|
|
IVTC, deinterlacing, postprocessing, etc...
Hey guys,
I am trying to wrap my head around some of this deinterlacing/IVTC/force film/etc.. stuff. I've read through a lot of posts and guides and I think I have a fairly good understanding of it, although I feel as though my head is about to explode.
I am in the process of ripping most of my store-bought DVDs to xvid for playback on my HTPC. For the most part I've been using quot;Force Filmquot; since most DVD's seem to fall in this category (97% or higher film). I have noticed that some have some combing effects on scene transitions or camera transitions. The duration of the combing is short so it's not so bad. I also have a few DVDs that seem to have interlaced frames randomly stuck in the middle of the movie (The Godfather for instance). I just used GK's standard options for IVTC on this. I haven't gone back to look at the result yet, but i figure I might get some choppiness on the telecined area(s).
Here are my questions:
1) For getting rid of minor, inter dispersed telecined frames should I just always add FieldDeinterlace(full=false) to my avs script? Will this create any adverse effects if the source is purely progressive?
As an alternative can I add FieldDeinterlace(full=false) as a ffdshow postprocessing filter to account for random interlaced frames for all sources without any adverse side effects?
2) What is the difference between IVTC on a progressive (film) source and using the quot;force filmquot; option in terms of quality? Why can't I just use IVTC all the time as it seems to be better suited to handle some minor interlaced frames (decimate mode 1 or 3)? It seems that it is always recommended to use quot;Force Filmquot; if you can, but I can't find any specific information as to why this is better than IVTC.
Thanks!
Edit*I was thinking that maybe I should've put this is a more general forum since it is not specific to Xvid.. feel free to move it*
-gt; Avisynth Usage forum?
oh you already noticed
FieldDeinterlace is the proper solution to that problem, though I think it's better to use TDeint:
tdeint(mode=0,full=false,cthresh=12,type=2,map=0)
which should be faster, safer, and better looking.
Normal IVTC is just slower and a little added complexity compared to force film. If it wasn't edited after telecining force film is perfect, but they do minor editing sometimes. It's safest to always ivtc unless film is 99-100%, or use this sort of workaround when it's a little lower. Hybrid mode decimate is meant more for entire swaths of video being pure interlaced, and generally should be identified first so you can process it separately; it can make everything a lot more complicated otherwise.
Thanks for the feedback. That settings does make it look a bit better. I still do see some problems in some areas, but I may just have to accept that. Here is a very small video clip of what I'm talking about.
d.aspx?id=1...305N7TGMX3VNVZ
In case the above doesn't work:
upload/133858.avi
You can see the combing effect at the scene transition.
If I change the threshold to 6 it will pick up the combing (I can tell using map=1 or map=2) but it doesn't seem to want to get rid of the effect. Is there anything that can be done in this case?
Thanks!
Was this resized in any way before uploading it? Because if not, those are the artifacts of a vertically-resized interlace, and no normal deinterlacer will work on those. I'd say fix it by either blending the prior and next field, or interpolating it with motion compensation. If it happens much you'd probably have to use a special detection (after the main deinterlace):Code:
corrected=bob()
tdeint(mode=2,full=false,cthresh=15,edeint=corrected,map=0)
That's not quite as nice as motion-interpolating the actual jacked up fields, but it's simpler and a lot better than what you have now.
Thank you! I will try these settings when I get home from work.
The video was originally lanczos resized during the encoding process but it wasn't touched after that except to cut out the segment that I uploaded.
link isn's working for me.
As foxyshadis says, always deinterlace before resizing. The order is these operations is key.
Get it from source again. Scrap your avi and upload part of the vob, if at all possible, along with your current script. Then I could probably give you a more useful suggestion. |
|