Avisynth 2.6 MT - Page 27

the_weirdo · 30th August 2012, 03:51

Quote:

Originally Posted by SEt

Give me exact dlls and script that reproduce the error – I'm building RemoveGrain myself and don't have problems.

Here are some scripts and plugins that can reproduce the error:
https://dl.dropbox.com/u/18695757/Test_Avisynth_2.6_MT.7z

Like Boulder, I also get those plugins from this post:
http://forum.doom9.org/showthread.php?p=1568142#post1568142

After I replace those plugins with their original versions, the error doesn't occur any more. But they're worked fine with previous builds.

Tempter57 · 30th August 2012, 04:45

SEt
Error on SeeSaw.avsi (function sharpen2), GrainFactory3.avs, plugins with original versions

the_weirdo · 30th August 2012, 05:55

Quote:

Originally Posted by Tempter57

SEt
Error on SeeSaw.avsi (function sharpen2), GrainFactory3.avs, plugins with original versions

With SeeSaw.avsi, you can open it in an editor like Notepad and change float "Spower" to int "Spower" (line #33). That means you cannot use a float value for "Spower" parameter. If you want to keep using float for "Spower" then you need to change int power to float power in function sharpen2 (line #95).

About GrainFactory3.avs, I don't know.

Tempter57 · 30th August 2012, 06:29

the_weirdo
Thank you, with GrainFactory3.avs all good

forclip · 30th August 2012, 06:42

SEt
Thanks for the build

Quote:

Originally Posted by the_weirdo

Like Boulder, I also get those plugins from this post:
http://forum.doom9.org/showthread.ph...42#post1568142

These plugins must be recompiled (+small changes) to be compatible with the "new-new" AviSynth 2.6 interface.

-----

Attached DirectShowSource.dll for 2.6.

SEt · 30th August 2012, 16:00

Plugins in that post, as forclip said, need to be recompiled for the "new-new" interface. Personally I suggest using classic interface if you don't need something from the "new-new" one.

In SeeSaw.avsi it's clearly the discussed situation of working bugged code due to ints masking as floats. The script should be changed this way:

Code:

function sharpen2(clip clp, float strength, float power, float zp, float lodmp, float hidmp, int rgmode) 
{
STR  = string( strength )
PWR  = string( 1.0/power )

Keiyakusha · 30th August 2012, 16:24

So everything that stopped working I have to modify by hands (in case of script) or get recompiled version (in case of plugin)? Why not just roll back changes that caused this?
Btw this is only MT build, or normal one affected too?

SEt · 30th August 2012, 16:59

What "everything" is stopped working? I see only plugins built especially for that one build of Avisynth and one previously hidden bug in one script surfaced (what is a good thing).
No idea what "normal one" build you are referencing to.

Keiyakusha · 30th August 2012, 18:08

Well this is my bad English. Perhaps "anything" suits better there? I have a bunch of scripts that stopped working in new build and that doesn't use any plugins from -Vit-. They return message that invalid arguments was passed, maybe some other messages too. But it works with previous build. I'm not going to investigate which scripts are problematic and where exactly these problems appear.

P.S.
There was build (official?) without MT capabilities. This is what i called "normal"

SEt · 30th August 2012, 20:09

The int-float change just exposed the hidden problems that you can stumble upon anyway with different parameters for your scripts.
The last "official" build is from 25.05.2011 I think. You are free to use it, but personally I find nothing great there.

Keiyakusha · 30th August 2012, 20:53

Quote:

Originally Posted by SEt

The last "official" build is from 25.05.2011 I think. You are free to use it, but personally I find nothing great there.

I have compiled version from 08.02.12. What I'm actually asking is if this change that "broke" some things is in the trunk or this is your custom upgrade? I said about "normal" version before because some time ago MT stuff wasn't part of the avisynth 2.6, not officially. maybe things changed I don't know. I don't follow actual avisynth development so I'm lost a bit. I don't really need MT stuff anyway. If I still can compile it without MT and this latest change - this is fine. But If everything affected by this change- it really should be rolled back.

06_taro · 31st August 2012, 02:04

So if anything relying on a buggy behaviour stopped working after a new version fixed it, this bugfix should be rolled back?

the_weirdo · 31st August 2012, 04:05

Quote:

Originally Posted by 06_taro

So if anything relying on a buggy behaviour stopped working after a new version fixed it, this bugfix should be rolled back?

This actually happen with VSFilter.

How about an additional build until those buggy scripts are fixed?

SEt · 31st August 2012, 08:49

If you want previous behavior just rollback 1 build back.

Keiyakusha · 31st August 2012, 09:35

Quote:

Originally Posted by 06_taro

So if anything relying on a buggy behaviour stopped working after a new version fixed it, this bugfix should be rolled back?

