Log in

View Full Version : "Rainbows" in anime


archon231
24th January 2003, 02:48
Hey all ^_^,

I've recently been encodeing some Cowboy Bebop (not the best DVDs, not as bad as Evangelion though! :P) and there is a pretty big problem with the "rainbows" that appear, may be called "moire" dont know, they mostly appear when black outlines meat a light color, especially if there are a bunch of close, parallel lines. Well anyway, i've found a few filters to get rid of this, which seem to work pretty well, but they just take _way_ to long, and what i want to know is; Is there any new filters (that i missed) that do this well, and quickly?

Actually got another question, i also would like a good noise reducer, that operates directly from the .vob so i dont have to convert to avi then go through it again.

-thanx

Zarxrax
24th January 2003, 03:42
Guava Comb is great for removing rainbows.
I'm not sure I follow you on your second question, but maybe MPEG2DEC3 is something along the lines of what your looking for. It lets you use postprocessing on the VOBs. As far as I know though, any filters can be used without having to convert to avi.

archon231
25th January 2003, 21:02
Thanx, i'll check it out now ^_^.

Yeah sry about the second question, phrased it poorly, i'm using temporalsmooth works great, for what i wanted, a little slow T_T but worth it :) Actually if any knows of one just as good, or better that doesn't take so long, a link would be great, thanx.

angelyote
25th January 2003, 22:23
I've had really good luck with removing rainbows using Tim Parks smoothIQ (not to be confused with smoothHIQ). I don't believe this filter supports YV12. Nothing is mentioned about it in the docs.

Last seen here: http://www.doki.ca/filters/

I haven't tried guava comb so I don't know how they compare. I do know that smoothiq has done great things but also is extremely slow with the default values. For some reason I got better results with version 0.5 then 0.6. Go figure.

The diameter of this filter defaults to 11 which is huuuuuge and is the cause of the slowness. I've found that decreasing the value however takes away most of the benefit I'd gotten from using it.
Version 0.6 makes some attempts at optimizing the smoothing routine for diameter 11. ymmv.

Dave