Log in

View Full Version : Black vertical lines


mg262
6th December 2005, 14:14
I've found these on a DVD:

http://people.pwf.cam.ac.uk/mg262/posts/black_vertical_lines.jpg

or PNG: http://people.pwf.cam.ac.uk/mg262/posts/black_vertical_lines.png

One of those vertical lines persists for most of the episode, whereas the other is only found on this frame and nearby frames. The lines always stay in the same place, even though the underlying picture wobbles -- so presumably whatever caused them happened at quite a late stage in the history of the source.

Does anyone know what might have caused them? Ideas on how to remove them in script would also be welcome... the best I can think of is to isolate them with a mask and replace them with a horizontally blurred picture, but I'm sure there's something more intelligent. But it's the cause that is particularly annoying me!

Mug Funky
6th December 2005, 15:13
the film was scratched by the film transport at some stage (probably after copying, otherwise it'd have greater registration errors).

there's a film-slang term for it, but i've no idea what it is.

as for removing it, i don't really know. motion-compensation goes some of the way, but it comes back when the movement stops, obviously. i've got crap like this on some of my astroboy transfers.

inpainting would be useful, and possibly some kind of edge detection (hint: it's always vertical and always in the same spot, and goes for the full height of the frame, so it should be easy to detect :)).

i'm sure filling it with some kind of median of neighbor pixels would do the trick, then maybe some artefact prevention if necessary.

Didée
6th December 2005, 15:33
Try w=last.width()
h=last.height()
turnright()
lanczosresize(h*2,w)
assumetff().separatefields()
removegrain(4)
assumetff().weave()
lanczosresize(h,w)
turnleft() and use this with MaskedMerge and a suited mask. Not tested, braindead errors excepted ... but you get the idea.


@ MugFunky - your PM box is bursting.

Fizick
6th December 2005, 17:42
Try my Descratch plugin

mg262
12th December 2005, 11:04
Thanks for the ideas all! I shall experiment with them and see what happens.

The black lines had puzzled me because I had assumed that film strips were oriented like comic strips -- i.e. that the pictures were put together using stackhorizontal. Reading your answer, it seems that it is actually stackvertical, which would explain why the line was always in the same place...

foxyshadis
12th December 2005, 13:36
Horizontal is still camera. Vertical is motion camera. ;) (Although of course this varies between formats.)

Utterly off topic: A filter to simulate the effect of a frame sticking and burning out would be neat. =D Happened the first time I saw fellowship of the ring, and it's a really trippy effect.

Serg Belyansky
12th December 2005, 18:23
Film people call those "tramlines". As all defects persistent between frames, this one is basically impossible to remove without introducing afrtifacts.

Ebobtron
12th December 2005, 20:14
Hi all,
Scratch the negative and the lines would be black but they would have to wobble also.

Scratch the print and the lines are white but they would move with the frame also.
Sorry, it is a bad scan with a bad line scanner, or whatever they are called. Underlying cause, $$$.

Ebobtron

Serg Belyansky
13th December 2005, 17:47
Ebobtron, negative marks are white. Release print / positive marks are black.

Ebobtron
13th December 2005, 19:09
Ebobtron, negative marks are white. Release print / positive marks are black.Serg,

If I take my pen knife out of my pocket and scrape off the emulsion from the print film, the scratches I make with the knife let more light though to the screen, just like the scratches the print film collects over time do.

If I scratch the negative I let more light through to expose the print film making it darker when developed. Otherwise why would we call it a negative.

If you want to continue this debate pm me and we can move this to general topics or something.

:)

Ebobtron