Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Capturing and Editing Video > Avisynth Development

Reply
 
Thread Tools Search this Thread Display Modes
Old 1st May 2006, 18:01   #621  |  Link
foxyshadis
ангел смерти
 
foxyshadis's Avatar
 
Join Date: Nov 2004
Location: Lost
Posts: 9,173
If your system can't handle realtime tdeint, try leakkerneldeint. It's much simpler and faster, even though it tends to have more issues than tdeint.

A couple of ways to speed up tdeint are, use mtnmode=0 or 2, denoise=false; you should try mtnmode first.
__________________
There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order. ~ Ed Howdershelt
foxyshadis is offline   Reply With Quote
Old 1st May 2006, 21:34   #622  |  Link
tritical
Registered User
 
Join Date: Dec 2003
Location: MO, US
Posts: 999
@Alizar
As foxyshadis said, the file is created in tdecimate's constructor, but the data isn't actually written into the file until the tdecimate object is destroyed (happens on script close). The main reason for doing it that way is that it ensures the metrics will be in order even if frames aren't requested in linear order. That point doesn't really matter for reading the data back in, but it helps if you want to manually examine the stats file. I could change it to write to the file each time a frame is processed, but I don't see any real advantage... if you want to see the results for a frame immediately use the display option.

@JaunC
An exclusion band for tdecimate is a good idea... though it's a little more tricky to do than for tfm. The other issue brought up by your example is that the y0/y1 bands in tfm currently only effect field matching comparisons and not combed frame detection which doesn't make sense. I'll change that in the next release.
tritical is offline   Reply With Quote
Old 1st May 2006, 21:40   #623  |  Link
foxyshadis
ангел смерти
 
foxyshadis's Avatar
 
Join Date: Nov 2004
Location: Lost
Posts: 9,173
Alizar told me offline that it actually wouldn't write the stats even after closing, even after multiple open-play-close cycles, so I'm afraid it was something else. No error message came up.
__________________
There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order. ~ Ed Howdershelt
foxyshadis is offline   Reply With Quote
Old 2nd May 2006, 00:37   #624  |  Link
tritical
Registered User
 
Join Date: Dec 2003
Location: MO, US
Posts: 999
I take it that there isn't a set of steps that always cause the problem? I've personally never had it happen. The only thing I can think to do is put up a special build with some extra debug output to see what happens. I'll try to put one up later tonight.

EDIT:
Test Build

Just have debugview open to catch the output strings and post the log here or pm it to me... it should show why it is failing.

Last edited by tritical; 2nd May 2006 at 00:58.
tritical is offline   Reply With Quote
Old 2nd May 2006, 01:57   #625  |  Link
johnmeyer
Registered User
 
Join Date: Feb 2002
Location: California
Posts: 1,770
Say, if you're going to do anything with extra debug output, is there any chance that you could include a "verbose" function for the output file that would include the MIC values? Currently I am having to output via the debug mode and then parse that file to get both the p, n, c decision made by TFM, along with the MIC values. In my application, I have created an external parsing application that looks at the MIC values and uses that to determine where to override the TFM decisions. This is not a huge deal, but it would save me some extra work of parsing the debug file. What I am proposing is taking the current TFM Output file that looks like this:
Code:
31	n	-	[56]
32	c	-	[8]
33	c	-	[0]
34	n	-	[3]
and add the MIC values:
Code:
31	n	-	[56]	96	83	56	0	104
32	c	-	[8]	104	8	71	56	61
33	c	-	[0]	61	0	20	71	91
34	n	-	[3]	91	79	3	20	56
johnmeyer is offline   Reply With Quote
Old 3rd May 2006, 02:40   #626  |  Link
JuanC
Registered User
 
Join Date: May 2002
Posts: 220
Quote:
Originally Posted by tritical
@JuanC An exclusion band for tdecimate is a good idea... though it's a little more tricky to do than for tfm. ...
It's great you like the idea. I hope there could be a chance for it in the future.
Quote:
Originally Posted by tritical
@JuanC ... The other issue brought up by your example is that the y0/y1 bands in tfm currently only effect field matching comparisons and not combed frame detection which doesn't make sense. I'll change that in the next release.
You know, I found it convenient, so that I can use PP=6 to deinterlace the subtitles at the first and/or last frame(s) where they appear. Is there a chance 'combed frame detection' could be optional for the exclusion band ?

