echo
28th January 2003, 18:47
I noticed that some people like to do only one pass with I-frame constant quant 2 and P-frame constant quant 3. When I tried to test the same thing by putting 2 in the I-frame min and max quantizers and 3 in the P-frame min and max quantizers I noticed through the ffdshow's OSD that all frames, I and P, were quant 3. This happened with both the CBR 1-pass mode and the quality 1-pass mode. After doing a constant quant 3 encode, the file sizes were exactly the same. I tested all kinds of unstable builds and they all had this problem as well as the stable build by koepi.
The only way to use quant 2 for I-frames and quant 3 for P-frames seems to be only a 2-pass encode where after a short test I did seemed to work. But even then I get some I-frames with quant 3 and some P-frames with quant 2 as reported in DebugView, not many though, about 1 in every 1000 frames.
Anyone else can confirm this or is it just me?
Regards
echo
The only way to use quant 2 for I-frames and quant 3 for P-frames seems to be only a 2-pass encode where after a short test I did seemed to work. But even then I get some I-frames with quant 3 and some P-frames with quant 2 as reported in DebugView, not many though, about 1 in every 1000 frames.
Anyone else can confirm this or is it just me?
Regards
echo