditional'', bool ''pc_range'', 0, 120)

It not work, please tell me?

martino

2009-04-08, 10:19

I really, and really suggest you read this (http://avisynth.org/mediawiki/AviSynth_Syntax) first.

You're just copy-pasting without understanding how it works, that's not the way to get around.

jofori89

2009-04-08, 19:43

Sorry, demo i actual don't understand it mean.

Orochi^

2009-04-15, 09:26

Hello, everyone :]

I'm having some encoding trouble which is driving me crazy with an anime here. I tried a lot of things, like encoding without any filter, encoding with some filters, encoding with the Insane x264 profile [0.6 fps D:], and the result is always the same. When I open the .avs in vdub, it looks good. Not perfect, but, good. Then, when I encode it, the output comes out really terrible.

Here is some comparison screenshots:

http://img254.imageshack.us/img254/4462/vdub.th.png (http://img254.imageshack.us/my.php?image=vdub.png)
.avs opened in vdub

http://img140.imageshack.us/img140/7379/x264.th.png (http://img140.imageshack.us/my.php?image=x264.png)
encoded

http://img24.imageshack.us/img24/5991/vdub2.th.png (http://img24.imageshack.us/my.php?image=vdub2.png)
.avs opened in vdub

http://img18.imageshack.us/img18/711/x2642.th.png (http://img18.imageshack.us/my.php?image=x2642.png)
encoded

Of course, that's only in dark scenes. Bright ones are always perfect D:

I understand that the vdub opened avs is compressionless, so it naturally should look better, but... The encoded one is just AWFUL D:

So, this is why I'm here, asking for the pros some help. I rather do this than releasing something that might make everyone blind D: I never studied a lot x264 encoding, since the default profiles and some simple filters usually did the job pretty well. But, this one... :\

Here's the avs script:

LoadPlugin("C:\x264\fft3dfilter.dll")
DGDecode_mpeg2source("air.d2v")
ColorMatrix(hints=false, interlaced=true, threads=0)
tfm(order=1).tdecimate()
crop( 4, 0, -4, 0)
Lanczos4Resize(848,480)
fft3dfilter()
gradfun2db()
Trim(6,0)

And here's the x264 settings I used to those screens encode:

"C:\x264\x264.exe" --pass 2 --bitrate 1150 --stats "air.stats" --ref 12 --mixed-refs --bframes 16 --b-pyramid --direct auto --deblock 1:1 --subme 8 --trellis 2 --psy-rd 0.6:0 --partitions all --8x8dct --vbv-maxrate 17500 --me umh --threads auto --thread-input --aq-mode 0 --progress --no-psnr --no-ssim --output "air.mkv" "air.avs"

Thank you so much, any help will be hugely appreciated :}

Scab

2009-04-15, 10:56

x264 fails to retain the grain you add with gradfun2b. Re-enable AQ and tweak psy-rd. Adding --no-dct-decimate can help too. Note that when encoding to a target bitrate these are all likely to raise your quants noticeably. Consider running some --crf tests to see the effects properly.

Orochi^

2009-04-15, 16:57

TheRyuu

2009-04-15, 18:28

Thank you for the answer, Scab. I tried out those settings you adviced me, but, it's still bad. Just a little better, but, not even close to what we see on vdub. Even with --crf 1 the result remains bad. So I assume I'd need to filter the .avs in someway that x264 keeps up the input as close as possible.

Any ideas?

Try gradfun2dbmod.
This introduces grain in areas that are prone to banding instead of just dithering it.

Also in order for x264 to keep said grain, start with stuff like --psy-rd 0.8:0.2 --aq-strength 0.8.
Adjust as necessary. You should definitely NOT disable aq.

And as scab said, run a --crf 18/21 encode to see what bitrate you might need to achive good results.

Edit:
Furthermore, fft3dfilter() on defaults (sigma=2) is quite strong (this is also a factor which introduces banding).
I would also suggest doing anamorphic encoding (if it's a dvd) instead of resizing it to 848x480, but that's just a minor suggestion.

Dark Shikari

2009-04-15, 18:33

If you can see any visual quality loss at CRF 1, there's something wrong in your decoder or processing chain, not x264.

Quarkboy

2009-04-15, 18:55

If you can see any visual quality loss at CRF 1, there's something wrong in your decoder or processing chain, not x264.

Yeah, I think I agree... This may be a colorspace conversion issue in you h.264 decoding chain using the codecs on your system versus the uncompressed video decoding chain that is used when loading the .avs

Try taking one of your encoded files and loading it in avisynth with Directshowsource and playing that .avs...

Orochi^

2009-04-16, 18:16

Thank you, everyone.

In fact, opening with avs's DirectShowSource showed a better image, but, still not good. Then I adjusted the other settings you adviced me, changed to gradfun2gbmod, changed some other minor settings, and it got really better.

Thank you A LOT for the help, you all saved me :}

Anything you need (if I can help), PM me or something.

Thanks again! :]

SHiN-gx

2009-04-18, 00:14

TheRyuu

2009-04-18, 00:37

Quarkboy

2009-04-18, 00:59

I have an AVI clip of an old anime with 29.976 fps (not ripped by me) and it's riddled with what you call "ghost" artifacts. Is there a way to remove those artifacts? It's a different story compared to decimating it using the original source.

Also I append three clips into one AVI file and each clip has different set of filters, etc. and I ended up with this:

AviSource("op.avi").crop(6, 0, 0, 0).unblend(threshold=5, dthresh=0.50, show=false).lanczos4resize(640, 480) + AviSource("ep85.avi").crop(8,2,-2,-2).Levels(0, 1.3, 255, 0, 255).Tweak(0.0, 1.0, 3.0, 1.0).Stab().unblend(threshold=5, dthresh=0.50, show=false).lanczos4resize(640, 480) + AviSource("ed.avi").crop(6, 0, 0, 0).unblend(threshold=5, dthresh=0.50, show=false).lanczos4resize(640, 480)
Is there a way to simplify this?

Thanks.

You're SOL, generally. Unblending-type filters only really work in really pristine conditions, which old anime does not have.

Normally the blending that you'd get is because of the deinterlacing/poor IVTC that was done, probably in the ripping step, and older shows often have horribly composited scenes where, say, the background has a different cadence than the foreground and true IVTC is basically impossible.
If this happens and is blend deinterlaced but not decimated than you could get a pretty horribly blended rip, with zero way to recreate original frames outside of like, manually redrawing each frame. Which is probably what you have.
Frankly, I'd try hard to find the original source, probably DVD somewhere. That's the only way you'll really be able to improve the quality.

SHiN-gx

2009-04-18, 01:05

^ Bummer. =/

Thanks for your extra info and I guess I'll go cry in a corner, deal what I have or look for the source. :P

comatose

2009-04-18, 05:55

Actually, I found that in many shows, there's minor field blending - so you have a frame that's blended followed by a frame that isn't (because anime has dupes, hooray!)... so in many cases FreezeFrame-ing is a decent solution that works for everything but pans.

Obviously though, it's very frustrating and boring as hell.

TheFluff

2009-04-18, 06:19

If the blended frame/clean frame pattern is consistent (haha yeah right) YATTA has some semi-automated ways to freezeframe stuff one scene at a time or so.

Quarkboy

2009-04-18, 06:58

jofori89

2009-04-18, 09:33

Overlay(clip, clip overlay, int ''x'', int ''y'', clip ''mask'', float ''opacity'', string ''mode'', bool ''greymask'', string ''output'', bool ''ignore_conditional'', bool ''pc_range'')
int ''x'', int ''y''

rtfm

Overlay(last, logo, 600, 80, clip ''mask'', float ''opacity'', string ''mode'', bool ''greymask'', string ''output'', bool ''ignore_conditional'', bool ''pc_range'')
What dose it mean?

50119

2009-04-18, 10:02

TheRyuu

2009-04-18, 21:07

Overlay(last, logo, 600, 80, clip ''mask'', float ''opacity'', string ''mode'', bool ''greymask'', string ''output'', bool ''ignore_conditional'', bool ''pc_range'')
What dose it mean?

Read the 'fine' manual. </stolen>
http://avisynth.org/mediawiki/Overlay

and

I really, and really suggest you read this (http://avisynth.org/mediawiki/AviSynth_Syntax) first.

You're just copy-pasting without understanding how it works, that's not the way to get around.

jofori89

2009-04-19, 19:25

Read the 'fine' manual. </stolen>
http://avisynth.org/mediawiki/Overlay

and

Thanks a lot, :D

comatose

2009-04-20, 10:24

Yeah, that's great, except it'll take in insane amount of time and has to basically be done manually.
It doesn't take that long with YATTA (I manually wrote a script once with a shorthand for FreezeFrame, and it did take me ages)... if you do it together with IVTC, then I found that (guessing) 95% of the time it's picked up as V difference from both sides - 20. You can also go to section starts and ends, because it happens there when two scenes are blended.

If the blended frame/clean frame pattern is consistent (haha yeah right) YATTA has some semi-automated ways to freezeframe stuff one scene at a time or so.
:( Can you point me in the direction, just in case? :P

TheFluff

2009-04-20, 16:54

Rightclick -> set pattern -> freezeframing; n replaces with next, p replaces with previous, o does nothing. Set the pattern you desire, then scroll through the video and use the t/u keys (try pattern/use pattern) as appropriate.

comatose

2009-04-21, 08:58

Oh, yay. Thanks :)

TGBatman

2009-04-21, 21:39

Hi guys.

Some days ago, I bought Saber Marionette J Again DVDs, because I wanted to play with something old. But man, isn't this UGLY...

First, I have been testing deinterlacing methods, and none of them convince me, but of course, I have almost no experience with really interlaced sources.