Thanks for your comments, Juan
JuanC is offline   Reply With Quote
Old 3rd May 2006, 07:32   #627  |  Link
tritical
Registered User
 
Join Date: Dec 2003
Location: MO, US
Posts: 999
@johnmeyer
Yep, I can add that.

@JuanC
I'll make using the exclusion band in combed frame detection optional.

I'm still busy working on stuff for school, but should finally have free time next week to do some work on tivtc.
tritical is offline   Reply With Quote
Old 7th May 2006, 03:21   #628  |  Link
swaaye
Registered User
 
Join Date: Jun 2004
Posts: 144
either TFM or TDecimate is crashing on me for the first ep of Stargate SG-1. It will run for a few mins of the D2V but then the encode will just stop and Staxrip will move to the next pass and this will stop out too. I switched to telecide/decimate and that works so it must be a TIVTC bug. Can I send anything to help with this?

I've also been having a problem with a interlaced frame sneaking in after every scene change. Telecide/Decimate seem to not have this problem. I read that it could be a frame order switch problem with the DVD.

Code:
tfm(d2v="%source_file%")
tdecimate(hybrid=1)
vs. (didn't bother setting it up properly for hybrids cuz it isn't very good at that anyway)
Code:
Telecide(guide=1).Decimate()

Last edited by swaaye; 7th May 2006 at 03:23.
swaaye is offline   Reply With Quote
Old 7th May 2006, 04:18   #629  |  Link
tritical
Registered User
 
Join Date: Dec 2003
Location: MO, US
Posts: 999
If you could post a clip that causes a crash it would help. Otherwise, I have no idea where to start looking for a bug. Also, does it crash if you open that script with only tfm+tdecimate in vdub and encode? If so, then try taking out tdecimate and see if it still crashes (would at least narrow it down to either tfm or tdecimate).

If combed frames are getting through then you need to tweak the combed frame detection settings (MI, cthresh)... using display=true in tfm will help.
tritical is offline   Reply With Quote
Old 7th May 2006, 05:34   #630  |  Link
swaaye
Registered User
 
Join Date: Jun 2004
Posts: 144
Well it's TFM that's crashing. With just MPEG2Source and TFM in the filter stream, the clip will crash within 20 seconds of the show. I'm not really sure how to send you a part of the VOBs though....

I don't know if this is useful or not, but DGIndex says "Warning! Opening GOP is not closed. First few frames may not be decoded correctly". I've ripped the DVD a few times and it always has this error. I also tried using Trim to force TFM to work on a piece deeper in and it still crashes in that area too.

Last edited by swaaye; 7th May 2006 at 05:38.
swaaye is offline   Reply With Quote
Old 7th May 2006, 09:05   #631  |  Link
tritical
Registered User
 
Join Date: Dec 2003
Location: MO, US
Posts: 999
You can upload the vob to my ftp:

ip/port = 68.184.14.253:17252
user/pass = upload/upload

If you want to you can cut a chunk from the vob using dgindex (save project and demux video) or just stop the transfer after enough has been uploaded to reproduce the problem.

The open gop warning from dgindex wont effect things.
tritical is offline   Reply With Quote
Old 8th May 2006, 02:05   #632  |  Link
swaaye
Registered User
 
Join Date: Jun 2004
Posts: 144
I uploaded a test set of D2V, M2V, and AVS to your FTP. swaaye prefix on the files. It will crash Vdubmod very quickly on you.
swaaye is offline   Reply With Quote
Old 8th May 2006, 04:03   #633  |  Link
tritical
Registered User
 
Join Date: Dec 2003
Location: MO, US
Posts: 999
It doesn't crash here on my comp so I've got a few more questions... What type of cpu do you have? If you set opt=0 in tfm does it crash? If you set pp=0 in tfm does it crash? What version of avisynth and tivtc are you using?
tritical is offline   Reply With Quote
Old 8th May 2006, 05:43   #634  |  Link
swaaye
Registered User
 
Join Date: Jun 2004
Posts: 144
Quote:
Originally Posted by tritical
It doesn't crash here on my comp so I've got a few more questions... What type of cpu do you have? If you set opt=0 in tfm does it crash? If you set pp=0 in tfm does it crash? What version of avisynth and tivtc are you using?
Setting pp=0 solved the crashing. pp=3 crashes as well. All other levels of pp are stable. Opt didn't help.

