View Full Version : To many artifacts. Need the suitable filters!!
plomon
2nd February 2006, 11:45
You are already losing all the details. I do not recommend using eDeen, ever!
You could try using Didée's MCNR_simple2.
I don't know the settings of MCNR_simple2. Please help me by specifying the correct settings. What's it's purpose & ow do I use it ?
My next question is - Even after removing noise, a lot of artifacts are still present like small size blocking, faces looking like they are bleeding at some places - especially the edges bleed a lot(Cap 02, Cap03, Cap 05 in Post#47 describe them best).
plomon
2nd February 2006, 12:37
@Didée
Well here is the script using fft3dfilter and the corresponding captures:
LoadPlugin("d:\programs\dgindex\dgdecode.dll")
LoadPlugin("d:\program files\avisynth 2.5\plugins\DeComb.dll")
LoadPlugin("d:\program files\avisynth 2.5\plugins\warpsharp.dll")
LoadPlugin("d:\program files\avisynth 2.5\plugins\masktools.dll")
LoadPlugin("d:\program files\avisynth 2.5\plugins\fft3dfilter.dll")
import("d:\program files\avisynth 2.5\plugins\LimitedSharpen.avs")
mpeg2source("e:\sol\sol.d2v")
Telecide(order=1,post=0,guide=1).Decimate(cycle=5,mode=0)
fft3dfilter(sigma=8,sigma2=2,sigma3=2,sigma4=2, bt=4,plane=4,sharpen=0.7)
Crop(4,14,-8,-14)
LancZosResize(640,344)
LimitedSharpen(ss_x=1.0,ss_y=1.0,smode=3,dest_x=640,dest_y=344,strength=0)
Caps for the above script :
http://img75.imageshack.us/img75/231/fft3d_cap04.png
http://img75.imageshack.us/img75/9837/fft3d_cap10.png
http://img75.imageshack.us/img75/6816/fft3d_cap06.png
http://img75.imageshack.us/img75/5830/fft3d_cap09.png
Well, it surely does retain maximum detail, but is quite ineffective in removing certain noise as compared to the previous filters. However, ruling out the eDeen() filter(since it's too strong), RemoveGrain() emerges as the best according to me, coz, if one RemoveGrain() cannot remove noise, then I'll use4 in a series, and make sure that noise is removed. And I'll try not to loose too much detail by reducing the no. of consecutive usages of RemoveGrain(). I prefer this filter for two reasons:
1.) It's much faster than fft3dfilter.
2.) It's easier to understand and use. So, more flexible. fft3dfilter looks very complex to me.
With the above settings for fft3dfilter, it's clear that a lot of noise still remains. If I try strengthening this filter more and more, it will take huge amount of time to encode. But, that's not the case with RemoveGrain(). Moreover, I feel that all I've to look after in RemoveGrain() is about which mode I'm using. I prefer modes 1,2,3,4 & 17(4 & 17 for strong denoising). I guess, a combination of RemoveGrain()'s consecutive filters using the above modes may help in retaining detail and yet remove fair amount of noise.
My imageshack.us credits ran out. So, the next posts with pictures may be delayed.W
Boulder
2nd February 2006, 12:48
Remove LanczosResize and add some strength for LimitedSharpen. With strength 0 you do no sharpening whatsoever.
foxyshadis
2nd February 2006, 13:37
I d'no, he obviously loves filtering the hell out of it, and can't even see results with more subtle filters, so what's wrong with that? It may go against the sensibilities of the rest of us, but it's his encode.
But plomon, if you proceed with such heavy, fast filtering you're guaranteed to kill edges and cause bleeding; unfortunately there's not really any way to have the one without the other.
Oh, and you may want to turn your monitor's brightness down. I can't be sure if that's it, but jacking the brightness up is a sure way to spot all kinds of "artifacts" that wouldn't show in a normal playback brightness at all.
plomon
2nd February 2006, 14:12
It may go against the sensibilities of the rest of us, but it's his encode.
I don't understand, how you arrived at such a conclusion, but I apologize for any irrelevant discussion I lifted up.
But plomon, if you proceed with such heavy, fast filtering you're guaranteed to kill edges and cause bleeding; unfortunately there's not really any way to have the one without the other.
I guess, finally you're indirectly insisting upon a single statement - "Don't use any filter!!"