Also, the image shakes a lot, as if in drugs. I haven't tested any deshaker, but I'd like to know first a way to properly deinterlace this. I hope you can help me clean this :(

Here, I uploaded some demuxes from dgindex. (http://www.mediafire.com/?sharekey=d4f8f27e5ff886be7069484bded33bcdf1fb0b2f c25e394c)

martino

2009-04-22, 06:02

For shaky footage try this:
temp = last.TemporalSoften(7,255,255,25,2)
Interleave(temp.Repair(last.TemporalSoften(1,255,2 55,25,2)),last)
DePan(last,data=DePanEstimate(last,trust=0,dxmax=1 0,dymax=10),offset=-1)
SelectEvery(2,0)Needs DePan, DePanEstimate and Repair (RemoveGrain package) plugins. I probably won't have the time myself to look at the rest so you might have to wait until somebody else takes a proper look at them and helps you with your issues.

mad_cracker

2009-04-22, 06:33

Hello everybody.
I'm recently having some problems loading an avs inside another avs, for example:
src = DirectShowSource("c:\workpath\Episode.avs")
ovl = AviSource("c:\workpath\ovleray.avi")
Overlay(src,ovl.ConvertToYV12(),mask=ovl.ShowAlpha ())

The error it's like: Avisynth read error: DirectShowSource: The video Grapgh failed to restart after seeking. Status = 0x80004005

This happened after I formated my hard drive. Before it, this code worked perfectly.
I guess that before formating I had Avisynth 2.5.7 instead 2.5.8.
Don't know what to do anymore. I've tryed so many things. Please help.
Thanks

Notes: The AVS I'm trying to load contains a mkv h264 video and I deactivated the filters. The second AVS only contais the code I mentioned before because I removed the filters too.

martino

2009-04-22, 06:56

You load an already existing script with Import(). This goes both for a script that contains just some function or a video clip. So...

src = Import("c:\workpath\Episode.avs")

mad_cracker

2009-04-22, 07:05

It works fine now. Thank you very much.
Just curious now, why did it work back then with DirectShowSource and now doesn't? Does it have anything to do with Avisynth versions?

Thanks again.

martino

2009-04-22, 07:10

I think it more has to do with DirectShow on your system. Maybe some of your filters aren't registered properly so it can't build the filter graph correctly. My knowledge of DirectShow is close to non-existent, so I wouldn't be able to tell you much more. You could try asking on Doom9 or hope that somebody here with some knowledge will be able to tell you more.

TGBatman

2009-04-22, 11:31

Thanks martino, i'll give a look to that script :)

EDIT: Well, so far the shaking has been almost completely eliminated. This is the script I'm using.

LoadPlugin("C:\yatta_6-129\PLUGINS\fft3dfilter.dll")
MPEG2Source("C:\SMJADVD1\FullDisc\SABERJ_AGAIN_D1\project\VTS_0 6_1.d2v")

temp = last.TemporalSoften(7,255,255,25,2)
Interleave(temp.Repair(last.TemporalSoften(1,255,2 55,25,2)),last)
DePan(last,data=DePanEstimate(last,trust=0,dxmax=1 0,dymax=10),offset=-1)
SelectEvery(2,0)
tcomb()
fft3dfilter(sigma=4,sigma2=4,sigma3=13,sigma4=4,bt =4,plane=1)
fft3dfilter(sigma=4,sigma2=4,sigma3=13,sigma4=4,bt =4,plane=2)
dfttest(sigma=0.4)
aaa(720,480)
awarpsharp(10,1,0.5,1).toon(0.2)
spline16resize(640,480,8,2,-8,-2)

And these are the comparisons, left is untouched, right is using the script.

http://img9.imageshack.us/img9/7974/1before.th.png (http://img9.imageshack.us/my.php?image=1before.png)http://img9.imageshack.us/img9/6677/1after.th.png (http://img9.imageshack.us/my.php?image=1after.png)

http://img12.imageshack.us/img12/9315/2before.th.png (http://img12.imageshack.us/my.php?image=2before.png)http://img12.imageshack.us/img12/54/2after.th.png (http://img12.imageshack.us/my.php?image=2after.png)

http://img12.imageshack.us/img12/2664/3before.th.png (http://img12.imageshack.us/my.php?image=3before.png)http://img9.imageshack.us/img9/2671/3after.th.png (http://img9.imageshack.us/my.php?image=3after.png)

I can notice a lot of banding and haloing (that I think blinddehalo3 can get rid of), and not even gradfun2dbmod gets rid of the banding. If I don't use AAA, it goes away, but if I don't use AAA, I'll be left with some horrible aliasing.

Any sugestions?

Tofusensei

2009-04-22, 13:27

Had an interesting glitch yesterday.

mp3 in an avi file was borked. While it would load in the .avs with directshowsource (it errored on AVISource), it would not play any audio if I began playback/encode from frame 0. However, if I advanced it a little bit and played, the audio would play.

When I'd go to do "save wav" in virtualdub, it would save an empty .wav file, even if I tried clipping/saving from a frame higher than 0 to start from. Same thing if I tried to extract out with ffmpeg. Same thing if I tried to open the .avi or .avs in a Sound editor. I couldn't even get the audio out using graphedit.

However, if I opened the .avi directly in Virtualdub, I was able to extract the audio out via "save wave" but there were these error messages:

[!] AVI: Stream 1 (audio) has a non-zero start position of 2 samples (+52 ms).
VirtualDub does not currently support a non-zero start time and the stream
will be interpreted as starting from zero.

[!] AVI: Truncated or invalid MP3 audio format detected (18 bytes, should be
30). Attempting to fix.

[!] AVI: A non-zero wBitsPerSample value was detected in the MP3 audio format
structure. This has been corrected in-memory to zero for better
compatibility with commonly installed MP3 audio codecs.

[!] AVI: Variable bitrate (VBR) audio detected. VBR audio in AVI is
non-standard and you may encounter sync errors up to 0ms when attempting
to extract WAV files or playing in some players. If this is a problem, use
Full Processing mode is recommended to decompress or recompress the audio.
(This warning can be disabled in Options, Preferences, AVI. bitrate: 256.0
± 0.1 kbps)

--I've been using Virtualdub on a regular basis since about the year 2000 and Avisynth since the project began and I've never had an audio stream that was as difficult to process as this one.

Can anyone tell me what was going on here?

comatose

2009-04-22, 22:10

Well, uh, the late hour is preventing me from really reading that post, but... did you try muxing to mkv then demuxing to mp3 and seeing if it works in a standalone player, like foobar2000 (which you can also use to decode to WAV)?

Tofusensei

2009-04-23, 01:58

comatose

2009-04-23, 07:29

Wait, what? What does that have to do with muxing to mkv? I suggested that because sometimes mkvmerge manages to avoid screwing up when demuxing from nasty files, so it might help :\

Tofusensei

2009-04-23, 10:12

Wait, what? What does that have to do with muxing to mkv? I suggested that because sometimes mkvmerge manages to avoid screwing up when demuxing from nasty files, so it might help :\

You can mux to mkv in graphedit, too. And you never mentioned mkvmerge in your post :D

But yes, I tried to mux to mkv in graphedit as well as mkvmerge and it didn't change a thing.

I just want to know what was wrong with it and how it got so messed up. I was able to salvage it.

comatose

2009-04-23, 13:23

Well, if I had to guess I'd say it was corrupted?

jnazh

2009-04-28, 00:04

Hi
I have 1 question
How to clean raw through Avisynth??
can I do that??

jfs

2009-04-28, 10:24

Hi
I have 1 question
How to clean raw through Avisynth??
can I do that??

By "clean" you probably mean "remove noise", "perform inverse telecine", "remove rainbowing" and doing various other things to video that can make it more watchable or more compressible.

What filters you actually should use depends on what kinds of problems you want to try fixing in the video and how the video looks in general. There is no one "miracle cure".

But yes, Avisynth "can be used", but you need to know how to use it, you need to understand how it works and you need to understand what you really mean by "clean".

jnazh

2009-04-28, 11:30

We in the Arab world to use Avisynth
in production and other things.:rolleyes:

One of them said that we can purify raw through Avisynth
Or, as you said
"remove noise", "perform inverse telecine", "remove rainbowing"

But I did not understand good when you say:confused:
you need to understand how it works and you need to understand what you really mean by "clean".

As you said Avisynth "can be used", And it also depends on the video as .. right?
I want more information on this matter ,But in any case thank you jfs :).

That1GuyTim

2009-04-29, 10:23

Hey guys, I've got a problem that I need some help with. I couldn't find any answers elsewhere, so I'm turning to y'all.

Last night I started to run a lossless in VirtualDub and everything was going fine. Cut to right now and I have this error message that has popped up and caused the encode to stop prematurely by 2 frames:
http://i615.photobucket.com/albums/tt231/TheLanguid/AviSynthError.jpg

Now, I clicked OK and it still gave me a video, but it's missing the very last 2 frames that the original source had. Any idea what this error means or is causing it?

Meltingice

2009-04-29, 12:27

It is most likely a filter conflict/issue.

That1GuyTim

2009-04-29, 12:52

But why would it happen with only 2 more frames left?

Meltingice

2009-04-29, 15:36

edogawaconan

2009-05-02, 22:57

how about telling us what's your avisynth plugins folder contents, and the .avs causing that error.

That1GuyTim

2009-05-07, 10:36

Nevermind, I believe it was a problem with the raw I was using. I used a different raw with the same avs and same settings and it worked just fine.

SHiN-gx

2009-06-24, 23:10

I need some suggestions on how to remove this artifact (sorry, don't know what it's called.)

http://img141.imageshack.us/img141/4447/ttt12.th.png (http://img141.imageshack.us/i/ttt12.png/)

I used FFT3D and fluxsmoothst() but it really didn't do much.

Thanks in advanced.

Akatsuker

2009-06-24, 23:24

"AVISource: couldn't locate a decompressor for fourcc WVC1"

My AvsP keeps telling me that, even when I configure my FFDShow codec to decode it, whatever, libavcodec or wmv9. I couldn't find anything about a good decoder for this. It's Microsoft.

Directshowsource opens the file, but since I heard that it could be a little frame inaccurate, I ask you:

- "DirectShowSource() is not totally accurate"...? Is it true?
- Being it or not, what do you guys do (or would do...) when you was in that trouble? Any suggestions?

P.s.: I've got this message after I was manipulating some MKV vfr raw.

Thanks a lot.

Zycle

2009-06-25, 15:37

SHiN-gx, try gradfun2db.

SHiN-gx

2009-06-27, 11:56

SHiN-gx, try gradfun2db.

