View Full Version : Judder issue found in NTSC R3 source
PeRFeCK
18th October 2008, 23:13
Hi All,
I need some help with a NTSC R3 DVD source with a up/down judder every time the camera shot changes/cuts to a different angle.
Here's a quick 50mb sample of what the encode looks like. Please take a look and let me know if anything can be done, I have now found this to be in 2 of my R3 DVD sources.
http://rapidshare.com/files/149197678/enctest.mkv.html
You will notice the effect I am describing in the following areas: happens twice between 6-9 seconds, once at 24-25, 31-32, 1:03 - 1:04.......etc. Note that I used SelectRangeEvery(2400,120) when trying this out so the effect only occurs on a true camera cut.
Does anyone know of some sort of filter I can use to fix this judder?
Thanks in advance!
setarip_old
19th October 2008, 00:35
Hi!
Too obvious a question, perhaps, but do the original DVDs exhibit this same behavior when played on the same system?
PeRFeCK
19th October 2008, 00:39
hi :) i'm definitely seeing it on the VOBs and I do remember getting a headache when playing it back on a standalone dvd player :)
setarip_old
19th October 2008, 02:59
i'm definitely seeing it on the VOBs and I do remember getting a headache when playing it back on a standalone dvd playerReturn the DVDs to the place of purchase...
PeRFeCK
19th October 2008, 03:03
:X its an official Korean DVD and I've now seen this on two different R3 korean movies - Beautiful and Sex is Zero. so there's nothing that can be done? :(
thetoof
20th October 2008, 03:08
Can you upload a 10mb (max) m2v sample where the problem can be seen? (dgindex-select a small section-save and demux video)
PeRFeCK
20th October 2008, 03:30
sure, thanks for the reply
here's a 8mb rar that has 2 really short m2v's
http://rapidshare.com/files/155691900/Desktop.rar.html
thetoof
20th October 2008, 03:52
Maybe the "judder" you're talking about is simply non-ivtced material, as a simple "forced film" worked perfeclty fine for me (+ clicking "yes" to correct the field order transition).
Here (http://www.doom9.org/ivtc-tut.htm) is some (dated, but still good) reading.
Edit: oh wait, I see the issue now. Try this stabilizing script I took from g-force a little while ago:
orig=mpeg2source("forced film.d2v")
temp=orig.TemporalSoften(7,255,255,25,2)
rep=Repair(temp,orig,mode=16).TemporalSoften(1,255,255,25,2)
source=Interleave(rep,orig,rep)
mdata=DePanEstimate(source, range=1, trust=0, dxmax=1, dymax=1)
DePan(source, data=mdata, offset=1)
SelectEvery(3,2)
PeRFeCK
20th October 2008, 04:16
hmmmm...either I don't know how to do what you're telling me or you aren't seeing what i am :)
in DGIndex i changed Video > Field Operation > Forced Film
i'm still seeing the judder in between camera shots. it happens very quickly.
i have seen the "would you like to correct field order" before on other movies but it doesn't seem to be popping up for me on this. am i missing something fairly obvious?
thetoof
20th October 2008, 04:43
Maybe the field order thingy was introduced when extracting the m2v... Anyways, as you can see in my edit, I noticed the judder after posting and included a script. Have you tried it?
PeRFeCK
20th October 2008, 05:33
did not see your edit. sorry I'm a bit of a noob and can't get that script to work. i keep getting "DePanEstimate: Can not load FFTW3.dll!" on line 6, i have that dll in the plugins folder so i'm not sure whats up.
thetoof
20th October 2008, 05:36
It must be in your system32 folder.
PeRFeCK
20th October 2008, 05:40
still not working :X im running vista x64 if that makes a diff :(
thetoof
20th October 2008, 06:04
I think it's another folder for Vista... run a search on your HDD to find avisynth.dll and it should be in that folder (can someone else confirm?).
PeRFeCK
20th October 2008, 06:13
ok i threw it into a few folders in my windows directory after reading some post about it on google, got the script working now :)
i can't say for 100% sure but during playback (before encoding) the judder still appeared although it seemed to be a tad less, and it might have added some artifacts.
Leak
20th October 2008, 09:02
i can't say for 100% sure but during playback (before encoding) the judder still appeared although it seemed to be a tad less, and it might have added some artifacts.
Ummm... I didn't download your sample, but if this is an older, non-digitally produced anime that "judder" are simply artifacts at scenechanges caused by different films being spliced ("taped") together in the production process.
In that case you most probably want a script that detects scenechanges and replaces the frame at the scenechange with the one after it.
PeRFeCK
20th October 2008, 13:04
this is a live action movie who's dvd was produced in 2008, a few months ago. it spans much longer than one frame so they cannot be removed
Mug Funky
21st October 2008, 04:46
is this like a kind of vertical stretching? (you can see it in the woman's face on the smaller sample you sent, just after the scenechange).
if that's the case it is indeed a splice causing it, but a different kind of artefact.
most telecine machines run in constant motion (ie the film never stops, and it's motion provides the vertical scan in the final picture). the beauty of this is that it is very gentle and the film has less chance of damage. this kind of film transport however is tension based - pull on the film or the tension take-up rollers and the film will wind faster.
now a cement splice is a big chunk of glue between different bits of film. when this chunk goes over the tension take-up thingies, the frame currently being scanned will wobble for an instant as the machine corrects the tension. this shows up as a small vertical stretching in the middle of the affected frame.
if a splice is really bad, and sticks out the side of the film, then the frame will move side-to-side briefly in the gate. this is evident on the baseball shot there. the splicing crap pushes against the rollers and takes the film with it. in an extreme case the film can actuall jump upward and the picture gets blurry, and in really extreme cases the film can jump out of alignment on the rollers and scratch up against them, putting a thick white vertical line in the film (that is there permanently). when this happens it's time to stop the machine in panic, ring up the lab and abuse them for being careless in their splicing.
@ setarip:
returning the DVD wont help. it'll be that way on the master tapes, possibly also on the inter-negative that the release prints were struck off.
[edit]
look at those white dots! the DVD must have been transferred off an interneg. that's a dumb way to do things in the age of digital intermediates.
PeRFeCK
21st October 2008, 05:34
interesting, thanks for the info behind it. you're right it is too bad because i found watching the movie quite annoying and it doesnt sound like there's a way to fix it. thanks for trying to help guys
setarip_old
21st October 2008, 05:50
According to Amazon, the "Sex is Zero" DVD is:Completely Re-Mastered Video & AudioThis makes it hard to understand why the quality of your original DVDs would be so problematic...
PeRFeCK
21st October 2008, 05:54
we are talking about different versions of the movie :) i was talking about the region3 (original korean) release of Sex is Zero. the R1 did not seem to have that but was DVD5 only, also im no longer in posession of that anymore to check...but i found a special edition of R3 which i'm crossing my fingers wont have it.
as far as "beautiful" goes though, that just came out on DVD a few months back and there is only one version.
g-force
21st October 2008, 20:39
Untried with your source, but should work. You will need to implement a scene change detector, or manually choose the output of the following filter only on the frames that are bad.
Interleave(last.TemporalSoften(7,255,255,25,2),last)
DePan(last,data=DePanEstimate(last,trust=0,dxmax=0,dymax=100),offset=-1)
SelectEvery(2,0)
-G
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.