View Single Post
Old 13th April 2004, 09:32   #16  |  Link
violao
Registered User
 
Join Date: Feb 2004
Posts: 252
@gvm - glad to hear you're making progress

@Richard - I didn't get that about Sony Hi-8. I have Sony Hi-8 camcorder and it has both manual focus and manual shutter speed (exposure), or is it aperture - who knows. Probably both. It appears that exposure doesn't change automatically when I set it to manual and move the camera towards darker or brighter objects, at least that's what I can see in the viwefinder. I can also confirm that all auto-exposure programs (like those for daylight, spotlight, mountains, etc.) are turned off once I set manual exposure.

It's certain that camcorders are limited to minimum shutter speed not greater than the length of a single field (20 ms in PAL) - that corresponds to 1/50th of second in tradditional shutter speed scale. With that shutter speed it's almost certain that camcorder will occasionally get 2 consecutive film frames on a single field. I can confirm that this happens and it's very annoying. I still can't figure out the reliable way to get rid of such fields/frames.

EDIT: After reading through some other posts I realize that this is known as "blending fields effect". I played with Donald Graft's Decomb and MultiDecimate in order to get rid of this and find that there are some duplicates still left after decimation, as well as some of the frames are deinterlaced by Telecide, so that they clearly differ from the original. Have not tried restore24 script by Didée yet. @Richard, did you manage to get rid of blended fields with your method(s)?

As for deflicker, I don't believe this is a way to go. I believe the flicker may be caused by recording half black lines, therefore missing fields, or even missing complete frames. We should be able to record original projected frames as seen by our eyes, not fabricate them with software.

EDIT #2: I've done some searching on "blended fields" effect and it appears that the biggest problem is having 2 consecutive top or bottom field blended from 3 different frames, like:
Code:
A  AB BC C ...
a  b  c  d ...
In this example it is impossible to restore B field. Using some simple math it can be shown that there is a minimum frame rate for this effect to happen and that one is directly related to the duration of blanking interval in projector (when the frame advances) and video field length of course (20 ms for PAL). For example if blanking interval is 1/5th of complete frame interval then maximum frame rate for fail-safe capture is 20 fps. This assumes that camcorder is allowed to set maximum possible shutter speed (1/50 s per field for PAL). If shutter speed was larger then max frame rate would increase.

Last edited by violao; 14th April 2004 at 11:31.
violao is offline   Reply With Quote