I assume the template u used got a lmin value of 1 (lmin=1) ... so do another test using lmin=0.5 and try again. BUT Phil reported some days ago that lower lmin values at 1cd encodings (= low avg bitrate) do mess bitrate curves up. I dont remember if that only happened at 1pass VBR encodings.
So do a test and lower the lmin value as desribed above and do a
2pass encoding (generally recommended when using mencoder).
That was exactly the reason why I made packshot purposed only for dvd encodings (=higher avg bitrate), as no matter which parameters are used:
WE STILL didnt found the "general" full k(S)VCD compilant parameters where all these issues do not appear in general:
a) No buffer underruns
b) constant quality in the whole stream
c) no max bitrate overruns (max 2500)
I did read in here so many issues related to 1cd encodings using mencoder that in the "parameter experiences"-state now ... I do not recommend using mencoder at 1cd encodings. But that could change very fast as its a matter of figuring out the perfect paramaters for that format.