Log in

View Full Version : Did I go wrong with Telecide on PAL interlaced


GrofLuigi
16th October 2005, 03:43
Hello all,

The story: I suspected that the capture driver (BTwinCap) or my VCR shifts the field order on dropped frames (my setup is quite all right and I get no dropped frames without a reason - usually I could see a tape glitch that caused the drop). After such a drop, the following (up to 10) frames are somehow 'different' than others - i.e. the interlacing 'jaggies' are not so well defined and 'sharp', but kind of 'triangular', and the picture is fine, although kind of blurry. A screenshot would explain it better, but at the moment my HDD is filled with the filtered video, and when I fix this problem, I will make another capture and post if necessary (no requests for avi's, PLEASE, I'm on dialup!)

To cut to the chase: After capture, the video was cleaned by the following script (Avisynth 2.56 RC2):

---------------

AviSource ("d:\video\capture.avi")
KillAudio ()
Crop (16,0,0,-16, align=true)
AssumeTFF() # as a precaution, but obviously didn't help
Telecide (guide=2, post=0, show=false) # I guess this is the culprit
DeGrainMedian(limitY=2,limitUV=3,mode=1,interlaced=true)
DeGrainMedian(limitY=2,limitUV=3,mode=1,interlaced=true)
ConvertToYUY2 (interlaced=true) # captured with PicVideo v3.0.13 and it's YV12!
AddBorders (0,0,0,16)
BilinearResize (352,576)

----------------

, saved as avi and sent to CCE for encoding.

The problem: The field dominance changes! About 1 minute it's TFF, then BFF and so on... (verified with AssumeTFF().SeparateFields()). The header (or whatever it is) says it's TFF and it doesn't change (verified with Info ()). Same with the MPEG. The original doesn't change field dominance.

The stupidity: I didn't check the result thoroughly (or maybe have just hit some good parts) and deleted the original capture.

The dissapointment: I still haven't discovered the behaviour of my setup on dropped frames or how to treat them.

The irony: Trying to avoid one particular problem created it where probably it wasn't.

The question: Can I get away without recapturing? I have the filtered avi and the MPEG made out of it, but no original. Some header trickery, or avisynth, or... I returned the tapes to the friend whose it were (this is his wedding video) and promised him a DVD in a day or two. :) How to avoid making fool of myself?

Thank you!

GL

Guest
16th October 2005, 04:42
If it's a wedding video, it's probably pure interlaced, which means that Telecide should not be used! Telecide is for progressive content only.

Is the field order reversal in the input before any filtering? If not, just make an interlaced DVD. If yes, try deinterlacing with a deinterlacer that is not field order sensitive, such as FieldDeinterlace().

GrofLuigi
17th October 2005, 00:03
If it's a wedding video, it's probably pure interlaced, which means that Telecide should not be used! Telecide is for progressive content only.

Yeah, I found out afterwards in Decomb's tutorial (and only there) that telecide (guide=2) should not be used on interlaced material. Otherwise, on other places, mainly this forum, it is usualy mentioned that telecide is good for PAL "to maintain phase lock", but rarely about interlacing. I guess I wasn't paying attention enough and I will declare this as a failed experiment.

Is the field order reversal in the input before any filtering? If not, just make an interlaced DVD. If yes, try deinterlacing with a deinterlacer that is not field order sensitive, such as FieldDeinterlace().

It's not in the input, but I didn't keep the input. :mad:
Is 352x576 PROGRESSIVE a valid DVD resolution?

Thank you,

GL

Mug Funky
17th October 2005, 06:56
yep, interlaced and progressive in DVD is usually a formality.

most stuff is encoded interlaced even if it's progressive because it's better to do that than to encode interlaced stuff as progressive (yuck).

so half-D1 progressive is fine. if your source is 720 wide, make sure to crop 8 off the left and right before resizing to 352.