> On Sat, 2004-12-11 at 09:06, David Johnstone wrote: > > Any experiences out there from Australian users on which commercial > > detection scheme works best ? I turned off commercial detection a while > > back in frustration but now I'm getting weary pulling commercials out by > > hand. > > Hi David, > > I've also turned off commercial detection. Have you noticed the line on > the top left of the screen, it maybe the reason the detection methods > don't work. If we remove it then blank frame detection may work. > > Paul
I've made a few improvements in the current detection code in my source tree, one of those being the fact that it ignores the outer 20 pixels. I've sped up blank-frame detection and made it true "blank" vs "black" as it was previously. Do your recordings have blank frames at all between commercials or commercials & the show? I'm currently working on an "all methods" detection method. It isn't based on exact commercial lengths as the current blank-frame detection code is, it instead looks at blocks of video (separated by blank frames) and gives each one a score based on various things. It seems to be working a bit better on the 10 or so shows I'm testing with currently. The code also handles detecting blank frames as truly blank even if a logo is present. I hope to have this code in CVS in a week or so. Still doing some final testing on it before I start to run it on my production Myth setup, then once it's run there for a while and seems to be OK, I'll commit the code to CVS. Do either of you run CVS? If so, I could supply a patch against current CVS for you to give a shot to see if it helps. -- Chris
_______________________________________________ mythtv-users mailing list [EMAIL PROTECTED] http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