Hot dang, this worked like a charm! :D Thanks!:kisskiss:

EDIT:

I'm kinda stuck here. I can't seem to get the desired output. I can't get the ending credits for Candy Boy episode 7 to scroll smoothly (source is DVD). I used two methods namely tritical's tdecimate and animeivtc() to make a VFR encode but none of them worked.

I know I'm doing this wrong but I don't know what I'm missing...

K!R4

2009-06-27, 19:01

Scab

2009-06-28, 02:07

x264 can retain the grain pretty well thanks to AQ/psy-rd at very reasonable bitrates, but yes, gradfun was designed as a postprocessing filter. Unless CCCP has started using it by default hardly anyone has it activated on playback though, so it's really not such a ass backwards idea to apply it before encoding if you know how to retain the effect.

TheFluff

2009-06-28, 20:08

Hot dang, this worked like a charm! :D Thanks!:kisskiss:

EDIT:

I'm kinda stuck here. I can't seem to get the desired output. I can't get the ending credits for Candy Boy episode 7 to scroll smoothly (source is DVD). I used two methods namely tritical's tdecimate and animeivtc() to make a VFR encode but none of them worked.

I know I'm doing this wrong but I don't know what I'm missing...

if they were anything like the credits in an earlier ep I looked at a while ago, it's one of those cases where you get to pick if you want to make the background or the credits jerky; there is no way to get both smooth

Yumi`

2009-06-29, 20:27

if they were anything like the credits in an earlier ep I looked at a while ago, it's one of those cases where you get to pick if you want to make the background or the credits jerky; there is no way to get both smooth

Somebobber().p60to24p()
Might make the credits too blurry though.

TGBatman

2009-07-04, 18:46

Kazu-kun

2009-07-04, 21:24

Hi there guys. I have a question, regarding cropping and stuff.

I've read once that it was OK to crop 8 pixels from X and -X always when using DVD sources, but I never really found why. I've been doing it since then supossing it was true, as I've read it from this topic I think.

I mean, I've my source at 720x480, without any black borders on the side, but somewhere I've read that the 8 pixels at the right/left were not used, so I always crop 8 pixels at x and 8 pixels at -x, and it ends being 704x480.

Is it true? If so, why?

No, it isn't. You only crop if the picture has black borders on the side. If it doesn't, then of course leaving the full 720x480 is the way to go.

The whole "8 pixels at the right/left are not used" thing only applies when they're not really used (aka, when the picture has black borders on the side).

To sum it up, some DVDs use 704 pixels (has black borders) and some use the full 720 pixels (doesn't have black borders). You can tell just by looking.

TGBatman

2009-07-04, 22:10

Oh crap, what have I been doing until now :(.

Thanks for the answer, I think I might have missread the whole think.

Scab

2009-07-05, 06:48

You're not wrong, Kazu-kun is. You crop to 704x480 to get closer to the AR of 710.85x486, i.e. the real active picture size, while staying mod16. If you don't, your encodes will have a slight AR error. Alternatively, you could of course keep it 720x480 and AR-flag the stream accordingly to compensate, i.e. 8640/4739 if your source is 16/9 or 6480/4739 if your source is 4/3.

Read this (http://lipas.uwasa.fi/~f76998/video/conversion/).

Kazu-kun

2009-07-05, 10:13

Alternatively, you could of course keep it 720x480 and AR-flag the stream accordingly to compensate, i.e. 8640/4739 if your source is 16/9 or 6480/4739 if your source is 4/3.

He didn't say anything about resizing so I assumed he was going to use AR flags to beging with. And it really makes no sense to crop if that the case.

TGBatman

2009-07-05, 10:33

Wow, that's a really complete guide, but a bit hard to understand.

Well, yeah, I tend to use the AR flags in Matroska and try to avoid resizing when it's not needed, so I guess cropping at the sides when there's no blackbars at all is pointless.

But I didn't get the "8640/4739 if your source is 16/9 or 6480/4739 if your source is 4/3." part. Aren't those some BIG numbers there? I tend to check the AR error ratio with Yatta, 853/480 seems to be the one with less of it if I crop the black bars at the sides, and 872/480 seems to be the one with less AR error if I don't crop them (when not needed, of course). And that's supossing I don't have black bars in Y and -Y.

I better always check the AR error in Yatta from now on :p.

Also, another question. If I have a 4:3 DVD NTSC source, what's the best idea? Not doing ANY resize in avisynth (only the needed cropping if any), setting the display size in Matroska as 640x480, or is a better idea to use 768x576, OR resizing at the desired resolution?

I've always done the third one (cropping and resizing, and that's to 640x480), but I'd like to know your opinion.

Thanks for the help guys.

Scab

2009-07-05, 13:54

They are big to avoid decimals. If you don't do any cropping, they're correct.

Not resizing is always the best idea if possible; resizers are nothing but bad for quality. Upscaling just a few pixels has a negative effect on the full frame and downscaling obviously kills resolution. That goes for your 4:3 content as well. Ideally, just do a crop() to get rid of the edge padding and AR flag the stream accordingly. DAR can be confusing and requires recalculation based on how much you've cropped, so unless you absolutely want to flag it in the container, consider using --sar in the x264 command line instead. It flags the pixel aspect ratio of your clip which stays the same no matter how you've cropped it.

TGBatman

2009-07-06, 00:23

It flags the pixel aspect ratio of your clip which stays the same no matter how you've cropped it.

So, does this means that even if I crop whatever number of pixels at each side of the image (staying mod16 of course), 8640/4739 for 16:9 or 6480/4739 for 4:3 stays like that?

And, what if I need to crop 4 pixels in Y, but not in -Y? 476 is not mod16, the next mod16 number would be 464 I think, and that's too much crop. I have one of those. :heh:

Again, thanks for the help and sorry for so much questions.

Scab

2009-07-06, 02:20

8640/4739 and 6480/4739 are DAR numbers. IIRC the SAR for 4:3 NTSC is 10/11 (704*(10/11)=640) and 40/33 for 16:9. And yes, those stay the same as long as you don't resize.

If you can't crop to mod16 you either overcrop (good) or resize (bad; watch the AR error %). If you feel adventurous I suppose you could crop to a non-mod16 resolution; x264 can handle it just fine but playback results may be unpredictable.

TGBatman

2009-07-06, 10:48

Well, I'd like to stay mod16. I guess it depends on the case, sometimes overcropping might be a good idea, sometimes not. Guess I can't have the Holy Grail huh. :p

Thanks for the help guys, have a nice day. :)

Zergrinch

2009-07-06, 11:24

This may sound like a retarded question, but bear with me. I'm still inexperienced :p

I'm in the process of ripping a DVD with meGUI, which is in 720x480 resolution.

I want to merge this with AviSynth to a video that's in 640x480 resolution.

Question - do I need to resize the video to match the DVD [e.g. Lanczos4Resize(720,480) ] or can I just merge as is? Suppose I don't merge, but use ordered chapters in Matroska instead - will that change the answer?

I ask with the understanding that meGUI automatically passes on an aspect ratio to x264, telling it to dynamically use 4:3 (or 640x480 resolution) for the NTSC DVD.

Thanks in advance :)

jfs

2009-07-06, 12:43

Zergrinch, read the discussion that has JUST been had in this very thread, it's an AR issue you're having.

@TGBatman: Nobody cares about the outer 16 pixels on either edge of the picture, the animation is generally made to take overscan into account and thus there won't be any interesting picture that near to the edges, you can safely just crop 8 pixels at every edge and not resize at all. (Or only crop at left/right edges if the picture touches the top and bottom perfectly all the way through; but it's ugly if some sections have small black pixel lines near the top or bottom, IMO.)

Zergrinch

2009-07-06, 20:04

I did read it, and my decision was to rip the DVD as is and let the --sar option take care of aspect ratios.

But I'm looking for something slightly different. I want to get a 720x480 set to 4:3 AR to play with a 640x480 set to 1:1. Can I do that or do I have to resize said 640x480 first?

jfs

2009-07-07, 07:20

No if you have multiple segments with different pixel resolutions you must make them all have the same pixel resolution.

Kristen

2009-07-07, 21:57

TGBatman

2009-07-07, 22:24

@jfs

That's good to know, I'll try to crop 8 pixels from each side next time, as long as I don't lose important detail and stay mod16, I'm ok with it.

SHiN-gx

2009-07-13, 11:54

I thought gradfun2db sole purpose is for post processing (during playback), not for preprocessing (encode) because this filter wont work unless you have a higher bitrate for your encode.

To see what Im talking about: encode your video with this filter then once its done play it on mediaplayer classic. While its playing, check ffdshow video decoder below your right taskbar (there are two of these: the red and blue icons) right click the red one, it will list some filters. you will see gradfun and it has check mark, heres a interesting part of this filter: turn off gradfun and you will see the real outcome of your encode. Either your encode will look shit (banding are back on your encode) or your encode will look good because you have higher bitrate you used in your encode to compensate.
Figured that. Well, I need some tips on how to remove the banding in my encodes. Here's a different sample:

Original:

http://img20.imageshack.us/img20/6920/sample1i.png

With Tweak() and Levels() in Preview:

http://img515.imageshack.us/img515/2255/sample2.png

Original's quality is not so good. You can still see banding and it became worse if I add in Tweak and Levels. I'm looking for a filter to remove or lessen the banding.

Thanks.

fireshark

2009-07-13, 12:35

...it's called gradfun2db. Or try gradfunkmirror if that suits you. either way you have to throw more bitrate at it or it's not going to help no matter how much gradfun you do.

SHiN-gx

2009-07-13, 17:23

...it's called gradfun2db. Or try gradfunkmirror if that suits you. either way you have to throw more bitrate at it or it's not going to help no matter how much gradfun you do.
Already did that but it's for post-processing only. I want to actually remove the banding in my encodes.

TheFluff

2009-07-14, 02:14

it works exactly the same when encoding as it does when postprocessing, the problem is to get the encoder to keep the nice smooth gradients and not re-introduce the banding; usually this is only a problem in dark scenes

one solution is to add a very small amount of grain (think addgrainc(1.0) or even less) and encode with a small amount of psy-rd.

TheFluff

2009-07-14, 02:16

Recently, about 1/8 h.264 encode come out for me with fine video and audio to start with. But at a select point in the encode, the video will change to the following screen:
http://i27.tinypic.com/2ignojm.jpg
And it will remain that way for the rest of the encode.

