Am Sonntag, 13. Januar 2008 10:26:03 schrieb Francesco Romani:
> Please provide the full command line used, possibily in the various
> stages tried, in order to better understand the transcode (wrong)
> behaviour.

Alright, this the way I first tried it and which, ideally, would be only one 
run:

transcode -y xvid -w 1600 -J 
yuy2toyv12,dilyuvmmx,text=range=60-$frames:string="somestring":font=/usr/share/fonts/truetype/FreeSans.ttf:points=12:dpi=120:fade=1:antialias=1:posdef=3,logo=file=/path/to/logo.png:posdef=2:rgbswap
 -x 
raw -i "$input" -o "$output"

After finding out what went wrong here (and after using mencoder to convert to 
YV12), I simply took out the yuy2toyv12 filter and tried again. 
Seeing that the logo will not be rendered (and after experimenting with the 
logo itself), I split it in at first two steps:

transcode -y im,lame -F png -J 
dilyuvmmx,text=range=60-$frames:string="somestring":font=/usr/share/fonts/truetype/FreeSans.ttf:points=12:dpi=120:fade=1:antialias=1:posdef=3
 -x 
raw -i "$input" -o "$output"

And yes, I know that jpg would probably have been a lot faster, but 
uncompressed jpg naturally tend to grow quite large, and I'm pushing enough 
data around as is.
After compositing all frames, I went on with:

transcode -y raw,null --export_fps 0,5 --use_rgb -z -k -g 680x480 -x 
imlist,null -i "$wd"/list -o "$wd"/tempfile.avi

(I actually experimented with the audio, but it was all basically the same as 
above and didn't change anything regarding the video)

After finding out that -k produced nonsense, I left that out and fed the video 
to mencoder (after merging the audio with avimerge).
>From then on, it's a simple encoding, with the options provided by xvid4.cfg:

transcode -y xvid -w 1600 -x 
raw -i "$wd"/tempfile.avi -o "$wd"/"$t_output".avi

That's about it.

> Are these streams in DV format? Or maybe uncompressed YUY2 frames packed
> in AVI or anything else?

No, they are not DV. As you guessed, uncompressed YUY2 in AVI. These are 
produced by DScaler when capturing from a tv-card. 

> Uhm, I should really improve detectclipping filter and its documentation
> as well :)

Probably, at least I was somewhat confused and then let it be ;)

> filter_logo was seriously improved during 1.1.0 cycle, maybe you could
> give a try using transcode 1.1.0-alpha6.

> Those anomalies are supposed to be fixed as well in 1.1.0

I will certainly give it a try later on.

> That's hard maybe seamples will be needed to fix that, and this raise
> various problems, first and most important being copyright issues that I
> absolutely want to avoid.

Yes, this is quite obvious. Though most of these are captures from video 
games, which are usually openly distributed on the net. Well, still probably 
a somewhat grey zone.

> No, this is definitively not acceptable. Let's see how this can be
> improved.

Let me know if you need more information.
I will try 1.1.0 later this day. Hope to report better news by then ;)

Regards,
Toni

Reply via email to