AVISynth 2.5.6a (StaxRip installs it)
TIVTC - 2006/04/09 v1.0 RC 3

I'm running a dual core Opteron 165. I've encoded roughly a bazillion () vids with this setup and I think this is the first time I've had crashing problems with your filter. Though I haven't used this version of TIVTC much.

Last edited by swaaye; 8th May 2006 at 05:52.
swaaye is offline   Reply With Quote
Old 8th May 2006, 07:57   #635  |  Link
tritical
Registered User
 
Join Date: Dec 2003
Location: MO, US
Posts: 999
Yep, it's a problem with the combination of field=0 and cubic deint (pp=3/6) due to a change made in the last release. I'll put up a fixed version tommorrow.
tritical is offline   Reply With Quote
Old 8th May 2006, 13:55   #636  |  Link
tritical
Registered User
 
Join Date: Dec 2003
Location: MO, US
Posts: 999
[link removed], changes:
Code:
TFM:
   + output file generation is now effected by the micout parameter
   + if micout > 0 then calculate/output mics even when an override 
         or d2v match is used
   - fixed a bug introduced in the last release in cubic deint pping (pp=3/6)
   - fixed a bug in one of the yuy2 checkcombed mmx routines

Last edited by tritical; 27th July 2006 at 00:28.
tritical is offline   Reply With Quote
Old 8th May 2006, 18:33   #637  |  Link
swaaye
Registered User
 
Join Date: Jun 2004
Posts: 144
... glad to help.
swaaye is offline   Reply With Quote
Old 8th May 2006, 19:57   #638  |  Link
johnmeyer
Registered User
 
Join Date: Feb 2002
Location: California
Posts: 1,770
tritical,

Many thanks for the verbose output option. Makes my job a lot easier.
johnmeyer is offline   Reply With Quote
Old 29th May 2006, 20:36   #639  |  Link
Chainmax
Huh?
 
Chainmax's Avatar
 
Join Date: Sep 2003
Location: Uruguay
Posts: 3,103
In my latest encode, using EEDI2 v0.91, TDeint v1RC7 and TIVTCv1RC4 resulted in some artifacting on several scenes. Here are some sample comparison screenshots:

MPEG2Source + ConvertToRGB:


Filtered:



Here's the script I used:
Code:
mpeg2source("X:\wherever\GNS.d2v")

ColorMatrix()

AssumeTFF()
Interp = SeparateFields().SelectEven().EEDI2(field=1)
Deinted=TDeint(order=1,field=1,edeint=Interp)
TFM(mode=6,order=1,PP=7,slow=2,mChroma=true,Clip2=Deinted)
TDecimate(mode=1)

Crop(8,2,700,466,align=true)

Levels(0,1,255,16,235)

hqdn3d(0,0,3,3)

DeGrainMedian()

aWarpSharp(depth=16,cm=1)

Lanczos4Resize(656,448)

DeHalo_Alpha()

LimitedSharpenFaster(SMode=4,LMode=3,Strength=200)

AddBorders(24,16,24,16)

ConvertToRGB()
(I put the ConvertToRGB lines in both scripts for a more accurate comparison as I'm using VDubMod)

This is much more frequent and evident in a previous high motion scene. I'll be trying just TIVTC soon, and get back with results. In the meantime, what do you think is the cause of this?
__________________
Read Decomb's readmes and tutorials, the IVTC tutorial and the capture guide in order to learn about combing and how to deal with it.
Chainmax is offline   Reply With Quote
Old 30th May 2006, 01:47   #640  |  Link
foxyshadis
ангел смерти
 
foxyshadis's Avatar
 
Join Date: Nov 2004
Location: Lost
Posts: 9,173
Try using upconv=true in the Mpeg2Source, that might be bad chroma sampling. You can convert back to yv12 progressive after ivtc.

Since it's R1 the chance of an interlaced resize is pretty slim, and the field-blending would be more noticable. That'd be the other main reason for the artifacts.
__________________
There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order. ~ Ed Howdershelt
foxyshadis is offline   Reply With Quote
Reply

Tags
tdeint, tivtc

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 06:21.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2017, vBulletin Solutions, Inc.