My avisynth script is usually:

AVISource("C:\Lossless.avi")
Textsub("C:\subs.ass")

And when I run the encode a second time without changing anything, it encodes perfectly.

Does anyone know what's going wrong so I don't have to waste 3-6 hours?

Odd issue, especially since it's not reliably reproducible, but the general solution to odd Avisynth crashes is to empty out the plugin autoload directory and only leave the dll's that ship with Avisynth in there, and load the plugins you need manually instead.

andy_blah

2009-07-14, 09:45

I am having again the same blended frames problem as I previously explained here (http://forums.animesuki.com/showpost.php?p=2211033&postcount=258), the only difference now is that this is from the actual interlaced source (DVD). I tried a few deinterlacers such as Yadif, TDeint, TempGaussMC, LeakKernelDeint, all with correct settings, and all seem to produce those blending artefacts. Is there any solution for this problem? If needed I can give some samples of the source.

jfs

2009-07-14, 09:52

First tell what it looks like if you SeparateFields() and otherwise do not process the source at all. Do the individual fields contain blending? If so, it's impossible to save.

andy_blah

2009-07-14, 10:06

do not process the source at all
By this, you mean to not include any filters?

Tried it anyways and it seems that unfortunately the blending is still there :(

Raislin

2009-07-14, 10:49

This might be a rather broad question. I'm trying to collect myself a number of quality plugins for AviSynth and I'm wondering what some people here would consider essentials.

Ha, I phrased my question as a statement. Anyways, suggestions are welcome of course. I'm not really in a rush, though, because I'm lacking some anime sources to work with at the moment.

jfs

2009-07-14, 11:17

By this, you mean to not include any filters?

Tried it anyways and it seems that unfortunately the blending is still there :(

Yes, following this idea for the complete Avisynth script:

mpeg2source("myfile.d2v")
separatefields()

(Switch mpeg2source for whatever the actual source filter you use is.)

If you see blending in that, your source is impossible to fix, except with ten thousand hours in mspaint.

andy_blah

2009-07-14, 11:37

Tried without any filters as you suggested above, and unfortunately the blended artefacts are still there. But then again it is fortunate that the DVD doesn't have this problem all the time ^_^

If you see blending in that, your source is impossible to fix, except with ten thousand hours in mspaint.

Haha, yes, I can imagine that, would be a pain for the PC, and a even greater pain for me :heh:

Raislin

2009-07-14, 14:43

Alrighty, I just finished playing around with AviSynth and an anime raw for the first time. Cut out commercials, made some adjustments here and there. I've started my encode to h.264 and the length of the first pass is estimated at ~20 hours. I've no idea how long the second pass will take.

This is my first time doing more than what's automated in MeGUI in AviSynth. Does it typically increase the encode time this much? Thus far, I've typically just encoded my movies that I've backed up, without much AVS. I usually get great quality and the length of the encoding is about the same as what I'm experiencing now. But that's with feature length films, not a 24 minute anime.

Anyways, It's been a long time since I've done any encoding at all, and I'm diving into AVS this time as well as expanding what I know about the various x264 settings (which, admittedly, isn't nearly as much as I'd like) and trying not to bury myself in the process.

I hate it when I do this. I typed all of this out and now I forget where I was going with it. Anyway, is this encoding speed typical? I probably should have just encoded a 30 second clip of it, but I really don't think it's going to turn out poorly.

Dark Shikari

2009-07-14, 23:46

Alrighty, I just finished playing around with AviSynth and an anime raw for the first time. Cut out commercials, made some adjustments here and there. I've started my encode to h.264 and the length of the first pass is estimated at ~20 hours. I've no idea how long the second pass will take.

This is my first time doing more than what's automated in MeGUI in AviSynth. Does it typically increase the encode time this much? Thus far, I've typically just encoded my movies that I've backed up, without much AVS. I usually get great quality and the length of the encoding is about the same as what I'm experiencing now. But that's with feature length films, not a 24 minute anime.

Anyways, It's been a long time since I've done any encoding at all, and I'm diving into AVS this time as well as expanding what I know about the various x264 settings (which, admittedly, isn't nearly as much as I'd like) and trying not to bury myself in the process.

I hate it when I do this. I typed all of this out and now I forget where I was going with it. Anyway, is this encoding speed typical?Did you pick absurdly slow settings, or are you encoding on a Pentium 2?

Raislin

2009-07-15, 05:41

I did pick slow settings, but I didn't think they'd be this slow. Ha. I've been learning about what the settings in x264 do, but not much about how they affect encoding speed.

Anyways, I was just wondering if a non-trivial AviSynth script normally affects the encoding speed much and whether or not it's worth it.

Quarkboy

2009-07-15, 07:12

Did you pick absurdly slow settings, or are you encoding on a Pentium 2?

Off topic, but I'm curious: What's the crappiest processor you know of that's been used to encode with x264?

Could it run on a 386?

Kristen

2009-07-15, 17:45

Odd issue, especially since it's not reliably reproducible, but the general solution to odd Avisynth crashes is to empty out the plugin autoload directory and only leave the dll's that ship with Avisynth in there, and load the plugins you need manually instead.

That's the weird thing. The only dll in there that don't ship with it are ffmpegsouce and tivtc. I don't think it's likely they're causing the problem, since they're common usage, but I'll try it.

Dark Shikari

2009-07-17, 16:24

Off topic, but I'm curious: What's the crappiest processor you know of that's been used to encode with x264?

Could it run on a 386?x264 requires i686 or later unless you --disable-asm, in which case it can probably run on basically anything; the only requirements we specify are two's complement arithmetic and 8-bit bytes, which pretty much means any CPU from the past 25 years. sizeof(int) also has to be at least 4.

We're working on optimizing for ARM now, so you'll be able to run x264 on your iPhone 3GS.

TrickyDream

2009-07-26, 22:18

I was reading those posts about SAR and DAR.
Info:
DVD Source
16/9 - 720x480
In order to keep mod16 resX&resY I crop 8 at each border so I get 704x464. (I have black borders everywhere but less than 16 total)

I thought you could just encode setting sar to 1:1 and then mux it setting "Aspect Ratio" to 16/9 (mkvmerge).

Now, as someone (Scab) said, I'm trying to make use of the x264 sar param. But I'm not sure how to do it, I think the correct SAR is 32:27. I took it from the "ARS Calculator", 704:464 as PFS and 16:9 as MAR.
I played the new file and the VOB file, and it looks pretty equal, mine looks a little bigger but that's because it has been cropped by 8 at each border.

So, did I do something wrong?

Kazu-kun

2009-07-26, 23:22

I was reading those posts about SAR and DAR.
Info:
DVD Source
16/9 - 720x480
In order to keep mod16 resX&resY I crop 8 at each border so I get 704x464. (I have black borders everywhere but less than 16 total)
So, did I do something wrong?

The SAR should be 40:33, not 32:27. It doesn't matter how much you cropped, because SAR does not specify the aspect ratio, it specifies the factor to be stretched by.

EDIT: I just want to be sure that you understand you're aiming to have the content of the picture displayed in its correct proportion (not deformed), not the picture itself. It doesn't matter if the picture isn't displayed exactly at 16/9. In fact, it won't, because you cropped at the top and bottom of the picture. But again, it doesn't matter.

EDIT: Geez, I messed up (that's what I get for writting at 4 AM ><)

TrickyDream

2009-07-27, 07:27

Ups, actually it has black borders at the top, left and right. I cropped from the bottom to get the mod16 at resY.

Should I consider the VOB as the correct proportion? Because when I set --sar 40:33 it looks deformed (little) if I compare it to the VOB.
"ARS calc" says that 40:33 is the SAR to 704x480 at 16/9, not 720x480 (which was 32:27), I said it wrong before, for 704x464 is 116:99.

Really, I'm very confused.

I understand that what I want is the correct proportion. The picture itself is not going to be 16/9 because I cropped it, and I don't want to resize, so I don't care about it. What I'm trying is to set the correct proportion.

BTW, thanks for the explanation.

Scab

2009-07-27, 07:33

Crop as you see fit then encode with --sar 40:33 and the proportions will be correct.

TrickyDream

2009-07-27, 10:12

I was thinking about that 1.85... The DVD box says 16:9 LB (I guess it means Letterbox) but the VOB file does not show any black border at the bottom, just around 3 lines at the top.

Edit: I played the DVD with WMP and MPC and the file encoded with --sar 40:33 looks wider (deformed).

TrickyDream

2009-07-27, 11:00

Here, some screens (http://img37.imageshack.us/gal.php?g=vobc.png)

http://img37.imageshack.us/img37/4567/vobc.png (http://img37.imageshack.us/i/vobc.png/)

http://img34.imageshack.us/img34/8767/3227.png (http://img34.imageshack.us/i/3227.png/)

http://img188.imageshack.us/img188/1084/4033.png (http://img188.imageshack.us/i/4033.png/)

http://img252.imageshack.us/img252/8695/12699.png
(the image says 126:99, it's a typo) (http://img252.imageshack.us/i/12699.png/)

To appreciate the difference, I recommend to download and open them with the image viewer.

Kazu-kun

2009-07-27, 11:20

Here, some
40:33 is the correct one.

TrickyDream

2009-07-27, 11:48

Ok.. and do I need to set something on the mux process?

Scab

2009-07-27, 11:50

Not if you encoded the video track with --sar, no.

TrickyDream

2009-07-27, 13:24

I'm sorry but I really want to understand.
I don't know why it's 40:33.

DVD 720x480, 16:9, no crop:
to have a correct Aspect Ratio:
--sar 1:1 + resize 720x480 > 853x480 or 873x480?
or
--sar 40:33 + noresize (source 720x480), then I get 873x480 = It is not 16/9

cropping 8 at X and -X:
704x480
--sar 40:33 + noresize (source 704x480), then I get 853x480 = It is 16/9

But If I crop 8, -8, 8, -8, the source is:
704x464
--sar 40:33 + noresize (source 704x464), then I get 853x464 ?
The image looks wider because it expands the difference of 16 lines of the X-axis without expanding the 16 of the Y-axis.

