Telerecording/kinescope lag

[Floatingshed]

Can someone please suggest how it may be possible to reduce the lag/smear exhibited by telerecordings? i.e. films made by pointing a camera at a tv monitor.

This problem varies greatly between recordings, some have almost none. The one I'm working with at the moment has buckets full!

I know a sample would be helpful but that is not easy at present. I know a couple of forum users are familiar with telerecordings, I hope that they can help.

Thanks.

[johnmeyer]

You need to post a few seconds in order to show what kind of smear you are dealing with. Dideé posted a great solution yesterday to one type of "lag/smear,"

Deblending again, need some help (http://forum.doom9.org/showthread.php?t=161224)

but his solution to that problem may not have any relevance to your problem.

A "real" Kinescope has many issues, and some of the real early ones, shot with 1950s cameras, do sometimes exhibit ghosting. I have similar problems with video I shot with my first 1981 consumer camera, an RCA "Vidicon" tube. It lagged and smeared like crazy. So if this is the problem you have, and there is some sort of solution to suppress these artifacts, I'd love to see them.

So, try the solution posted above, and if that doesn't work, post a few seconds of your video.

[Didée]

Very probably, the script mentioned script in the "deblending again" thread will not work for a telerecording.

I've close to zero practical experience in that field, but the technical implications are quite obvious. For a good telerecording, it is ultimately necessary that the recording device is carefully synced/locked to the refresh rate of the display device. If the recording has been done "freely" without any syncing/locking, then blending of two frames will appear very frequently.

- If the display device was a purely progressive display, then there will be a relatively "clear" blending pattern. This case eventually can be processed reasonably. (Good chances if the recording framerate was at least 2x the source framerate. Else, chances get poor.)

- If the display has shutter blackouts (film projector), then the blendings will include that black. (Very difficult, not much hope for a good result.)

- If the display device was a CRT tube, then the blendings will include frame tearing. (dito)

[Floatingshed]

Thanks for the help so far...
Here is a sample: http://www.mediafire.com/?i5ebpq4jsneqdp6
It is cut directly from the .vob, no processing of any kind. It is only a handful of frames but the problem is very clear if you look at the hands.
I guess that this is caused by the blending of two fields onto one film frame and cannot be improved but I'm hopeful...
Thanks.

[johnmeyer]

I see what you mean about the hands, but since they are moving, some blurring is to be expected. The video looks quite pristine to me, and not in the need of much help. Exceptionally excellent exposure for a Kinescope, if that is what it really is. Some variation of Didée's recent script might work, although I'll admit to still being a little stumped about how his sharpen mask was used to remove the ghosts in that sample a few days ago. Pure "FM" as far as I'm concerned (the "M" stands for Magic, the "F" stands for, well ...). You might try tweaking parameters in that script and see if you can get a little improvement. I tried the script with the parameters that were tuned for the other clip, and it didn't do much. I don't have time right now to take this further.

[Floatingshed]

Yes its most definitely a kinescope, telerecording actually as it is British. It is a 16mm film recording of a 625 line colour VT source. The original tape was destroyed, only this neg remains. From the few comparisons I have made I would say that British telerecordings seem to be better quality that US kinescopes, a different process maybe or something to do with the frame rate...

I reasise that with movement comes blurring but this leaves a kind of trail too. I tried the script mentioned and it made matters much worse!

[Didée]

Sure, the script as-is can't work here. The other case was an interlaced source, hence the vertical sharpening effectively was a "temporal sharpening" operation. This source here is progressive, hence the fitering would have to be changed to a true inter-frame temporal operation.

What makes me wonder is that the blending/smearing seems to be mostly in areas with relatively low contrast. If you look e.g. at the movement of the women's shoulder (relatively high contrast between cloth and background), there is not any sign of blending or smearing. Judging by this, it ~could~ be the original recording was rather clear, and the damage was done by postprocessing with a temporal smoother.

[Mug Funky]

yep, looks like a DVNR in the telecine chain.

it isn't too bad though.

removing the ghosting looks like a mission and a half. anything i can think of to kill it would probably increase the noise level by a large amount.