(2017-09-08, 10:13 PM)BronzeTitan Wrote: Right. So the 25th Anniversary release (24fps) then is from the original, progressive (24fps) source?

Yes.

What has me curious is if they made changes to release a 24p version.
It isn't uncommon for anime to have variable frame rates, including segments at 29.97 for added smoothness.

It's possible this was originally 29.97fps, but made some changes because people complained about the interlaced releases.

Anything released theatrically would be made 24fps. You'd have to speak to an anime expert with regards to TV/video productions and their framerates.

Can't remember where I saw it, but Ghost in the Shell is supposed to be one of the few anime made at a native 24fps.

However as Doctor M said, it is common for anime to be variable refresh, this just happens to not be one of them.

A thread about GITS on doom9 (too lazy to read it all, but maybe could be interested for someone):
https://forum.doom9.org/showthread.php?t=174849

Interesting link, but I think I must read it a few more times ... not only to get what they're stating, but also to figure out what it means for a final conclusion or inconclusion.  

I did find out something though ... diff.pics (the screenshotcomparison-like example they referred to) can't be saved complete by any of the usual methods (including "save as archive" - Firefox ZIP-compatible format .maff ).  

(2017-09-08, 07:06 PM)BronzeTitan Wrote: I'll work up (fingers crossed) the proof-of-concept again ...

So much for "fingers crossed".

It's fairly simple process in the paint program -- the wrong field (which was exposed into the ghosted-field) is "differenced" with the ghosted-field at a low percentage. That should remove the ghosting. Then the result is brightness & contrast adjusted to compensate for the dulling effect from this processing. Only, it's not coming through clean.

As the present paint program is a different one than I used on my old proof-of-concept, it could be, that the labels (or worse, the functions) don't correspond to one another. I'll test this a little more, now, but I'll probably need to go back to that previous program, if I can find it, to try again. (Hopefully, the GITS problem is the same type of overlapping-field problem as was in my previous fix-concept.)

Could one not just write a Avisynth script?
I figure one could take the progressive frames, do a selectevery() for every 3rd and 4th frame, interlace them and then select the correct fields and merge? I don't know if this would work, just an idea.

(2017-09-11, 07:30 PM)Beetwaaf Wrote: Could one not just write a Avisynth script?
I figure one could take the progressive frames, do a selectevery() for every 3rd and 4th frame, interlace them and then select the correct fields and merge? I don't know if this would work, just an idea.

Yeah, I thought something like that; doom9 is the place where ask for help, there are Avisynth gurus there!

(2017-09-11, 10:27 PM)spoRv Wrote:

(2017-09-11, 07:30 PM)Beetwaaf Wrote: Could one not just write a Avisynth script?
I figure one could take the progressive frames, do a selectevery() for every 3rd and 4th frame, interlace them and then select the correct fields and merge? I don't know if this would work, just an idea.

Yeah, I thought something like that; doom9 is the place where ask for help, there are Avisynth gurus there!

Indeed, I've learned most of what I know from doom9, however most of the time the threads get way to complicated and I stop being able to follow what they're doing.