I mean, if the source was 704x480, 40:33 seems to be OK but not for 720x480.
Or what I have to do is not a 16/9 AR in anime but 1.81875, or just the video should be treated as 704x480.

It's only a guessed theory. This is what I see, if you could explain why is it wrong, I would appreciate it.

Thanks.

TheFluff

2009-07-27, 15:44

Why are you using SAR at all, is your target container something that doesn't support a DAR header? Getting the SAR right usually involves a lot more maths than getting the DAR right, and the end result is the same, so why bother?

I mean, if the source was 704x480, 40:33 seems to be OK but not for 720x480.
It actually IS 704x480, not 720x480. Or, at least it's closer to 704 than to 720. Read http://lipas.uwasa.fi/~f76998/video/conversion/ for some interesting revelations.

Scab

2009-07-27, 15:49

Assuming you don't resize, SAR requires no math. If you do resize, DAR makes a lot more sense.

TrickyDream

2009-07-27, 16:05

Yes, I do. MKV as container. If I'm not using --sar in x264 what should I set in the mux app?
DAR = 853x480? or 16:9?

Edit: I wrote 852x480 insted of 853x480.

Kazu-kun

2009-07-27, 16:17

Yes, I do. MKV as container. If I'm not using --sar in x264 what should I set in the mux app?
DAR = 853x480? or 16:9?

Edit: I wrote 852x480 insted of 853x480.
No, if you're using the AR flag on MKV, you should set it at 853x464. 853x480 or 16:9 will screw the proportions because you cropped at the top and bottom.

But Scab is right. SAR is pretty simple, and you should always use 40:33, even if you cropped the picture. As long as you don't resize, it's always 40:33.

TrickyDream

2009-07-27, 16:28

Thanks a lot. =)

Kazu-kun

2009-07-27, 16:52

I still wanted to reply to this. Maybe it'll help you to understand this thing.

I'm sorry but I really want to understand.
I don't know why it's 40:33.

DVD 720x480, 16:9, no crop:
to have a correct Aspect Ratio:
--sar 40:33 + noresize (source 720x480), then I get 873x480 = It is not 16/9

It's because what matters is the proportion of the drawing inside the picture, not the size the picture is displayed at. The DVD, according to the ITU standard, assumes an active frame of 710.85×480 (that's the "real" picture inside your 720x480 picture). That's the picture that should be the base of any AR calculation (no matter how much you cropped), but we take 704 instead of 710.85 because the latter isn't mod16. So, your active frame is 704x480.

Now do the math:

40:33 = 1,2121212121212121212121212121212.

1,2121212121212121212121212121212 x 704 = 853,33333333333333333333333333333

So you get 853,33333333333333333333333333333 x 480, rounded to 853x480 which is 16/9.

Now, even if you have the full 720 pixels with content (no black bar on the sides), you still use 40:33. It will result in the picture size of approximately 873x480 when displayed, but it doesn't matter because the proportion of the content will be correct. This is because the DVD was designed according to the ITU standard, with an active frame of 710.85 pixels in mind, even if they put more content on the sides.

But If I crop 8, -8, 8, -8, the source is:
704x464
--sar 40:33 + noresize (source 704x464), then I get 853x464 ?

Yes, you get a different size when displayed (853x464 instead of 853x480), but the proportion of the content inside the picture is still the correct one, and that's what matters.

So there, always use 40:33 and you'll never go wrong. For 4/3 DVDs use 10:11 btw.

TrickyDream

2009-07-27, 17:06

Ok, now it is clear. Thanks again. =)

Daiz

2009-07-28, 04:47

...Or just use 16:9 as the DAR because no-one will ever notice an AR distortion of 1-2% or less. That way it won't have black borders when watched on a 16:9 screen either.

Zergrinch

2009-07-29, 12:04

Just curious about this. Is it preferable to set the relevant aspect ratio options in x264 when encoding, or is it better to do this when muxing into Matroska?

twc

2009-07-29, 12:26

Set your SAR in x264. This way, the aspect ratio is guaranteed to always be correct regardless of how much you crop and you don't have to calculate a "display resolution" (lol) or "display aspect" (lolol).

Scab

2009-07-29, 13:03

Makes no difference. You can't set SAR in Matroska so that flag would go in the video track. DAR would be the other way around; can't be set in x264 so would go in the Matroska container.

MoonLight-

2009-08-02, 19:41

Hello,,

dunno if this is the right place to ask or not.

anyway.. any clue how to install mkvmerge on site, so I can merge files through the site?

my ul speed sucks.

thanks :)

jofori89

2009-09-01, 09:07

I try using sangnom and deen and some other filter to deinterlace video from DVD but it looked no good, have some grain, dot and rainbow...

DGDecode_mpeg2source("E:\My documents\Anime\RAW\[Raws]_Clamp_School_Detectives\[DVD_extract] Clamp School Detectives OVA - One day with The Clamp School Detectives\One day with The Clamp School Detectives 08.d2v", info=3)
ColorMatrix(hints=true, threads=0, interlaced=true)
crop(6, 0, -6, 0)

sangnom(1, 30)
deen("a3d", 2, 5, 7)

#Interleave(SangNom(1,15), SangNom(0,15))
#LeakKernelBob(order=1)#if TFF
#tomsmocomp(0,2,0)

LanczosResize(640, 480)

removegrain(mode=3)
TemporalCleaner()
#deblock(quant=25)
undot_undot

Anyone have any idea to fix it, please?

twc

2009-09-01, 14:07

Problems:

1) ColorMatix doesn't do anything
2) SangNom is a stupid way to deinterlace
3) Deen is a stupid way to denoise
4) RemoveGrain, LanczosResize, TemporalCleaner, and UnDot are also stupid.

Try:

1) Using a comb filter
2) Deinterlacing with Decomb v5 by Donald Graft
3) Using one denoiser after deinterlacing that is not Deen

TheFluff

2009-09-01, 17:18

jofori89

2009-09-01, 19:54

decomb is not primarily a deinterlacer package even if it does contain functions to do that; if you meant inverse telecine, say inverse telecine, and in that case use tivtc, not decomb

if you actually meant deinterlacing, try yadif (fast) or some nnedi-based motion adaptive bobber (very slow); either is much better than fielddeinterlace() (which is what decomb provides)
Thank you, i'll use yadif with temporalCleaner.
But it have some aliasing, what antialiasing fillter can i use, maa, ediaa or antialiasing.avsi very slow T_T ?

DGDecode_mpeg2source("E:\My documents\Anime\RAW\[Raws]_Clamp_School_Detectives\[DVD_extract] Clamp School Detectives OVA - One day with The Clamp School Detectives\One day with The Clamp School Detectives 08.d2v", info=3)
ColorMatrix(hints=true, threads=0, interlaced=true)
crop(6, 0, -6, 0)
Load_Stdcall_plugin("C:\Program Files\AviSynth 2.5\plugins\yadif.dll")

Yadif(mode=1)
denoise(lx=350,ty=370,rx=400,by=430)
removegrain(mode=3)
TemporalCleaner(5, 10)

LanczosResize(640, 480)
undot_undot

jfs

2009-09-02, 02:26

jofori89

2009-09-02, 02:39

Are you sure you're doing it right?
Interlaced video and interlacing is not the same thing (http://www.mod16.org/hurfdurf/?p=12).
Almost all anime is telecined, and that isn't interlacing that should be de-interlaced but a process that should be reversed.
Thank all guys, i tryed a lot of filters, some looked very well

Telecide()
Decimate(cycle=5,mode=1,threshold=50)

removegrain(mode=3)
TemporalCleaner(5, 10)

Spline64Resize(640,480)
undot_undot

TGBatman

2009-09-03, 09:19

Hi guys.

I see you've been talking about interlaced/telecined material, so I guess it's the perfect time to ask about this.

I have a BD source of .Hack//G.U Trilogy, and using DGAVCIndex, it gets reported as "Frame structure: Field (TFF)".

Now, first thing I noted, the BD is not 1080, looks like an upscale from 720.

Second thing I note, there aren't any combing problems at all, and that's using honor pulldown flags, so it looks like a progressive 29.970 stream.

Third thing is, there are some frames where some lines are flickering.

Nnedi with field=1 does nothing actually.

So, my question is this: Can a BD be encoded as interlaced when having progressive frames too (being that DVDs can)? If so, can I get away with just resizing and encoding? If needed, I'll upload a sample later.

TheFluff

2009-09-03, 17:49

bd's have the exact same interlacing/progressive/ivtc/hybrid options available to them as dvd's do, so yes it could be that it's actually 29.97fps progressive and the retards at the studio coded it all as interlaced for some reason, who knows

TGBatman

2009-09-03, 19:55

Thanks for the answer TheFluff, it looks like it is 29.97 fps progressive, but it seems I was quite wrong here.

Now that I see it with "hardware deinterlacer" on in CoreAVC (or using DXVA with MPCHC internal decoder in overlay mixer), it LOOKS as if it's not an upscale, completely 1080.

Yet, if I pause the video, it looks aliased. I guess that's why it is played at 60 fps, it's doing some sort of bob.

I'm uploading a sample, because I'm working for the first time with content like this, and I really need help here. I'd like to know if it's possible to retain 1080p without having to use bob. CoreAVC hardware deinterlacer is great, but it doesn't seem to work when using dss2. And sadly, I don't own an nvidia :(.

Here, a sample:

Part 1 (http://www.mediafire.com/?ww3meymqxdo)
Part 2 (http://www.mediafire.com/?1i3wzwonmnk)

Sorry for it being more than 100 mb, and I had to split it for mediafire to let me upload it.

If it is interlaced, there should be some combing, no? Hell, I hate interlacing...

twc

2009-09-09, 17:44

Do RFF/TFF flags even exist on BDs? Every report of hybrid BDs I've heard of mentions that the 24 fps segments are coded as interlaced/telecined.

@Batman: This looks pretty progressive to me. Of course, since you uploaded a retarded .264 file, there's no way to determine what the framerate is/was/will be.

Edit: There's also garbage on the first few frames. Did you remove an I-frame?

Edit2: Bobbing effect from setting "DEINTERLACE=BLOB" is probably from aliasing created by duplicating frames, throwing away half the fields, and then guessing them.

