VidFIRE-type function?

[segfault]

I'm processing some old Doctor Who episodes from NTSC VHS for transfer to SVCD, and I'm looking for an AVISynth plugin or function that can duplicate the function of the BBC's VidFIRE process. Steve Roberts has already explained the process better than I can:

http://www.purpleville.pwp.blueyonder.co.uk/rtwebsite/vidfire.htm

What a mess. 25fps PAL vadeotape transferred to film by pointing a camera at a monitor, then telecined for NTSC release. Can this material even be IVTCed? I have tried it with Telecide(guide=1,post=false,chroma=true).Decimate(cycle=5,mode=0,quality=3) but there still seems to be some combing.

So is such a thing possible at all with AVISynth?

[Guest]

If you turn off postprocessing it's not surprising that you get some residual combing!

Make a source clip available for us to examine if you hope for improvements to your process to be offered. It will make a good challenge for new generation Decomb (nearing alpha release).

[Chez_Wimpy]

You might try rePAL, which is specifically tuned for reconstructing PAL from NTSC transfers. However, going from PAL - which (as the restoration team has described) came from kinescoped transfers, freezing both fields into one "frame" - back to interlaced NTSC in a way that improves upon the previous efforts... I am not sure. My guess is that a method does not exist to recover the original fields from the 405line TV standard (as most of the kinescoped episodes are pre-PAL) because this would involve "mapping" the distored geometry of the moniter used for playback. Optically reverse this process during the telecide, record the input with a 405 line camera, then process it digitally. Otherwise "interpolating" software such as vidfire is the only other option. Looking at the results of some of the recent DVDs (Seeds of Death for example) the interpolation process is stunning, and wholy convincing.

Of course, other strange Dr. Who related findings have occured, like recovering part of the color signal from B&W kinescope film transfers of early lost Pertwee tapes. Those kinescopes, when converted to video, still trigger the PAL color decoder, and Pertwees cape appears RED!?!? That one still confuses me.

-Chez

[bobsobol]

I have to say I've been looking for similar functionality in AVISynth, for much the same reason. (BIG Who Fan) ;) But also for other old BBC archives, Steptoe, Hankock etc.

It's my feeling that the NTSC telecine would confuse the issue far too much, and would have to be correctly reversed before vidFIRE type correction, but vidFIRE (from the descriptions I've seen) isn't as complex as Chez suggests. (I don't think you intended that suggestion as strongly as it seems on paper... or screen lol) It simply takes one frame and assumes it is two images (like a double exposure) which is pretty much what each kinescope frame is. Then it makes a further assumption that the two images in the frame contain the same objects moving somewhere between the two images in the previous frame and the two images in the next frame. It then decides how much of which intensities are the first frame, and how much are the second by means of motion estimation and compensation, and produces 2 frames from every one.

The people working on motion compensated temporal softening and frame rate change could probably give a better idea of how practical this would be to perform as a AVISynth plugin.

It would be nice to see however, not only from the point of view of restoration of old kinescope material, but also from early tape to digital dubs, where blur deinterlacing was commonly performed. Particulaly on VHS to VCD conversions, where the blur was more from resizing without taking any notice of the interlace fields. And the result is almost identical to kinescope footage, except if you scan film at a high enough resolution (HD res for example) some idea of the fields remains, tho blurred. This is what Chez referes to in his last paragraph. But much recovered footage is already in video format from the kinescope transfers so scanning at higher resolution is not possible.

So boffins, we've postulated. What can your genius minds come up with for us. ;)

-Ron

[Guest]

I'll ask once again. Can you make available an unprocessed source clip?