On Mon, 2004-05-24 at 18:09 +0200, Bernhard Praschinger wrote: > Have you also taken a look at the lav2yuv secenlist feature ?
Not beyond having seen is presence in the usage: statement for lav2yuv when I was digging through the code to see just how difficult it would be. > There lav2yuv splits the inputfile into several scenes. Where is uses a > similair approach. I thought that might be an example I could modify for my purposes -- there is one fairly major rub with it though... > The first few lines are often black if you have 16:9 or a cinema movie. Right. That would be a small penalty over 4:3 material. Unavoidable. But fortunately currently, on non HDTV anyway, only a small handful of programs are being shown 16:9 so they are the exception rather than the rule. > You you have to take a look at the center, oft the first field or > something like that. Yeah, funny enough, the black frame search I did in mplayer looked at the centre of the frame and worked it's way towards the top and bottom of the frame for just that reason, however, that defeats the efficiencies of looking at the top of the frame -- that being the need to only jpeg-decode a minimum number of scanlines to determine that a frame is not black. If I use the centre lines, I need to decode whole frames. Even with 16:9 material, I have to decode more lines, but certainly not as many as decoding a half a frame or more. > I think you should take a look at lav2yuv, maybe it does something like > you want. So you can easy chack if it is usefull, and may be adapted. Yeah, I think it might. Now just for some time to actually do it. ~sigh~ Thanks much for your input Bernhard, b.
signature.asc
Description: This is a digitally signed message part
