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. |
20th August 2002, 08:31 | #1 | Link |
Registered User
Join Date: Feb 2002
Posts: 9
|
crazy anime IVTC
ok here's my problem
i've been combing the forums for days trying to solve my strange problem, but i've yet to find any info so plz don't flame me if the answer is obvious, i tried!! NTSC, interlaced, 3:2 pulldown alrite, i'm working on some anime dvds that no one's touched yet anyways, examining some of my finished products, i noticed to my dismay that when the camera pans at a certain speed, the content looks fuzzy... curious, i ran it through nandub and noted that the fuzziness was caused by teh decombing, not doing a good job on the frames...or so i thought... when i examined it closely, i realized that it was a kind of ghosting effect that caused the bluriness well after tweaking decomb in every possible way, e.g. disabling post, adding area based deinterlacers, smart.. etc. nothing worked.. the problem still remained, i was desperate enuff that i even tried tomsmocomp, meant only for pure interlaced that whacked my progressive frames, and still didn't fix my deinterlacing so, finally when i went back to the original i noticed smthg rather shocking i'd counted 3 progressive and 2 interlaced, but during the panning scene at that 'certain' speed, i'd failed to notice, that in a set of 5, 1 was a good progressive frame, 2 were fully interlaced, and the next 2, which should've been good, the bottom halves of hte frames were interlaced so now.. i'm not quite sure what to do... by tweaking decomb, area based and smart deinterlacer in taht order i've been able to somewhat clean it up, but it still looks a little fuzzy can anyone offer some advice? thx |
20th August 2002, 19:46 | #4 | Link |
Registered User
Join Date: Feb 2002
Posts: 9
|
Kodomo no Omocha
r2 Dvds no matter what i try on the half-interlaced frames i always get a ghosting effect on the arms i've been playing with area based trying to lower the edge detect, but even only fades the lines a little so far the best setting i've found decomb with post disabled, followed by area based, and then a smart deinterlace anyone have any suggestions on how to fade the 'ghost' lines that i don't want without damaging the picture too much? |
20th August 2002, 21:04 | #5 | Link |
Asker of Questions
Join Date: Oct 2001
Location: Florida
Posts: 433
|
It's slow, but you should try the double-framerate deinterlacing process described at 100fps.com.
If that solves the problem, you'll know that you need more optimal deinterlacer settings.
__________________
"The real trick to optimizing color space conversions is of course to not do them." --trbarry, April 2002 |
21st August 2002, 08:27 | #6 | Link |
Registered User
Join Date: Feb 2002
Posts: 9
|
i'm having trouble finding that double-framerate method on 100fps
perhaps i'm just blind i've managed to reduce the problem now to 1 frame in 4 that look blurred but its still annoying to watch if there r enough ppl who r interested i'll post some screenshots OUTPINGED, what was the problem in Ranma? i also notice that in area based deinterlacer... when i lower the edge setting it preserves the wrong edge, the one form the next frame, the interlaced one, not the edge i want frustrating |
22nd August 2002, 21:43 | #8 | Link |
Registered User
Join Date: Feb 2002
Posts: 9
|
ok, in a set of 4 frames, ABCD
i've foudn that decomb, telecide/decimate post=true handles all but frame B well whereas, decomb telecide, post = false, followed by area based deinterlacer 5/30 handles frame B well, but gives the ghost on frame D is there any way to specify exactly which frames i want decomb/area based to work on? for the sample clip. would it be best to follow the svcd guide and create an mpeg2 sample? i'm not familiar with svcds.. |
23rd August 2002, 10:07 | #10 | Link |
Registered User
Join Date: Feb 2002
Posts: 9
|
i'm sorry mr. graft
i'm not sure how to cut a sample of the vobs is there a guide to that somewhere? as for the problem yes the fields r blended however if i run decomb, it messes up one frame in 4 which area based handles well is there a deinterlacer that can be loaded in avisynth that can deinterlace specified frames? for example, frames, 1, 5 or 4n+1? if it can do that it'd be great my plan woudl be to run decomb, post = false selective deinterlace then area based deinterlace that should take care of the problem |
23rd August 2002, 11:33 | #11 | Link |
Retired AviSynth Dev ;)
Join Date: Nov 2001
Location: Dark Side of the Moon
Posts: 3,480
|
Just import your file in avisynth, don't do any processing, and add:
trim(start_frame,-10) Compress the frames with huffyuv in vdub, zip it, attach here. Remeber to find a critical spot as startframe.
__________________
Regards, sh0dan // VoxPod |
26th August 2002, 07:57 | #12 | Link |
Registered User
Join Date: Feb 2002
Posts: 9
|
ok, i cut a sample and compressed with huffyUV
5 megs though so if u would like to help me, plz private msg me and i'll give u a link to my server perhaps it could be easily solved if somone could point out to me a SELECTIVE DEINTERLACER i just need one, that i can specify EXACTLY which frames it is to work on |
26th August 2002, 13:45 | #14 | Link | |
Asker of Questions
Join Date: Oct 2001
Location: Florida
Posts: 433
|
Quote:
blah=SelectEvery(parameters) DeinterlaceThingy(blah,parameters,parameters,etc)
__________________
"The real trick to optimizing color space conversions is of course to not do them." --trbarry, April 2002 |
|
27th August 2002, 20:23 | #15 | Link |
Registered User
Join Date: Feb 2002
Posts: 9
|
defiler
ur method would work great but according to the description of the select*** function its output is a clip created by the parameters of the function clip = mpeg2source(blah... temp = selectevery(clip...algorithm) deinterlacer(temp, parameters...) how do i merge the deinterlaced frames in temp back into clip? i'm going to up my sample to a different host if this doesn't fix the problem Last edited by xionx; 27th August 2002 at 21:28. |
28th August 2002, 01:36 | #16 | Link |
Asker of Questions
Join Date: Oct 2001
Location: Florida
Posts: 433
|
Damn. Good question.. The trick isn't merging the two clips, it's merging the deinterlaced clip with the original without mixing up the frame order. Nothing is coming to mind.. I'll sleep on it.
__________________
"The real trick to optimizing color space conversions is of course to not do them." --trbarry, April 2002 |
28th August 2002, 09:49 | #17 | Link | |
Retired AviSynth Dev ;)
Join Date: Nov 2001
Location: Dark Side of the Moon
Posts: 3,480
|
Quote:
__________________
Regards, sh0dan // VoxPod |
|
28th August 2002, 10:14 | #18 | Link |
Registered User
Join Date: Feb 2002
Posts: 9
|
ok sh0dan
i see where ur going with that and i think it just might work! i'm going to try it out in theory, since its only 1 frame in 4 that decomb doesn't handle so well i think i'll use SelectEvery to create 4 temporary clips A - 1,5,9...etc. B - 2,6,10, etc.. C... D... deinterlace C separately, interleave them, so that i preserve the original frame order then apply decomb to the clip with teh edited C frames perhaps it'll work... stay tuned |
Thread Tools | Search this Thread |
Display Modes | |
|
|