Yes. This is one of the cases that can can be called "don't fix what's not broken". I don't recall any problems with that behavior. Until now no one even noticed it! Buggy behavior was there long enough, we can call it feature now.
Edit: What will happen if google will fix buggy behaviour in vp8 bitstream? Or any of the old bugs in vsfilter will be fixed? It will lead to lost compatibility with something but will bring... I don't know what so good it will bring. To me situation with this latest avisynth change looks very similar.
And for example, I have tons of scripts and plugins that I don't even remember where I found and how many times modified. If any of these affected too, I don't see how they will be fixed unless I'll do this myself.

the_weirdo · 31st August 2012, 11:54

To be honest, I use these builds because they're up-to-date with CVS (which contain some fixes that I may encounter), not because of MT feature. I'm still using a Wolfdale desktop CPU so MT doesn't have much benefit for me. But I don't want to wait for a new official build. I wish there were someone willing to compile "nightly" builds for Avisynth 2.6.

Actually I didn't intend to complain or anything, I just want an additional option for those who cannot wait for those buggy scripts are fixed (I doubt they'll be fixed by their author), like Mosu have done with feature "header compression" of Matroska for MKVtoolnix. I don't use many function scripts and it's trivial for me to fix them (because it's just int-float misplace).

P.S. My English is really poor so if there're offending words or sentences then it's not my intent.

zerowalker · 2nd September 2012, 11:41

QTGMC() doesn´t work for me, something about NNEDI3 Arg 0, this is with the latest build, it work with 1 step earlier.

manolito · 3rd September 2012, 15:57

Quote:

Originally Posted by Atak_Snajpera

xp users must extinct. i don't care whether you like it or not. the same happened with win9x users and your whinning won't prevent that!

This is a childish statement.

Latest statistics about marketshare of different Windows versions:
http://marketshare.hitslink.com/oper...11&qpcustomb=0

42.52% of all users still use XP. It would be a very bad idea to drop XP support at this time already...

Cheers
manolito

Overdrive80 · 4th September 2012, 21:46

@Set After of updating avisynth with your current version, I got this error:

With version 2012.05.16, it didnt success.

Script used:

Code:

SetMemorymax(1024)
SetMtMode(5,4)

DGDecode_mpeg2source("D:\DBZ\DBZ1_12\071\Title_7.d2v", info=3)

ColorMatrix(hints=true, threads=0)

#NO deinterlace

#crop&resize (0,0,-0,-0)
	#Opening
	D0=Trim(0,2615).Crop(8,0,-6,-4).Spline64Resize(720,480)

	#Resumen
	D1=trim(2616,4077).Crop(2,0,-2,-0).Spline64Resize(720,480)

	#Titulo
	D2=trim(4078,4281).Crop(2,0,-2,-0).Spline64Resize(720,480)

	#Capitulo - 1 parte
	D3=trim(4282,18303).Crop(2,0,-2,-4).Spline64Resize(720,480)
	
	#Cortinillas
	D4=trim(18304,18595).Crop(2,0,-2,-0).Spline64Resize(720,480)

	#Capitulo - 2 parte
	D5=trim(18596,32431).Crop(2,0,-2,-4).Spline64Resize(720,480)

	#Ending
	D6=trim(32432,34735).Crop(2,0,-2,-0).Spline64Resize(720,480)

	#Avance
	D7=trim(34736,0).Crop(4,0,-2,-6).Spline64Resize(720,480)
	
	UnalignedSplice(D0,D1,D2,D3,D4,D5,D6,D7)

###################################
#	     FILTRADO             #
###################################

Setmtmode(2)

original=last

aa=spline64Resize(width(original)*2,height(original)*2).SangNom(aa=29).spline64Resize(width(original),height(original)).mergechroma(original)

prefiltrado= dfttest(sigma=4,tbsize=1,tmode=0,lsb=true).ditherpost(mode=2)


super=Msuper(prefiltrado)

backward_vec3 = MAnalyse(super, isb = true, delta = 3, overlap=4,search=4)
backward_vec2 = MAnalyse(super, isb = true, delta = 2, overlap=4,search=4)
backward_vec1 = MAnalyse(super, isb = true, delta = 1, overlap=4,search=4)

forward_vec1 = MAnalyse(super, isb = false, delta = 1, overlap=4,search=4)
forward_vec2 = MAnalyse(super, isb = false, delta = 2, overlap=4,search=4)
forward_vec3 = MAnalyse(super, isb = false, delta = 3, overlap=4,search=4)

prefiltrado.MDegrain3(super,backward_vec1,forward_vec1,backward_vec2,forward_vec2,backward_vec3,forward_vec3,
			\thSAD=370,plane=4,lsb=true).ditherpost(mode=2)

contrasharpening(last,aa)

toon(0.10)

gradfun3()

SEt · 5th September 2012, 06:35

Use the stable 2.5 API (i.e. mt_masktools-25.dll) if you don't want to rebuild the filters but want the latest Avisynth version.