Anyway, let me tell you, if I were to proceed with the fastest filtering, I obviously wouldn't have taken help of this forum and, that would have resulted in the most careless encode. Please think. I care about what I do, and I respect people's suggestions. But, without trying out on my system what has been suggested, how can you expect me to follow everything. I'm considering ideas, posting as many caps as possible, and was never in a rush to fix my mind upon a particular filtering process.
If I won't try what has been suggested, I can't post the results of that method, and consequently, nobody can suggest an alternative thereafter. That's the only reason I'm making so many attempts.
Regarding the brightness issue, it is completely normal.
foxyshadis
2nd February 2006, 14:55
The source is hardly perfect; it has some chroma noise and blocking, and a rather unusual lack of sharpness. (Most DVDs are oversharpened rather than under.) Of course filtering can help it. But our definitions of 'noise' differ, so while most of us would rather preserve most every wrinkle and texture under the grain in the original, you seem to prefer a much smoother, cleaner look. That's why our softer filters don't seem to work. For instance, you might something like
fft3dfilter(sigma=12,sigma2=5,sigma3=4,sigma4=4, bt=4,plane=4,sharpen=0.7)
which doubles the strength.
It's hard to give suggestions unless we know what you're looking for, and it's hard to know what you're looking for without having it to show us, so I can understand the difficulty. :p
Chainmax
2nd February 2006, 15:04
Would it really be that much of a trouble cutting a 10sec vob sample with something like ChopperXP (http://www.inchwest.com/digital-miner.com/chopperxp.html) and upload it to RapidShare (http://www.rapidshare.de/)?
plomon
2nd February 2006, 16:06
Here are a few clips from the movie:
http://rapidshare.de/files/12383031/sol_sample1.m2v.html
http://rapidshare.de/files/12383439/sol_sample2.m2v.html
http://rapidshare.de/files/12384438/sol_sample3.m2v.html
http://rapidshare.de/files/12385427/sol_sample4.m2v.html
http://rapidshare.de/files/12385103/sol_sample5.m2v.html
These may help understand about the quality in a better way.
Chainmax
2nd February 2006, 17:27
Thanks, now it will be much easier for us to help you.
Chainmax
3rd February 2006, 03:22
plomon, what do you think about this (http://rapidshare.de/files/12427861/SotLTest.avi.html)?
By the way, I haven't checked ther other samples, but sample 4 has an average bitrate of 1400-1500kbps! :scared:
plomon
3rd February 2006, 05:09
plomon, what do you think about this (http://rapidshare.de/files/12427861/SotLTest.avi.html)?
By the way, I haven't checked ther other samples, but sample 4 has an average bitrate of 1400-1500kbps! :scared:
i think there's not much difference compared to the orig., well, although some detail is removed(max. is preserved), noise(grainy noise) still exists. mainly the film still looks like what we call "the old reel"(which infact is the property of the orig clip). may be a bit of sharpness increase in the overall frame or in the chroma part will do some more good.
another problem if you noticed is that all parts of the movie are not the same. some are good and the other not. compare executing the same script on other clips. you may find irrelevancy of filter usage. i myself felt, i should use different filters for different parts of the movie.
can u post the script used for the clip u posted.
Chainmax
3rd February 2006, 12:47
My encode has some grain because I used a grain filter at the end of the script. Your source is horrible, and removal of the artifacts will always leave a very soft picture. The little extra grain helps mask that and whatever remaining artifacts there might be. Here's my script:
AssumeTFF()
Interp = SeparateFields().SelectEven().EEDI2(field=1)
Deinted=TDeint(order=1,field=1,edeint=interp)
TFM(mode=1,order=1,clip2=deinted)
TDecimate(mode=1)
FunkyDeblock()
DeGrainMedian()
Crop(6,18,704,448,align=true)
Lanczos4Resize(656,352)
DeHalo_Alpha()
dull=last
sharp=dull.LimitedSharpen()
Soothe(sharp,dull,20)
AddGrain(2,0,0)
plomon
3rd February 2006, 14:50
My encode has some grain because I used a grain filter at the end of the script. Your source is horrible, and removal of the artifacts will always leave a very soft picture. The little extra grain helps mask that and whatever remaining artifacts there might be. Here's my script:
AssumeTFF()
Interp = SeparateFields().SelectEven().EEDI2(field=1)
Deinted=TDeint(order=1,field=1,edeint=interp)
TFM(mode=1,order=1,clip2=deinted)
TDecimate(mode=1)
FunkyDeblock()
DeGrainMedian()
Crop(6,18,704,448,align=true)
Lanczos4Resize(656,352)
DeHalo_Alpha()
dull=last
sharp=dull.LimitedSharpen()
Soothe(sharp,dull,20)
AddGrain(2,0,0)
How much time did it take to encode the video with the above script ?? I guess it takes a very long time with so many filters in the script. The script looks extremely complex to me, especially in the deinterlacing region. Isn't Decomb enough for deinterlacing and pulldown ?? Because the above method uses some other filters like TDeint, TDecimate, which I need to learn now to understand their usage. Please use them only if Decomb is totally incapable of doing it's job for the current video source. Otherwise, usage of filters in your script needs me to learn again about all those.
Coming to other filters like FunkyDeblock() & DeHalo_Alpha() - they are not available with me. Please tell me, where I can download them.
plomon
3rd February 2006, 16:32
In the above script TDeint doesn't seem to have any parameter called as "edeint". Are you referring to the parameter "clip2" instead. The version of TDeint I'm using is 1.0b4, and my AVISynth version is 2.56(build Oct 28, 2005). Am I using the correct combo of Deint and AVISynth ??
I searched and downloaded a few plugins and scripts you used in the above script. I loaded all of 'em, and then this is what has happened:
After loading the following script, VDubMod v1.5.10.2 automatically closes off:
LoadPlugin("d:\programs\dgindex\dgdecode.dll")
LoadPlugin("d:\program files\avisynth 2.5\plugins\TIVTC.dll")
LoadPlugin("d:\program files\avisynth 2.5\plugins\TDeInt.dll")
LoadPlugin("d:\program files\avisynth 2.5\plugins\LoadPluginEX.dll")
LoadPlugin("d:\program files\avisynth 2.5\plugins\warpsharp.dll")
LoadPlugin("d:\program files\avisynth 2.5\plugins\masktools.dll")
LoadPlugin("d:\program files\avisynth 2.5\plugins\EEDI2.dll")
LoadPlugin("d:\program files\avisynth 2.5\plugins\AddGrain.dll")
LoadPlugin("d:\program files\avisynth 2.5\plugins\DeGrainMedian.dll")
import("d:\program files\avisynth 2.5\plugins\LimitedSharpen.avs")
import("d:\program files\avisynth 2.5\plugins\Soothe.avs")
import("d:\program files\avisynth 2.5\plugins\FunkyDeblock.avs")
import("d:\program files\avisynth 2.5\plugins\Dehalo_alpha.avs")
mpeg2source("e:\sol\sol.d2v")
AssumeTFF()
Interp = SeparateFields().SelectEven().EEDI2(field=1)
Deinted=TDeint(order=1,field=1,clip2=interp)
TFM(mode=1,order=1,clip2=deinted)
TDecimate(mode=1)
FunkyDeblock()
DeGrainMedian()
Crop(6,18,704,448,align=true)
Lanczos4Resize(656,352)
DeHalo_Alpha()
dull=last
sharp=dull.LimitedSharpen()
Soothe(sharp,dull,20)
AddGrain(2,0,0)
The only change made in the above script is in the Deint() function - I replaced "edeint" parameter with "clip2", b'coz, the "edeint" parameter doesn't exist according to my AVISynth manual.
Did I load any wrong plugins or scripts ??
plomon
3rd February 2006, 16:54
hooray!! I found the problem. The script without the DeHalo_Alpha() filter is loading fine into VDubMod.
