[XviD-devel] b-frames api question

[Christoph Lampert]

On Thu, 27 Feb 2003, suxen_drol wrote:

> 
> On Tue, 25 Feb 2003 15:03:04 +0100 (CET) Christoph Lampert <chl at math.uni-bonn.de> wrote:
> 
> > > What do you think? I'm in favour of the second solution, but maybe the
> > > first has better future...
> > 
> > I vote for the first.
> 
> agree: too many options is better than too few.. as long as theyre
> sensible.
> 
> because we force the applications to memset(,0,) strucutres before usage,
> when sensitity is zero, a default (~50%) sensitity should be used.
> 
> ALSO regarding regarding the api: does any think we make more use of
> float/double fields? ie. bquant ratio and offset could be floats.

No, let's stick to int with 100th or 128th or 256th resolution (it's not
for users eyes anyway). I can't think of anything that really needs float,
and who knows, maybe we can save a few conversions then. 
Bzw. 0 should mean a default value for quant_ratio...  we should check for
this. 

I think, floats should only be used in special routines, so one knows when
emms() or similar has to be called an when not. GMC is such a routine, but
bframe quantization I'd rather think not. 

gruel