AnimeIVTC() - All-in-one solution for interlacing and blending - v2.00 [2010-01-07] [Archive] - Page 7

thetoof

20th October 2008, 03:29

every once in a while there would be horrible artifacts, as big as a third of the frame that would last for only one frame.
Yeah... this is the big downsize of omode=1 (does the same as you did), but as you may have seen on the forum, I'm working on building a mask to properly restore the credits on the ivtced background.
using YV12 there is loss of quality?Your source is YV12 and you'll most probably encode to YV12, so no, there is no loss of quality.I have created a avi YV12 with Lagarith and I have load it with VirtualDub, but the speed does not change :(The speed of the bobbing won't change. High quality bobbing takes time, that's just how it is. However, the main purpose of a rendering pass is to use the new file for all subsequent processing, which will be faster.

@inurenegade & Oryon
Follow the advice of Merlin7777.

@RsVolcom
I'll have some time to look at your samples later tonight of tomorrow.If I use mode=6, there is no problem but if I use mode=3 or mode=4 then it will do the entire film telecined and then amend the interlaced parts to the end of the episode but with mode=6 it does the interlaced scenes separately as expected.Uh... do you mean that there is a difference on how the sections are spliced together? That'd be kinda weird since the splicing part of the script is the same for mode=3,4 and 6...

Kripsy

20th October 2008, 10:34

Uh... do you mean that there is a difference on how the sections are spliced together? That'd be kinda weird since the splicing part of the script is the same for mode=3,4 and 6...

Yeah, that is exactly what I mean.The splices are amended to the end of the video effectively making it four minutes longer (the length of my i1). I don't know what the problem is. I am using the latest version of the script as well.

inurenegade

21st October 2008, 05:58

Crashing? If I remember correctly, it never actually crashed. It was just a crc mismatch. Do you have an exact error? Can you describe what exactly you are doing?

It does not give me an error it just closes virtualdubmod and avsp whenever i load the script. The script does work fine when i am not using passes.

Adub

21st October 2008, 17:48

Does it still crash if you set MT=false?

inurenegade

21st October 2008, 18:36

Does it still crash if you set MT=false?

No it works fine now when mt=false

Adub

21st October 2008, 19:02

It sounds like it has to do with your MT MVTools version. For one, make sure that you are using Josey Wells' version of the MT MVtools. Secondly, make sure it is either up to date, or rolled back to a version that works. I cannot remember the exact version number, but I think it is mentioned somewhere in this thread. If it doesn't work either way, stick with the non-MT version of MVtools for now.

inurenegade

21st October 2008, 20:28

I used the mvtools.dll in the package posted under Intel10-Core2-SSE4 is that for ssse3 or actual sse4 instructions?

lansing

22nd October 2008, 04:52

toof, i'm trying to run SharpAAMCmod() function separately from AnimeIVTC, but the parameter names seems to be different than the one said in the documentation.

Adub

22nd October 2008, 07:15

@inurenegade
Yes, I think that that version is for the actual SSE4, but I am not one hundred percent sure. Try switching it for another compile and see if it fixes things.

inurenegade

22nd October 2008, 21:45

Using the Core 2 Duo SSE3 works great :)

thetoof

23rd October 2008, 03:36

The splices are amended to the end of the video effectively making it four minutes longer (the length of my i1). I'll check that in the next days...

toof, i'm trying to run SharpAAMCmod() function separately from AnimeIVTC, but the parameter names seems to be different than the one said in the documentation. Yup, my bad. It will be changed in the next version.

@ RsVolcom
I have asked the help of MOmonster for your problem.

Adub

23rd October 2008, 04:10

@inurenegade,
That's good to hear! I will have to remember that in the future.

Kripsy

23rd October 2008, 07:16

Yeah, that is exactly what I mean.The splices are amended to the end of the video effectively making it four minutes longer (the length of my i1). I don't know what the problem is. I am using the latest version of the script as well.

Forget about it, user error. :o

I had actually figured out my frames from mode=6 and when I changed it to mode=3/4 the frame numbers had changed without me realizing it making it have weird behavior.

Edit: I can however report that with my new script I am getting tons of blending artifacts in the credits.

