I broke telecide

[lord_terabyte]

I've encoded a few 100 DVDs with xvid which has brought me to a point where I am far from expert but sometimes I can ask intelligent questions. And maybe I've found an interesting bug.

I recently started using linux (opensuse 11.1) and I am running windows xp in VirtualBox so I can use Gordian Knot. I started getting artifacts in my rips and eventually I isolated their occurrence to the presence of the telecide filter.

http://i1022.photobucket.com/albums/af341/bubble_belly/Clipboard01-22.png

http://i1022.photobucket.com/albums/af341/bubble_belly/Clipboard02-21.png

http://i1022.photobucket.com/albums/af341/bubble_belly/Clipboard03-21.png

http://i1022.photobucket.com/albums/af341/bubble_belly/Clipboard04-19.png

http://i1022.photobucket.com/albums/af341/bubble_belly/Clipboard05-15.png

http://i1022.photobucket.com/albums/af341/bubble_belly/Clipboard06-5.png

http://i1022.photobucket.com/albums/af341/bubble_belly/Clipboard07-1.png

http://i1022.photobucket.com/albums/af341/bubble_belly/Clipboard08.png

I have been substituting tdeint(mode=1) for telecide when I needed to IVTC. I don't know if that was a good idea or not, but the results have been good enough I guess. I just found out about tfm().tdecimate() so I'll give that a try. I know ultimately my solution is to bite the bullet and learn how to encode without Gordian Knot.

[Inspector.Gadget]

In my decidedly non-expert opinion, your problem is a result of blending rather than a problem with telecide. Also, deinterlacing after field-matching is rarely useful.

[lord_terabyte]

After re-installing Gordian Knot and Avisynth more than once, trying different versions of xvid and different settings, and deleting the virtual machine and even Virtual box, I realized I was only having problems on rips that were IVTC'd. Removing telecide stopped the artifacts.

I made 2 rips last night with tfm().tdecimate() that are very nice, so my problem is solved I think.

Actually, I've had a couple more problems inside the virtual machine. srestore sometimes produces a rip of about 28.97 FPS but mrestore works properly, and I had a clip where tdeint followed by telecide produced an encode of about 48FPS before I decided the clip was indeed a transcode that needed to be restored to 25 FPS.