On Fri, Feb 01, 2019 at 05:50:37PM +0800, myp...@gmail.com wrote:
[...]
> > Ok, makes sense. Would you mind updating the comment to something like:
> >
> > /* Note: WEIGHT_LUT_SIZE must be larger than max_meaningful_diff
> > * (log(255)*max(h)^2, which is approximately 50 with the curr
On Fri, Feb 1, 2019 at 5:43 PM Clément Bœsch wrote:
> On Fri, Feb 01, 2019 at 05:19:53PM +0800, myp...@gmail.com wrote:
> > On Fri, Feb 1, 2019 at 5:11 PM Clément Bœsch wrote:
> > >
> > > On Fri, Feb 01, 2019 at 04:57:37PM +0800, myp...@gmail.com wrote:
> > > [...]
> > > > > > -#define WEIGHT_LU
On Fri, Feb 01, 2019 at 05:19:53PM +0800, myp...@gmail.com wrote:
> On Fri, Feb 1, 2019 at 5:11 PM Clément Bœsch wrote:
> >
> > On Fri, Feb 01, 2019 at 04:57:37PM +0800, myp...@gmail.com wrote:
> > [...]
> > > > > -#define WEIGHT_LUT_NBITS 9
> > > > > -#define WEIGHT_LUT_SIZE (1< > > > > +#define
On Fri, Feb 1, 2019 at 5:11 PM Clément Bœsch wrote:
>
> On Fri, Feb 01, 2019 at 04:57:37PM +0800, myp...@gmail.com wrote:
> [...]
> > > > -#define WEIGHT_LUT_NBITS 9
> > > > -#define WEIGHT_LUT_SIZE (1< > > > +#define WEIGHT_LUT_SIZE (80) // need to > 300 * 300 *
log(255)
> > >
> > > So the
On Fri, Feb 01, 2019 at 04:57:37PM +0800, myp...@gmail.com wrote:
[...]
> > > -#define WEIGHT_LUT_NBITS 9
> > > -#define WEIGHT_LUT_SIZE (1< > > +#define WEIGHT_LUT_SIZE (80) // need to > 300 * 300 * log(255)
> >
> > So the LUT is now 3.2MB?
> >
> > Why 300? 300*300*log(255) is closer to 500
On Fri, Feb 1, 2019 at 4:29 PM Clément Bœsch wrote:
>
> On Fri, Feb 01, 2019 at 10:45:24AM +0800, Jun Zhao wrote:
> > Remove the pdiff_lut_scale in nlmeans and increase weight_lut table size
> > from 2^9 to 80, this change will avoid using pdiff_lut_scale in
> > nlmeans_slice() for weight_lut
On Fri, Feb 01, 2019 at 10:45:24AM +0800, Jun Zhao wrote:
> Remove the pdiff_lut_scale in nlmeans and increase weight_lut table size
> from 2^9 to 80, this change will avoid using pdiff_lut_scale in
> nlmeans_slice() for weight_lut table search, it's will improve the
> performance about 12%. (i
Remove the pdiff_lut_scale in nlmeans and increase weight_lut table size
from 2^9 to 80, this change will avoid using pdiff_lut_scale in
nlmeans_slice() for weight_lut table search, it's will improve the
performance about 12%. (in 1080P size picture case).
Use the profiling command like:
perf