MPEG2Source("FLCLEp1.d2v",cpu=0)
AnimeIVTC(mode=3,omode=1,mix=4,estart1=0,eend1=31836,istart1=31837,iend1=34780,estart2=34781,eend2=35487,aa=0,MT=True,denoise=false,edimode="nnedi")
crop( 4, 0, -4, 0)

It seems that changing the fps from 30 to 24 isn't working out so well. Any ideas? Or is this one already in the midst of being addressed?

Edit2: Oh, I see. It is a problem with "motion" FPS conversion since there is a background image that moves a lot. I guess there is no easy fix.

thetoof

27th October 2008, 00:24

this one already in the midst of being addressed?Yes. Having a 20mb sample of your source with a lot of motion could help me (and the ones following this (http://forum.doom9.org/showthread.php?t=141996) thread) find a solution.

fleon

27th October 2008, 04:57

Hi I wanna try animeivtc with a film that I own it is Daft Punk's Interstella 555, Megui tells me that the the source is Hybrid film/interlaced. Mostly insterlaced, and also says that the field order is top field first,
and mediainfo tells me this

http://img159.imageshack.us/img159/6247/98935747mk5.th.jpg (http://img159.imageshack.us/my.php?image=98935747mk5.jpg)http://img159.imageshack.us/images/thpix.gif (http://g.imageshack.us/thpix.php)

So what settings should i use in animeivtc?

thetoof

27th October 2008, 05:14

Because of all that, any automated detection (DGIndex, megui, automkv... name it!) to know what type of treatment your source needs regarding IVTC/Deinterlacing will most likely fail. The only reliable source of information is the following... and your eyes.
Without a sample, I can't give any advice.

Nightshiver

27th October 2008, 12:33

fleon

27th October 2008, 17:11

Without a sample, I can't give any advice.
Give sample....

I used this as my script
DGDecode_mpeg2source("C:\INTERSTELLA 5555\VIDEO_TS\VTS_08_1.d2v",info=3)

__film = last
__t0 = __film.trim(5000, 6000)

__t0

and also I used the Unrestricted 1 pass max speed profile in megui

Here it is http://www.mediafire.com/?numumtjlf0f

Adub

27th October 2008, 18:26

It is best to use dgindex and the "Save and Demux" option. So cut the vob file using the DGindex markers, and then save and demux. Upload the resulting m2v file.

fleon

27th October 2008, 20:30

Without a sample, I can't give any advice.
Give sample....
It is best to use dgindex and the "Save and Demux" option. So cut the vob file using the DGindex markers, and then save and demux. Upload the resulting m2v file.

here it is the m2v file http://www.mediafire.com/?hdzdyjwnwg0

HymnToLife

27th October 2008, 23:55

Hmm, the documentation archive doesn't seem to exist on mediafire anymore, it gives me an error when I try to download it...

EDIT: Never mind, it's back.

Shiznadnaz

28th October 2008, 00:51

I've been using TDeint ever since I started encoding TV caps but a friend recommend that I use AnimeIVTC, I have tried a script from a friend which included AnimeIVTC and it worked great. I have this DVD, I recently downloaded but I'm not sure if I could use the same line from that script for it so I was wondering if you could help me out by checking the source - http://www.mediafire.com/?ofqvyentumd and also if you could help me identify what that is in the picture you can see all through the clip, its hard for me to describe but you can see it moving. Thank you very much, I look forward to your reply.

thetoof

28th October 2008, 04:23

I have this DVD, I recently downloadedCan't help, sorry. Check forum rule #6. The only thing I can say is: read the guide (especially the part about 3:2 pulldown and how to recognize your source... the answer lies there (and yes!!, animeivtc is much more appropriate than tdeint for your source))

fleon

28th October 2008, 04:37

thetoof

28th October 2008, 04:49

yes I did... answering to Shiznadnaz was simply quicker than playing with your source, as it's a bit trickier (at least at first glance).
I'm running some tests as we speak.

fleon

31st October 2008, 23:29

yes I did... answering to Shiznadnaz was simply quicker than playing with your source, as it's a bit trickier (at least at first glance).
I'm running some tests as we speak.

