[FFmpeg-devel] [PATCH] Remove imgconvert.c:avg_bits_per_pixel()

[Michael Niedermayer]

On Fri, Jul 23, 2010 at 05:29:53PM +0200, Stefano Sabatini wrote:
> On date Monday 2010-02-01 00:36:16 +0100, Stefano Sabatini encoded:
> > On date Sunday 2010-01-31 17:05:23 +0100, Michael Niedermayer encoded:
> > > On Sun, Jan 31, 2010 at 04:55:22PM +0100, Stefano Sabatini wrote:
> > > > Hi, as in subject.
> > > > 
> > > > Follows the difference between the avg_bits_per_pixel() and
> > > > av_get_bits_per_pixel() (the second value comes from
> > > > av_get_bits_per_pixel(), which is also right).
> > > > 
> > > > bgr4_byte: 8 4
> > > > rgb4_byte: 8 4
> > > > yuva420p: 12 20
> > > > rgb555be: 16 15
> > > > rgb555le: 16 15
> > > > bgr555be: 16 15
> > > > bgr555le: 16 15
> > > > 
> > > > Regression test passed.
> > > 
> > > [...]
> > > > @@ -882,7 +834,7 @@
> > > >          if (pix_fmt_mask & (1ULL << i)) {
> > > >              loss = avcodec_get_pix_fmt_loss(i, src_pix_fmt, has_alpha) & loss_mask;
> > > >              if (loss == 0) {
> > > > -                dist = avg_bits_per_pixel(i);
> > > > +                dist = av_get_bits_per_pixel(&av_pix_fmt_descriptors[i]);
> > > >                  if (dist < min_dist) {
> > > >                      min_dist = dist;
> > > >                      dst_pix_fmt = i;
> > > 
> > > your change is incorrect, this function selects the pix format that is most
> > > compact within the losses the user accepts. (should probably let the user
> > > specify how many bits he is willing to loose as well)
> > > after yor change it favors formats that are worse but not more compact
> > 
> > I don't see how my change could have do that, I simply removed a wrong
> > implementation of a function for computing the average bits per pixel
> > with a correct one.
> 
> So my question: what's wrong with the proposed change?
> 
> Is avg_bits_per_pixel() supposed to be equivalent to
> av_get_bits_per_pixel() to av_get_bits_per_pixel() (in which case the
> change is correct), or is it something different?
> 
> In the latter case what does it represent?

after 6 month my memory says that you mixed up the sample precission aka
what affects how many colors can be represented with how many bits are
physically used per pixel.

and ultimately the definition of functions dont matter that much,
what matters is that functions still work and iirc this one broke
some pixel format selection code.

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Those who are too smart to engage in politics are punished by being
governed by those who are dumber. -- Plato 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 190 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20100724/ea89436f/attachment.pgp>