Now, I don't know what's wrong with the DeHalo_Alpha.avs script that I've. But this is what I've in the DeHalo_Alpha.avs script:
function DeHalo_alpha(clip clp, float "rx", float "ry", float "darkstr", float "brightstr", float "lowsens", float "highsens", float "ss")
{
rx = default( rx, 1.0 )
ry = default( ry, 1.0 )
darkstr = default( darkstr, 0.5 )
brightstr = default( brightstr, 0.5 )
lowsens = default( lowsens, 25 )
highsens = default( highsens, 25 )
ss = default( ss, 1.0 )
LOS = string(lowsens)
HIS = string(highsens/100.0)
DRK = string(darkstr)
BRT = string(brightstr)
ox = clp.width()
oy = clp.height()
uv = 0
uv2 = (uv==3) ? 3 : 2
halos = clp.bicubicresize(m4(ox/rx),m4(oy/ry)).bicubicresize(ox,oy,1,0)
are = mt_lutxy(clp.mt_expand(U=uv,V=uv),clp.mt_inpand(U=uv,V=uv),"x y -","x y -","x y -",U=uv,V=uv)
ugly = mt_lutxy(halos.mt_expand(U=uv,V=uv),halos.mt_inpand(U=uv,V=uv),"x y -","x y -","x y -",U=uv,V=uv)
so = mt_lutxy( ugly, are, "y x - y 0.001 + / 255 * "+LOS+" - y 256 + 512 / "+HIS+" + *" )
lets = mt_merge(halos,clp,so,U=uv,V=uv)
remove = (ss==1.0) ? clp.repair(lets,1,0)
\ : clp.lanczosresize(m4(ox*ss),m4(oy*ss))
\ .mt_logic(lets.mt_expand(U=uv,V=uv).bicubicresize(m4(ox*ss),m4(oy*ss)),"min",U=uv2,V=uv2)
\ .mt_logic(lets.mt_inpand(U=uv,V=uv).bicubicresize(m4(ox*ss),m4(oy*ss)),"max",U=uv2,V=uv2)
\ .lanczosresize(ox,oy)
them = mt_lutxy(clp,remove,"x y < x x y - "+DRK+" * - x x y - "+BRT+" * - ?",U=2,V=2)
return( them )
}
function m4(float x) {return(x<16?16:int(round(x/4.0)*4))}
Did I download the wrong script for DeHalo_Alpha. This is where I downloaded the script from:
http://www.filefactory.com/get/v2.1/f2.php?f=2233f12daa16708d5d3d0a28
The above link is present in this site (http://www.hardwarexl.com/index.php?option=com_content&task=view&id=1192&Itemid=41&limit=1&limitstart=3&lang=en).
Well on an overall basis, keeping in mind that my source is a nasty one, usage of any number of filters may not help in making it better. May be, I should be content with whatever I've.
On the brighter side, I learnt a good deal of scripting.
foxyshadis
3rd February 2006, 17:07
Your Tdeint is old. Update. (http://bengal.missouri.edu/~kes25c/) Clip2 will work, but incorectly.
There's nothing wrong with dehalo_alpha, I made that mod (damn, that got around fast). Most likely you need up update masktools 2 and removegrain.
TDeint/TFM definitely game me better results than decomb. (Part of why it's my default now.) Otherwise I'll refrain from commenting on a few things.
plomon
4th February 2006, 07:24
Your Tdeint is old. Update. (http://bengal.missouri.edu/~kes25c/) Clip2 will work, but incorectly.
There's nothing wrong with dehalo_alpha, I made that mod (damn, that got around fast). Most likely you need up update masktools 2 and removegrain.
TDeint/TFM definitely game me better results than decomb. (Part of why it's my default now.) Otherwise I'll refrain from commenting on a few things.
Thanks a lot for the updated version of TDeint. I downloaded it, and this plugin is accepting "edeint" parameter. However, DeHalo() still poses trouble. The VDubMod still closes when DeHalo_Alpha plugin script is loaded. I loaded mt_masktools.dll(Is this what you were referring to as masktools 2 ?)
I attached the masktools 2 plugin that I'm using. Please check if I loaded the correct masktools plugin.
Guest
4th February 2006, 15:08
I attached the masktools 2 plugin that I'm using. Please check if I loaded the correct masktools plugin. I deleted the huge attachment because you can simply state the version you are using.
plomon
6th February 2006, 03:24
I deleted the huge attachment because you can simply state the version you are using.
it's okay if you deleted the attachment, but am I using the right version or not ?
Moreover, I took a big decision which I wasn't able to post, b'coz, yesterday doom9 site did not open at all.
I decided to get a new DVD, that's it! However, all the things I've learnt about repairing a bad video would not had been possible without the support and co-operation of you guys. So, thx a lot for all the help.
Chainmax
6th February 2006, 18:01
Yeah, you made the right choice by buying another DVD release. Yours was simply to bad to fix properly. In any case, read the documents and tutorials I linked for you, as they will greatly help in your future projects.
plomon
7th February 2006, 01:41
read the documents and tutorials I linked for you
What tutorials ?? You did not post any. If you like to, then u can.
Chainmax
7th February 2006, 13:21
The ones I linked to in your other thread (http://forum.doom9.org/showthread.php?t=106734).
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.