Any updates on my source? thanks in advance

thetoof

1st November 2008, 02:02

You have the same problem as RsVolcom. You can follow the development in this thread: http://forum.doom9.org/showthread.php?p=1208543#post1208543
The "fixable" thing he's talking about is your issue.

EuropeanMan

1st November 2008, 05:58

I have a progressive DVD9 source, but every 3 frames, you see ghosting; 2 frames no ghosting...what would be the proper way to remove ghosting for this type of source?

thetoof

1st November 2008, 06:12

ghosting? I think you're talking about combing.... can I have a sample? Quick call: try mode=2

EuropeanMan

1st November 2008, 06:15

Actually, I should have asked this question in restore thread, as this movie is NOT anime...sorry. Will cut a sample & upload it.

No, there is NO combing whatsoever, no interlacing lines whatsoever...pure progressive source, but there is ghosting.

It's a Hindi movie...what do you expect from those DVD manufacturers? :(

EuropeanMan

1st November 2008, 06:23

Sample of those ghosts: http://www.sendspace.com/file/3h4jz4

Thanks so much in advance for any suggestions...

thetoof

1st November 2008, 06:48

It's ok if the sample is not anime, since the script can also deal with "real life" material. However, the type of problem can't be solved by animeivtc().
Here is how the frames were blended together before being encoded as progressive video:
Film frames : A B C D
Video frames: 1=AA 2=AB 3=BC 4=CD 5=DD
The solution would be to take the common blend of frame 2&3 to recreate frame B, take the common blend of frame 3&4 to recreate frame C and discard the other parts to get back to the film rate... I know srestore deals with double blends, but I'd never heard of triple blends before!

EuropeanMan

1st November 2008, 06:52

Wowser, triple blends...my god. Are you an expert on SRestore? Will I need a bobber & will yadif suffice? I'm no expert on deinterlacing...would appreciate any headsup on this I can get...thank you so much.

thetoof

1st November 2008, 07:10

I can't say I'm an "expert" on srestore... MOmonster deserves that title ;)
You don't need to bob your source because it has no interlacing nor field blending (the whole frame is blended). If I fully understood how srestore works, I could help you adapt the function to your needs, but I currently have not enough time to investigate, so I asked some help in the "another restore function" thread. I'm afraid that I'd do more bad than good if I applied the tweaks I'm thinking of...

EuropeanMan

1st November 2008, 07:14

^ I'll take the risk if you have 5 minutes for me :( pleeeeeeeeeease

thetoof

1st November 2008, 07:34

Those 5 minutes were enough to know that this is still out of my league... I'd need a lot more time that I can't afford to take even if I want to.

EuropeanMan

1st November 2008, 07:40

Thank you - really appreciate your time...

regarding your solution above...how can I go about that ? is that through Srestore? or another filter I should be looking at?

thetoof

1st November 2008, 08:42

That'd be through the "double blend restoration" concept of srestore applied to work with triple blends. In other words, wait for MOmonster to be aware of the problem, then think a bit and post a new version of srestore with, hopefully for you, a new outputmode. Or either wait for me to master srestore...... which won't happen for, at least, a few months :p
Another option is to learn it by yourself, or hope someone on the forum will read this and know enough to adapt srestore ;)

EuropeanMan

1st November 2008, 08:52

Ok...will try. Thanks. I don't completely understand this...but will try.

Shiznadnaz

3rd November 2008, 15:54

I used

AnimeIVTC(mode=2,aa=0,killcomb=0,precision=3,decimate=4)

for this video I captured http://www.mediafire.com/?5zzcwm4nvdv but after it had encoded, there were still interlaced frames, do you think you could help me with this one?

thetoof

3rd November 2008, 18:36

O.o Was this really broadcasted??? If it was, I'll add a new mode for AnimeIVTC (I had only seen this on noobish encodes before, so I didn't think something "official" could be encoded this way)
You have a blend-deinterlaced telecined source encoded as progressive 30fps video.

For now:
#you can use another pp if you want... see srestore.txt for more info
srestore(omode="pp3", cache=10, other settings)
decimate()

I'd like to have a sample of the credits for that source to help the dev of animeivtc's new mode ((blend-de)interlaced credits on top of blend-deinterlaced telecined source).

fleon

5th November 2008, 23:47

You have the same problem as RsVolcom. You can follow the development in this thread: http://forum.doom9.org/showthread.php?p=1208543#post1208543
The "fixable" thing he's talking about is your issue.