People, these builds are not user-friendly: they are pretty stable and as bug free as possible, but API changes and breaking something old already broken are to be expected. Do not expect that everything will behave as in previous build, especially if you use 2.6 features.

30th August 2012, 16:00	#526 \| Link
SEt Registered User Join Date: Aug 2007 Posts: 374	Plugins in that post, as forclip said, need to be recompiled for the "new-new" interface. Personally I suggest using classic interface if you don't need something from the "new-new" one. In SeeSaw.avsi it's clearly the discussed situation of working bugged code due to ints masking as floats. The script should be changed this way: Code: function sharpen2(clip clp, float strength, float power, float zp, float lodmp, float hidmp, int rgmode) { STR = string( strength ) PWR = string( 1.0/power )

30th August 2012, 18:08	#529 \| Link
Keiyakusha 契約者 Join Date: Jun 2008 Posts: 1,576	Well this is my bad English. Perhaps "anything" suits better there? I have a bunch of scripts that stopped working in new build and that doesn't use any plugins from -Vit-. They return message that invalid arguments was passed, maybe some other messages too. But it works with previous build. I'm not going to investigate which scripts are problematic and where exactly these problems appear. P.S. There was build (official?) without MT capabilities. This is what i called "normal" Last edited by Keiyakusha; 30th August 2012 at 18:11.

31st August 2012, 11:54	#536 \| Link
the_weirdo Yes, I'm weird. Join Date: May 2010 Location: Southeast Asia Posts: 271	To be honest, I use these builds because they're up-to-date with CVS (which contain some fixes that I may encounter), not because of MT feature. I'm still using a Wolfdale desktop CPU so MT doesn't have much benefit for me. But I don't want to wait for a new official build. I wish there were someone willing to compile "nightly" builds for Avisynth 2.6. Actually I didn't intend to complain or anything, I just want an additional option for those who cannot wait for those buggy scripts are fixed (I doubt they'll be fixed by their author), like Mosu have done with feature "header compression" of Matroska for MKVtoolnix. I don't use many function scripts and it's trivial for me to fix them (because it's just int-float misplace). P.S. My English is really poor so if there're offending words or sentences then it's not my intent. Last edited by the_weirdo; 31st August 2012 at 12:02.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

30th August 2012, 04:45	#522 \| Link
Tempter57 Registered User Join Date: Jan 2011 Location: Donetsk Posts: 58	SEt Error on SeeSaw.avsi (function sharpen2), GrainFactory3.avs, plugins with original versions

30th August 2012, 06:29	#524 \| Link
Tempter57 Registered User Join Date: Jan 2011 Location: Donetsk Posts: 58	the_weirdo Thank you, with GrainFactory3.avs all good

30th August 2012, 16:24	#527 \| Link
Keiyakusha 契約者 Join Date: Jun 2008 Posts: 1,576	So everything that stopped working I have to modify by hands (in case of script) or get recompiled version (in case of plugin)? Why not just roll back changes that caused this? Btw this is only MT build, or normal one affected too?

30th August 2012, 16:59	#528 \| Link
SEt Registered User Join Date: Aug 2007 Posts: 374	What "everything" is stopped working? I see only plugins built especially for that one build of Avisynth and one previously hidden bug in one script surfaced (what is a good thing). No idea what "normal one" build you are referencing to.

30th August 2012, 20:09	#530 \| Link
SEt Registered User Join Date: Aug 2007 Posts: 374	The int-float change just exposed the hidden problems that you can stumble upon anyway with different parameters for your scripts. The last "official" build is from 25.05.2011 I think. You are free to use it, but personally I find nothing great there.

31st August 2012, 02:04	#532 \| Link
06_taro soy sauce buyer Join Date: Mar 2010 Location: United Kingdom Posts: 164	So if anything relying on a buggy behaviour stopped working after a new version fixed it, this bugfix should be rolled back?

31st August 2012, 08:49	#534 \| Link
SEt Registered User Join Date: Aug 2007 Posts: 374	If you want previous behavior just rollback 1 build back.

2nd September 2012, 11:41	#537 \| Link
zerowalker Registered User Join Date: Jul 2011 Posts: 1,121	QTGMC() doesn´t work for me, something about NNEDI3 Arg 0, this is with the latest build, it work with 1 step earlier.

5th September 2012, 06:35	#540 \| Link
SEt Registered User Join Date: Aug 2007 Posts: 374	Use the stable 2.5 API (i.e. mt_masktools-25.dll) if you don't want to rebuild the filters but want the latest Avisynth version. People, these builds are not user-friendly: they are pretty stable and as bug free as possible, but API changes and breaking something old already broken are to be expected. Do not expect that everything will behave as in previous build, especially if you use 2.6 features.