TGBatman

2009-09-09, 17:57

I've cut it with DGAVCIndex, didn't think about that when I did it. The frame rate is 29.97 fps and there are no dups. Don't know about the I-frame, shouldn't dgavcindex skip to an I-frame when seeking?

The only interlacing I can see is at the credits, where it shows combing in every frame if it's not deinterlaced, but the entire movie looks like that sample.

twc

2009-09-09, 17:59

Well, if it doesn't look interlaced, there's a 99.99% chance that it is in fact not interlaced.

TGBatman

2009-09-09, 21:03

Thanks for the answer, I didn't know if the fact of it being encoded as interlaced would affect encoding in some way, or if there was an obscure reason for it being like that.

Then, I'll just do a resize to 1280x720, as it seems to be an upscale. I really don't get why they didn't make this movie at real 1080p, it would have been great.

TheFluff

2009-09-10, 06:14

Do RFF/TFF flags even exist on BDs? Every report of hybrid BDs I've heard of mentions that the 24 fps segments are coded as interlaced/telecined.

In theory they most certainly do exist on BD's too, but no, I've never actually seen a hybrid-flagged BD in the real world either.

Sub-zero 4ever

2009-10-07, 22:01

clip=avisource("clip.avi").converttorgb32()
img=imagereader("01.png", pixel_type="rgb32", fps=23.976)
layer(clip, img)

isnt there a way to return the frame-rate of another clip in the same script (assuming that clip has CFR), for example

fps=clip.fps
(same way as accessing struct's members like in C)

and regarding Layer().. the problem is that the thing it can overlay pngs with alpha which i couldnt do with overlay(), but it is costy (converttorgb32())

isnt there another way to deal with images contain alpha pixels in pngs that costs less?

and sorry for my bad english btw.

jfs

2009-10-07, 22:20

fps=clip.fps

fps=clip.framerate

Avisynth actually has a decent manual, looking this up took less than a minute.

twc

2009-10-07, 23:14

Sub-zero 4ever, you can use Overlay(mask=clip) in combination with ShowAlpha() to process in YUV. Another thing you could do is to make the alpha clip using ShowAlpha() and then perform the overlay using mt_merge() from MaskTools2, which will guarantee full YV12 or YUY2 processing.

Zergrinch

2009-10-13, 01:35

I'm consistently getting an AVISynth error which goes like this:

CAVIStreamSynth: System exception - Integer Divide by Zero at 0x4ed5010

This error only happens to a specific AVI file I'm encoding. I have tried this on two different computers, with the same results. Other similar AVI files I've encoded using the same script and X264 codec settings don't have any problems, so it's limited only to this file. The error only happens on a specific frame - at the 7:58.591 mark (I assume this is frame 11485 at 23.997fps)

Searching for similar problems give a recommendation to add the following to the script (from Doom9 forums):
setmemorymax(20)
converttoyv12()

Unfortunately, the same error occurred. I am using the latest version of meGUI with Avisynth 2.58 build Dec 22 2008.

The contents of my script are as follows:

DirectShowSource("D:\Episode 13\[Cooking.Master.Boy][13][DVDRip][XviD.MP3](B36A64D4).avi")
setmemorymax(20)
converttoyv12()
dfttest(sigma=0.4)
limitedsharpenfaster(strength=30)
TextSub("D:\Episode 13\Episode 13 - Hardsubs.ass")

I have tried AVISource and DSS2, to no avail. I am currently going to try re-encoding with selective disabling of the last three commands, but am also interested in whatever recommendations I can get. If it will help, I have uploaded the input files, as well as the output file with the error, on my webspace (http://chuuka.jetsam.org/media/output/).

Thanks!

salty

2009-10-13, 02:31

you got a corrupted avi. redownload it. you said, its the only avi among all the (batch of) avi's you encode with the same avisynth script that gives this error. the frame you specified is causing the error, its corrupted. other than that... its a memory leak. check this one http://avisynth.org/mediawiki/FAQ_common_errors#When_frameserving_I_got_the_foll owing_message:_.22CAVIStreamSynth:_System_exceptio n_-_Access_Violation_at_0x0.2C_reading_from_0x0.22.3F on a way to fix it. hth

Zergrinch

2009-10-13, 03:33

Well, the memory leak was in response to the error saying "Access violation at 0x0, reading from 0x0", and I already specified SetMemoryMax at the beginning of the script. This one is a bit different - searching for division by zero problems was not very informative, honestly.

I have tried to encode without dfftest and limitedsharpenfaster enabled, with the same result. I will try one last time, without TextSub, before calling the AVI corrupt. The reason I'm reluctant to write it off as corrupt is because I have been able to encode this properly in the past.

salty

2009-10-13, 04:43

i only suggested you got a corrupted avi since someone in here (http://forums.animesuki.com/showthread.php?p=2368545#post2368545) have already experienced your problem but he never disclosed whats the exact problem he had and only said there was a problem with the raw video he was using, so i only inferred that he might had a corrupted file and yours might experiencing the same. well, i hope someone can help you with your problem.

Zergrinch

2009-10-13, 06:19

Okay. I tried re-encoding with just TextSub disabled.

No problems at all!

This is odd. It's the first time I've ever have an Aegisub-generated ASS file play havoc with AviSynth...

jfs

2009-10-13, 17:46

Okay. I tried re-encoding with just TextSub disabled.

No problems at all!

This is odd. It's the first time I've ever have an Aegisub-generated ASS file play havoc with AviSynth...

Care to share? I'd like to see if it could be a problem with the file, a bug in VSFilter (no I'm not going to fix any) or what.

Zergrinch

2009-10-13, 22:38

I've gone through the ASS file in question, and saw a typo which caused this error.

Instead of \blur30, it came out as \blur300.

This was more than enough to crash AviSynth :twitch:

Changing it to a more reasonable value like 10 completely fixed the problem.

Sub-zero 4ever

2009-11-12, 10:30

so like... FFVideoSource() can rip the timecodes out of any file it can open? if yes then should it give precise results? because i tried to extract the timecodes out of VOB file and it gave me strange times, for the clip it was about 19 minutes long and the timecodes were between 0 to 1780 with some duplicates of some times like
0.0000
0.0000
0.0000
0.6446
1.1544
1.1544
etc...

also seeking the video through the same function is totally missed up (in the same .vob). i tried the possible values for "seekmode" from -1 to 3.
am I fucking up something? because i tried the timecodes parameter on other containers (namely mkv-mp4-avi) and it worked completely and prefectly.

and all those attempts were to make sure that my source (*.vob) does not contain any softtelecined frames because if there is any the sound and video will play unsynced due to RFFs if imported through ffvideosource() (lol so n00b?)

thanks in advance

Desbreko

2009-11-12, 19:30

You should use DGIndex to index the VOB file and then load the .d2v into your AviSynth script with Mpeg2Source(). If it's fully soft-telecined, you can just use the Forced Film option in DGIndex to index it at the original 23.98 fps. If it's hard-telecined or a mix, you'll need to use the Honor Pulldown Flags option in DGIndex and then use IVTC and decimation filters in AviSynth.

The only time you'll need a timecodes file, though, is if the VOB has variable frame rate material. (I.e. a mix of 29.97 fps and telecined 23.98 fps video.) In which case you'd get the timecodes file from the decimation filter.

Emess

2009-11-12, 22:02

FFMS2 cannot handle transport streams, and that includes program streams like a VOB, at least as far as I can recall. VOB's won't have VFR marked in timecodes as they are of a constant framerate, so your shit is either telecined or full 30fps. Use mpeg2dec3/dvd2avi to index, or dgindex if you're a neuron2 supporter for some reason, and then IVTC filters.

@Desbreko: decimation is part of the IVTC process, they are not separate.

E~

Desbreko

2009-11-13, 04:01

Okay, if you want to be technical, you use field matching and decimation filters.

misteranonimous

2009-11-13, 16:52

http://hologfx.com/images/snapshot20091113214709.png

Any ideas on how I could correct the jagged lines?

I've tried using blur but then all the colour go funny, they go really bright.

I think its an upscale anyway, so I will downscale it so 720p, that should reduce the pixelization, but its still noticeable at 720p.

Any ideas?

TheFluff

2009-11-13, 17:32

http://hologfx.com/images/snapshot20091113214709.png
404

Any ideas on how I could correct the jagged lines?
antialiasing

FFMS2 cannot handle transport streams
wrong

misteranonimous

2009-11-13, 17:37

http://img.photobucket.com/albums/v478/misteranonimous/snapshot20091113214709-1.jpg

I'll try antialiasing, thx

jfs

2009-11-13, 18:13

http://img.photobucket.com/albums/v478/misteranonimous/snapshot20091113214709-1.jpg

I'll try antialiasing, thx

Try using a source that is not just an upscale.
AIR wasn't produced in HD, it was an SD production. The opening was later re-produced for HD, but that's the only thing.

D404

2009-11-14, 17:19

FFMS2 cannot handle transport streams,

wrong
Mind you FFMS2 doesn't like h.264 transport streams very much.

Zergrinch

2009-12-11, 08:36

I'm trying my hand at DVD ripping, and would like to get the best possible results that I can manage.

http://upload.jetsam.org/images//CMB-fix.png

My source has a problem with noisy edges. The picture isn't really obvious, but zooming it to full on MPC really shows the problem if you watch the area encircled in blue.

Here is a 10MB snippet from the VOB, which I cut using MKVmerge. I did no re-encoding:
http://upload.jetsam.org/movies/cmb01.mkv

Currently, my meGUI-generated Avisynth script I have consists of:

# Set DAR in encoder to 22 : 15. The following line is for automatic signalling
global MeGUI_darx = 22
global MeGUI_dary = 15

DGDecode_mpeg2source("F:\!DVD Images\!Cooking Master Boy\Disc 1\01\VTS_01_1.d2v", info=3)
ColorMatrix(hints=true, interlaced=true, threads=0)
tfm(order=1).tdecimate(hybrid=1)

#deinterlace
crop( 8, 0, -8, 0)
#denoise

TextSub("F:\!DVD Images\!Cooking Master Boy\Disc 1\01\Episode 1 - Hardsub.ass")

What combination of Avisynth filters would you gentle folks recommend to optimize this?

Scab

2009-12-11, 08:47

TComb with appropriately tweaked settings (read the included documentation) should take care of most of it. edgecleaner() can probably do a decent final clean-up job after that.

SHiN-gx

2009-12-20, 12:34

I'm having trouble fixing this:

At every scene change(?), the last 2-3 frames of the previous scene or the next scene and are always blocky. The problem is the source itself. Is there an Avisynth filter can fix this?

Example:
http://img684.imageshack.us/img684/7332/grab01875.png
http://img685.imageshack.us/img685/333/grab01876.png

EDIT: Also getting the same errors on 2 other DVD sources.

max2k

2009-12-20, 13:32

Try DeBlock or DeBlockQED. If you have a non rencoded DVD Source u could use the intern deblocking of MPEG2Source(CPU=/CPU2=).

Deblocking will smooth the Source. So if you use DeBlock or DeBlock QED and the Anime is realy worth you time, you could work with trim to filter the framechanges stronger.

TiGR

2009-12-20, 17:58

Or you could try applying the deblocking filter via mask, imagine MSCDetection.TemporalSoften. This would apply the deblocking both before the cut and after it, so it might be a better idea to duplicate the original mask and shift it by 1, 2, 3, 4 frames and then merge it into final mask with decreasing weight. I think it’s worth a try because trimming manually is only good for experienced YATTA freaks :D

(Not sure if MVTools’ scenechange detection is reliable enough with anime, though. It could just randomly deblock = blur things there and there.)

SHiN-gx

2009-12-20, 23:10

@max2k:

Setting the CPU made little difference and will try Deblock_QED() in a little why and...

you could work with trim to filter the framechanges stronger.
Please explain it to me in detail.

@TiGR:
TemporalSoften did the trick but it killed most of the details. Used lower settings but still to no avail. MSC Detection... searched that on Doom9 and it led me to RemoveDirt and led me to RemoveNoiseMC_HQ() and stopped using it because its requirements are conflicting with AnimeIVTC().

Tried DeDup() and on the default settings, it's not really helping.

TiGR

2009-12-21, 10:27

TemporalSoften did the trick but it killed most of the details. Used lower settings but still to no avail. MSC Detection... searched that on Doom9 and it led me to RemoveDirt and led me to RemoveNoiseMC_HQ() and stopped using it because its requirements are conflicting with AnimeIVTC().

You want to deblock poorly compressed frames near scenecuts but avoid unnecessary loss of detail everywhere else. In other words, you want to merge two clips, the one with deblocking applied and the one without it.

MSCDetection is an MVTools2 function that outputs a mask: scenecut frames get white, all other frames are black. This is good, but you wanted to do the filtering on more than just one frame, which means you have to (temporally) smooth the mask first, hence the dumb TemporalSoften. Imagine the luma values of the mask: 0 0 0 255 0 0 0; after softening you will get something like: 20 100 160 255 160 100 20. This means that the deblocked original clip (ie. “deblocking itself”) will fade in and out at scenecut rather than blink in for one frame.

original = last
deblocked = original.deblock()
super = original.msuper()
vectors = super.manalyse()
scenecut_mask = MSCDetection(original, vectors).temporalsoften(3,255,255,255,2)

mt_merge(original,deblocked,scenecut_mask)

max2k

2009-12-23, 02:44

@max2k:

Please explain it to me in detail.

Simple Version and this should be enough when you only need to change one filter. In your filter chain you insert something like this, instead the filter:

trim(0,278).
Deblock_QED(quant1=16, quant2=14)+trim(279,286).Deblock_QED(quant1=30, quant2=28)+trim(287,334).
Deblock_QED(quant1=16, quant2=14)+trim(335,342).Deblock_QED(quant1=30, quant2=28)+trim(343,0).
Deblock_QED(quant1=16, quant2=14)

The First trim defines the range: "trim(0,278)" form the start of your video (0) to frame 278. Then the dot and filter with the settings for this range: ".Deblock_QED(quant1=16, quant2=14)". "trim(279,286)" important the trims have to be aligned, the first segment ends with frame278 the second starts with frame 279.".Deblock_QED(quant1=30, quant2=28)" the same filter with other settings.

And so on ...

At the end "trim(343,0)" with filter and settings from frame 343 to the last frame(0)! of you video stream avisynth will using this settings.

The second way, would be making "functions" out of different filter chains, naming them and call them in the trims... Even more complicated...

If you don't use Yatta, there is still AvsP: http://avisynth.org/qwerpoi/ (http://avisynth.org/qwerpoi/). A very nice Tool to edit and compare your scripts. Should be very helpful.

I now think I really suck at making sentence in English. But I think I understand it very good so I should better shut up and read only.
p.s:
At the moment I cannot call myself “YATTA-mad”, but hope to get my ass down this holiday season and do some manual IVTC and decimating to get a nice vfr encode out of some OAD that I planed to encode for a long time. Yes I am that ignorant that I really think I could teach my self YATTA, thers no one who would tutor me anyway.

TiGR

2009-12-23, 11:24

I’d suggest using a script like this:
q1weak=16
q1strong=30
q2weak=14
q2strong=28

trim(0,279-1).Deblock_QED(quant1=q1weak, quant2=q2weak)+\
trim(279,287-1).Deblock_QED(quant1=q1strong, quant2=q2strong)+\
trim(287,335-1).Deblock_QED(quant1=q1weak, quant2=q2weak)+\
trim(335,343-1).Deblock_QED(quant1=q1strong, quant2=q2strong)+\
trim(343,0).Deblock_QED(quant1=q1weak, quant2=q2weak)
Rather than repeating parameters over and over inline, I prefer to define them as variables – you can comfortably adjust them that way (no need to search-replace all the time). As for the trimming, I’d write down the frame numbers first and then paste them into the script without doing the math manually. The horrible trim-line is easier to edit that way. For the same reason, I’d split it into multiple lines (that’s what the slash does).

[Kyuubi]Fansubs

2010-01-13, 15:18

Hey,

Im having a bit of a problem.
I've extracted the VOB files from a DVDISO and then merged them into 1 VOB file with VOBMerge, to use it in MeGUI and encode it into an mp4 video.

But the result is that the picture gets "scanlines" over the entire picture when its in motion, but when its still or not moving very fast, picture is as its supposed to.

Is there anyway of fixing this with some Avisynth filter?
And does anyone know what causes this phenonomen?

Will attach these screenshots so youll get what I mean:

Here is what it looks like in motion:
http://i46.tinypic.com/iwuqhg.png

And here is the same scene seconds later when its not moving very fast or its stillimage:
http://i48.tinypic.com/2qa750o.png

jfs

2010-01-13, 15:29

andy_blah

2010-01-13, 16:29

Open the VOB with DGIndex (http://neuron2.net/dgmpgdec/dgmpgdec.html) first..., save the project, and from there create an AviSynth script loading the Yadif (http://avisynth.org.ru/yadif/yadif.html) plugin (remember to load it with LoadCPlugin instead of LoadPlugin), then load also the DGIndex's dll file as prescribed at it's site and load it's project file with MPEG2Source("") then place .Yadif(mode=2, order=1) if the source is frame based and TFF (Top Field First, just make sure when you preview the video that you don't get jerky motion, if that happens switch to order=2 for BFF - Bottom Field First - or 0 to be automatically detected tho this isn't suggested for most cases), mode=1 is for field based (you have a double framerate than the vob's original). Basically you have to tinker with the settings until the source is properly deinterlaced. This is a clean and easy method that I prefer to use, and gets the job nicely done.

@ jfs: You always forget that there will be always a beginer needing for help, saying that doesn't help much.

jfs

2010-01-13, 19:00

Do not do what andy_blah suggests. Telecined video is not regular interlaced video, and the Yadif filter is for treating regular interlaced video, not telecined video.

TheFluff

2010-01-14, 13:49

salty

2010-01-15, 01:55

andy, i think all of your encodes are failure if you have used yadif on telecine videos. this is (http://avisynth.org/mediawiki/External_filters#IVTC_.26_Decimation) what jfs and fluff were referring to when they said ivtc filters. check tivtc first because its the one that most people use when they do ivtc.

[Kyuubi]Fansubs

2010-01-15, 15:24

The video is telecined and you need an inverse telecine (IVTC) filter to make it progressive.

Figure out the rest for yourself. (And if you don't want to spend some days learning about video theory, don't go into DVD ripping territory.)

Ah yeah, thanks alot.
Fixed the problem with Donald Graft's Decomb filter.
Slow process (took about 29 hours encode), but worth it.

Again, thanks!

Zergrinch

2010-03-09, 20:42

I'm running into some signs of overfiltering, and would like to get some feedback whether this is the case.

This scene is from the DVD:
http://upload.jetsam.org/images//Original.jpg

I added a few basic de-interlace and cropping:
ColorMatrix(hints=true, interlaced=true, threads=0)
tfm(order=1).tdecimate(hybrid=1)
crop( 6, 0, -10, 0)

To make it look like this:
http://upload.jetsam.org/images//Partial.jpg

There is an annoying dot crawl that occurs, so as per advice from here, I added a few more filters:
tcomb(mode=2)
MCTemporalDenoise(settings="high",edgeclean=true,GPU=false)

Giving me this:
http://upload.jetsam.org/images//Full.jpg

The script normally works very well. It's just this particular episode that's giving me a bit of problems, with the rainbowing and visible vertical blocks.

How best to proceed? Sample snippet from the DVD: Download (http://upload.jetsam.org/movies/Snippet.mkv)

TheFluff

2010-03-10, 03:05

That's not dotcrawl, that's deinterlacing artifacts, and they're most likely caused by tfm failing to find a good match for that frame, so it's falling back on its postprocessing mode, which means deinterlacing the frame. Try with other tfm settings (use a mode that lets it try more matches) or drop it into yatta and find out why it's failing. If that fails too just use a better deinterlacer as postprocessor, like nnedi2.

You do however have a lot of rainbowing there, try bifrost if you want to get rid of it.

this is dotcrawl, by the way:
http://mod16.org/randompics/screencaps/shit/dotcrawl-example.png

e: also, don't use colormatrix

Desbreko

2010-03-10, 05:00

TFM is deinterlacing because there's tons of dotcrawl which it sees as combing:

http://img191.imageshack.us/img191/5592/sourcer.png

Nothing I tried managed to get rid of it, though.

Zergrinch

2010-03-10, 05:43

Hmm, I've been stupidly applying deinterlacing BEFORE dot crawl removal. I'll reverse the process, add Bifrost, and report back.

@TheFluff: ColorMatrix was automatically put in by MeGUI, and I assume it's some kind of filter for DVD sources. What's the best filter to replace this with?

Thanks.

Edit: Downloaded YATTA 7.13b6 from http://ivtc.org/. Documentation is.... underwhelming. :(
Eh, this tool is probably above my skill level, I'll likely just live with having one ugly episode if I can't fix it with notepad.

Source:
http://upload.jetsam.org/images//Original.jpg

Old settings:
http://upload.jetsam.org/images//Full.jpg

New settings (modified order and deinterlace settings):
http://upload.jetsam.org/images//Episode%2036%20B.mkv_snapshot_11.48_%5B2010.03.11_ 09.07.27%5D.jpg
Basically changed the field order from 1 to 0 in tfm, and added Bifrost().

Oh well, I guess there's no way to get rid of all the artifacts. Whack the mole on the rainbowing of the blond guy's shirt, brand new artifacts appear on the front guy's red shirt...

I am going to blend the frames and see what happens :p

Desbreko

2010-03-10, 20:38

What are you using to take your screens? The source frames you've posted and what I saw in AvsP of the sample you posted look very different. Yours look like they've been blend deinterlaced.

Zergrinch

2010-03-10, 22:48

MPC-HC screenshots for all...

I suspect I might have some sort of postprocessing deband filter active. Probably gradfundb.

TheFluff

2010-03-11, 05:30

the dshow mpeg2 decoder included with cccp defaults to blend deinterlacing
if you want accurate source screenshots use dgmpegdec or mpeg2dec3 in avisynth and screenshot it in vdub or something

Zergrinch

2010-03-11, 12:46

Tried two tweaked settings:

http://upload.jetsam.org/images//Episode%2036-muxed.mkv_snapshot_11.48_%5B2010.03.12_01.41.22%5D .jpg
tfm(order=0,mode=5,PP=0).tdecimate(mode=1)

http://upload.jetsam.org/images//Episode%2036%20(Alt)-muxed.mkv_snapshot_11.48_%5B2010.03.12_01.42.17%5D .jpg
tfm(order=0,mode=5,PP=5,field=1).tdecimate(mode=1)

Short of doing it frame by frame a la YATTA, I think the second one is the best result I can hope for.

Thanks for the feedback, guys.

Desbreko

2010-03-11, 18:47

If you just use blend deinterlacing to smooth out the dotcrawl, you might want to set cthresh=-1 so that TFM() will deinterlace every frame. That will avoid flickering between frames that got deinterlaced and ones that didn't, as well as make sure it catches all the dotcrawl.

martino

2010-03-11, 21:20

FFT3DFilter is also good against rainbowing, you'll just have to search around for the specific settings, if you can't get good results with Bifrost.

Vendonis

2010-04-07, 11:02

Hy guys,
I have the following Problem.
I'm unable to load mkv or mp4 files with avysynth, through DirectShowSource.
The Thing is that this Situation is only since yesterday, before that it worked fine.
Now when i try to load a Mp4 with a script like this:
DirectShowSource("D:\[Raws-4U] Kaichou wa Maid-sama! - 01 (TBS 1280x720 H.264 AAC Chap).mp4", fps=23.976, convertfps=true).ConvertToYV12()
I get no Audio and depending on what VirtualDub version i use, the Video plays(1.9.0) or not(1.9.8).
When i try to load an avi via Directshowsource it works fine.
The mp4 and mkv doesn't have a Problem with mpc either.
The Things i have done until now:
Reinstalling CCCP
Trying to reset the Settings of CCCP
Reinstalling CoreAVC and the AACparserfilter.
The Problem appeared out of the blue, i changed nothing, neither rebooting the system or installing Something, my system runs 24/7^^.
I hope you can help me because with the current Situation i'm unable to work on the Projects of my group.

//edit
Sorry for my bad english. I'm not a native speaker^^

salty

2010-04-07, 20:14

DirectShowSource("D:\[Raws-4U] Kaichou wa Maid-sama! - 01 (TBS 1280x720 H.264 AAC Chap).mp4", fps=23.976, convertfps=true, audio=true)

hth.

Vendonis

2010-04-08, 00:28

Nope that doesn't work.
In Fact that was the first thing i tried^^
But it seems like i must clarify something.
It seems that the AAC loading isn't working not only since yesterday,
rather it seems like it hasn't worked since i upgraded(new installation)the Win7 Rc1 to the RTM.
I have come to this conclusion, because the files that worked yesterday contained vobis audio, and that works.
And the other thing is, i talked to our other encoder in the group and the aac loading doesn't work for him either, but he hasn't noticed that because he normaly exctracts the audio from the raw and convert it with an external Program.
So it seems it's a Problem with the Windows 7 rtm build.
The non Playing video in the new Vd was caused by the audio problem, because if i load the raw with parameter Audio=false. The Video plays.

newfag

2010-05-19, 22:00

What's the most common method for deinterlacing transport streams in avisynth amongst raw cappers?

Sorry for bumping a thread that's over a month old, but it was only half way down the first page so meh.

And to the post above me despite that I'm sure you've moved on by now... run a filter tweaker and set your AAC filters to ffdshow.

Meltingice

2010-05-19, 22:05

http://avisynth.org.ru/docs/english/externalfilters/tivtc.htm

newfag

2010-05-19, 22:18

http://avisynth.org.ru/docs/english/externalfilters/tivtc.htm

Thanks, I'll give it a try. Sleep first.

EDIT: Did exactly what I wanted. Thank you. :)

Kristen

2010-09-02, 16:25

Here's a problem I've had, but gotten around, and is coming up again.

I have a Blu-Ray. It's 24 bit PCM audio, 2.0. I want to trim it, and then convert it to FLAC.

I load it into eac3to with

eac3to "C:\Downloads\S1\BDMV\Stream\00000 PID 1000 delay 0ms.pcm" "C:\Downloads\S1\BDMV\Stream\00000 PID 1000 delay 0ms.wav" -2 -24 -big

And I get the WAV file. But when I load it into avisynth (WAVSource("C:\Downloads\S1\BDMV\Stream\00000 PID 1000 delay 0ms.wav") I get the error "No compatible ACM codec to decode 0xFFFE audio stream to PCM (03.avs, line 2)"

I usually get that WAV file, put it in audacity, and export it, and then it works. But I can only find how to get 16 bit output. I want 24 bit. How can I do this without having to resort to mkvmerge?

TheFluff

2010-09-03, 05:44

For some reason eac3to always sets the AudioCompression field of the fmt chunk of the WAV file to FFFE, which means "use the WAVEFORMATEXTENSIBLE header instead". I suspect it does this because WAVEFORMATEXTENSIBLE supports complex channel layouts (while the old fmt chunk doesn't), and madshi was too lazy to make a special case for plain ol' stereo. Naturally, to make your life more interesting, the old ACM model used by VirtualDub and Avisynth doesn't support WAVEFORMATEXTENSIBLE.

As for the solution, either find a program that supports WAVEFORMATEXTENSIBLE and can save 24-bit stereo WAV's using the old WAVEFORMATEX/fmt chunk header (sox might work), or take the interesting route: hex edit the WAV file produced by eac3to, find the fmt chunk, go to offset 0x04 relative to its start (if the file starts with RIFF, 4 bytes, WAVE and then "fmt ", it should be 20 bytes from the file start), and you should find two bytes with the hex values FE and FF (in that order, remember the endianness). Change them to 01 and 00, respectively, save the file, and see what happens. It may work, or then it may not. See also https://ccrma.stanford.edu/courses/422/projects/WaveFormat/

Edit: or find a wav source plugin for avisynth that supports WAVEFORMATEXTENSIBLE. I think there might be one; at least I know there is one that supports w64.

Edit edit: JEEB pointed out to me that eac3to supports W64 too, so ignore all of the above, just write to that instead and use either the W64 source plugin whose name escapes me at the moment or use ffaudiosource.

Edit edit edit: or you could let eac3to write FLAC and do the cutting with split_aud.pl which should give the same results as using Avisynth.

Edit edit edit edit: found the reason why eac3to does that:

Remarks
Any PCM format that has more than 2 channels, more than 16 bits per sample, or more than 44,100 samples per second must be described by [a WAVEFORMATEXTENSIBLE] structure instead of a non-extended WAVEFORMATEX structure.

Desbreko

2010-09-03, 18:40

That's odd. I've been encoding some BDs with 24 bit PCM audio and loading the WAV files eac3to spits out into AviSynth 2.58 for trimming just fine.

eac3to CLI:
eac3to "00002.m2ts" 2: "00002 - 2 - PCM, 2.0 channels, 24 bits, 48khz.wav" -2 -24 -big -48000

AviSynth script:
video = ffvideosource("00002.mkv", fpsnum=24000, fpsden=1001)
audio = wavsource("00002 - 2 - PCM, 2.0 channels, 24 bits, 48khz.wav")
audiodub(video,audio)

But anyway, you could try NicAudio (http://nicaudio.codeplex.com/) for loading them into AviSynth.

Kristen

2010-09-06, 08:18

Yeah, Desbreko, it does this on some computers, and not others.

I found a solution, which was in my eac3to command line, I added in -simple.

Aimbotter33

2010-09-06, 22:29

Hi im Having Problem When Im Inserting A Afx Picture, Heres my Code,
Everytime When i Delete the = Beside the insertsign, it will say no such function.

DirectShowSource("C:\Users\Aimbotter33\Desktop\[FFFpeeps]_Kaichou_wa_Maid-sama_23_[720p][B40216DC].mkv")
sign1 = ("C:\Users\Aimbotter33\Desktop\[FFFpeeps]_Kaichou_wa_Maid-sama_23_[720p][B40216DC].mkv")
logo = imagesource("C:\Users\Aimbotter33\Desktop\images.png", pixel_type="RGB32", end=100)

insertsign=("last, logo, 0")
insertsign=("last, sign1, 563, 900")