MOmonster have answered you about the solution of RsVolcom and mine here http://forum.doom9.org/showthread.php?p=1210301#post1210301 but I dont quite understand it, also since I am new to this i dont know where to place masktools, I assume that is in the avisynth plusgins directory but i am not sure, as the post said i downloaded the last one masktools-v2.0a35.

About the post of MOmonster as i said i dont quite understand it, what I think I have to do is to create a normal script, add the lines that MOmonster wrote to my script and then MOmonster says something about "Just bob the source and add this function. Decimation is still necessary", i also dont know how to do that, could you help me with that?

thetoof

6th November 2008, 04:08

I just saw his post; the solution will be added in the next release of AnimeIVTC

Comatose

6th November 2008, 17:06

@ post #340: Oh wow. Even in Israel Bleach looks better... and the company that subs it (they don't dub Bleach here) has a tendency to do a lot of video screwups ;_;

Haohmaru

7th November 2008, 12:32

I would like to encode my Dragon Ball Z collection (the remastered edition) using x264 so that I can store them on my HDD too with less space. To be honest I don't have much experience in deinterlacing so I'm asking you guys for help. I did try but I wasn't even shure in the type of the source.
Anyway I've uploaded a 1min sample of the source. (http://rapidshare.com/files/161464188/dbz.rar)
Please help me to find the correct command line for deinterlacing.

Any help would be great appreciated :)

Alcohor

7th November 2008, 14:07

you don't deinterlace, with these sources you can use the option "forced film" in dgindex or :

tfm(d2v="d2vpath")
tdecimate()

in avisynth (with the tivtc plugin) :)

Haohmaru

7th November 2008, 19:08

It worked :)

Thx a lot !

thetoof

9th November 2008, 07:09

and to be sure that you know where to get the info next time:

Possible situations:
1 - DGIndex tells you "100% Film" and you see a 3:2 pattern.
...
Source type:
1 - Soft telecined, so your stream can easily be extracted at the film rate
...
Solutions:
1 - You are happy! :p Use "Force film" in DGIndex and IVTCing/Deinterlacing/Deblending doesn't concern you anymore!

stpdrgstr

9th November 2008, 07:58

Hi, first than nothing, excuse my bad english, and thanks to thetoof for this awesome script.

For a while i've been using YATTA to do all this kind of stuff, but now I'm getting a bit tired of doing some things manually (as i'm not expert with it anyways), and I found this to be a very good ivtc.

But I'm running into a little issue.

My source is Negima!? Spring OVA R1J DVD, a source that needed mode=6. I was intending to do a VFR output so the scrolling credits we're not screwed up. Everything went fine until the moment of doing the two pass mode=7.

The mode=6 is giving me what it's needed, a 29.970 fps with dupes at the episode sections, and 29.970 progressive credits. It would be obvious that the episode parts need to be deduped, and the credits should stay untouched when dedupping with mode 7. But sadly, this last step is just where i'm having troubles. The mode 7 is deduping all the video, and is screwing up the credits (leaving them jerky). I end with a timecode that covers every part of the video as 24 fps - 17 fps, but the credits do not remain at 29 fps. Am I missing something?

This is my script for mode 6:

Load_Stdcall_plugin("yadif.dll")
SetMemoryMax(512)
mpeg2source("C:\KIBA_91309\VIDEO_TS\VTS_01_1.megui.d2v")
AnimeIVTC(mode=6,aa=4,bob=3,edimode="yadif",MT=true,degrain=2,omode=2,mix=4,istart1=83766,iend1=89161,estart1=0,eend1=41882,estart2=44581,eend2=44911)

And this is my script for mode=7 pass=1, to do the video analysis pass

AviSource("losslessrendering.avi").AnimeIVTC(mode=7,pass=1,denoise=true,block=8,showmetrics=false,hybriddup=1.1,vidthresh=1.1)

And for the next pass, I just change pass=1 to pass=2.

So, any way the dedupping doesn't screw up the credits?

:thanks: