<div dir="ltr"><br><div class="gmail_extra"><br><div class="gmail_quote">On Mon, Apr 11, 2016 at 7:36 PM, SÃ¸ren Sandmann Pedersen <span dir="ltr"><<a href="mailto:soren.sandmann@gmail.com" target="_blank">soren.sandmann@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">From: Bill Spitzak <<a href="mailto:spitzak@gmail.com">spitzak@gmail.com</a>><br> <br> If enable-gnuplot is configured, then you can pipe the output of a<br> pixman-using program to gnuplot and get a continuously-updated plot of<br> the horizontal filter. This works well with demos/scale to test the<br> filter generation.<br> <br> The plot is all the different subposition filters shuffled<br> together. This is misleading in a few cases:<br> <br> Â IMPULSE.BOX - goes up and down as the subfilters have different<br> Â Â Â Â Â Â Â Â numbers of non-zero samples<br> <br> Â IMPULSE.TRIANGLE - somewhat crooked for the same reason<br> <br> Â 1-wide filters - looks triangular, but a 1-wide box would be more<br> Â Â Â Â Â Â Â Â Â Â accurate<br></blockquote><div><br></div><div>Because you are not plotting the two dummy points at (0,Â±width/2), a 1-wide filter is actually just a single point.</div><div><br></div><div>You may be right that leaving the dummy points off the plot may make it easier to figure out what is going on.</div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> <br> Changes by SÃ¸ren: Rewrote the pixman-filter.c part to<br> Â Â Â - make it generate correct coordinates<br> Â Â Â - add a comment on how coordinates are generated<br> Â Â Â - in rounding.txt, add a ceil() variant of the first-sample<br> Â Â Â Â formula<br> Â Â Â - make the gnuplot output slightly prettier<br> <br> v7: First time this ability was included<br> <br> v8: Use config option<br> Â Â Moved code to the filter generator<br> Â Â Modified scale demo to not call filter generator a second time.<br> <br> v10: Only print if successful generation of plots<br> Â Â Â Use #ifdef, not #if<br> <br> v11: small whitespace fixes<br> <br> Signed-off-by: Bill Spitzak <<a href="mailto:spitzak@gmail.com">spitzak@gmail.com</a>><br> Signed-off-by: SÃ¸ren Sandmann <<a href="mailto:soren.sandmann@gmail.com">soren.sandmann@gmail.com</a>><br> ---<br> Â <a href="http://configure.ac" rel="noreferrer" target="_blank">configure.ac</a>Â Â Â Â Â Â |Â 13 ++++++<br> Â pixman/pixman-filter.c | 115 +++++++++++++++++++++++++++++++++++++++++++++++++<br> Â pixman/rounding.txtÂ Â |Â Â 1 +<br> Â 3 files changed, 129 insertions(+)<br> <br> diff --git a/<a href="http://configure.ac" rel="noreferrer" target="_blank">configure.ac</a> b/<a href="http://configure.ac" rel="noreferrer" target="_blank">configure.ac</a><br> index 6b2134e..e833e45 100644<br> --- a/<a href="http://configure.ac" rel="noreferrer" target="_blank">configure.ac</a><br> +++ b/<a href="http://configure.ac" rel="noreferrer" target="_blank">configure.ac</a><br> @@ -834,6 +834,19 @@ fi<br> Â AC_SUBST(PIXMAN_TIMERS)<br> <br> Â dnl ===================================<br> +dnl gnuplot<br> +<br> +AC_ARG_ENABLE(gnuplot,<br> +Â Â [AC_HELP_STRING([--enable-gnuplot],<br> +Â Â Â Â Â Â Â Â Â Â [enable output of filters that can be piped to gnuplot [default=no]])],<br> +Â Â [enable_gnuplot=$enableval], [enable_gnuplot=no])<br> +<br> +if test $enable_gnuplot = yes ; then<br> +Â Â AC_DEFINE(PIXMAN_GNUPLOT, 1, [enable output that can be piped to gnuplot])<br> +fi<br> +AC_SUBST(PIXMAN_GNUPLOT)<br> +<br> +dnl ===================================<br> Â dnl GTK+<br> <br> Â AC_ARG_ENABLE(gtk,<br> diff --git a/pixman/pixman-filter.c b/pixman/pixman-filter.c<br> index b2bf53f..af46a43 100644<br> --- a/pixman/pixman-filter.c<br> +++ b/pixman/pixman-filter.c<br> @@ -297,6 +297,117 @@ create_1d_filter (intÂ Â Â Â Â Â Â *width,<br> Â Â Â return params;<br> Â }<br> <br> +#ifdef PIXMAN_GNUPLOT<br> +<br> +/* If enable-gnuplot is configured, then you can pipe the output of a<br> + * pixman-using program to gnuplot and get a continuously-updated plot<br> + * of the horizontal filter. This works well with demos/scale to test<br> + * the filter generation.<br> + *<br> + * The plot is all the different subposition filters shuffled<br> + * together. This is misleading in a few cases:<br> + *<br> + *Â IMPULSE.BOX - goes up and down as the subfilters have different<br> + *Â Â Â Â Â Â Â Â numbers of non-zero samples<br> + *Â IMPULSE.TRIANGLE - somewhat crooked for the same reason<br> + *Â 1-wide filters - looks triangular, but a 1-wide box would be more<br> + *Â Â Â Â Â Â Â Â Â accurate<br> + */<br> +static void<br> +gnuplot_filter (int width, int n_phases, const pixman_fixed_t* p)<br> +{<br> +Â Â double step;<br> +Â Â int i, j;<br> +Â Â int first;<br> +<br> +Â Â step = 1.0 / n_phases;<br> +<br> +Â Â printf ("set style line 1 lc rgb '#0060ad' lt 1 lw 0.5 pt 7 pi 1 ps 0.5\n");<br> +Â Â printf ("plot '-' with linespoints ls 1\n");<br> +<br> +Â Â /* The position of the first sample of the phase corresponding to<br> +Â Â Â * frac is given by:<br> +Â Â Â *<br> +Â Â Â *Â Â Â ceil (frac - width / 2.0 - 0.5) + 0.5 - frac<br> +Â Â Â *<br> +Â Â Â * We have to find the frac that minimizes this expression.<br> +Â Â Â *<br> +Â Â Â * For odd widths, we have<br> +Â Â Â *<br> +Â Â Â *Â Â Â ceil (frac - width / 2.0 - 0.5) + 0.5 - frac<br> +Â Â Â *Â Â = ceil (frac) + K - frac<br> +Â Â Â *Â Â = 1 + K - frac<br> +Â Â Â *<br> +Â Â Â * for some K, so this is minimized when frac is maximized and<br> +Â Â Â * strictly growing with frac. So for odd widths, we can simply<br> +Â Â Â * start at the last phase and go backwards.<br> +Â Â Â *<br> +Â Â Â * For even widths, we have<br> +Â Â Â *<br> +Â Â Â *Â Â Â ceil (frac - width / 2.0 - 0.5) + 0.5 - frac<br> +Â Â Â *Â Â = ceil (frac - 0.5) + K - frac<br> +Â Â Â *<br> +Â Â Â * The graph for this function (ignoring K) looks like this:<br> +Â Â Â *<br> +Â Â Â *Â Â Â Â 0.5<br> +Â Â Â *Â Â Â Â Â Â |Â Â |\<br> +Â Â Â *Â Â Â Â Â Â |Â Â | \<br> +Â Â Â *Â Â Â Â Â Â |Â Â |Â \<br> +Â Â Â *Â Â Â Â Â 0 |Â Â |Â Â \<br> +Â Â Â *Â Â Â Â Â Â |\Â Â |<br> +Â Â Â *Â Â Â Â Â Â | \Â |<br> +Â Â Â *Â Â Â Â Â Â |Â \ |<br> +Â Â Â *Â Â Â -0.5 |Â Â \|<br> +Â Â Â *Â Â ---------------------------------<br> +Â Â Â *Â Â Â Â Â Â 0Â Â 0.5Â Â 1<br> +Â Â Â *<br> +Â Â Â * So in this case we need to start with the phase whose frac is<br> +Â Â Â * less than, but as close as possible to 0.5, then go backwards<br> +Â Â Â * until we hit the first phase, then wrap around to the last<br> +Â Â Â * phase and continue backwards.<br> +Â Â Â *<br> +Â Â Â * Which phase is as close as possible 0.5? The locations of the<br> +Â Â Â * sampling point corresponding to the kth phase is given by<br> +Â Â Â * 1/(2 * n_phases) + k / n_phases:<br> +Â Â Â *<br> +Â Â Â *Â Â Â Â Â 1/(2 * n_phases) + k / n_phases = 0.5<br> +Â Â Â *<br> +Â Â Â * from which it follows that<br> +Â Â Â *<br> +Â Â Â *Â Â Â Â Â k = (n_phases - 1) / 2<br> +Â Â Â *<br> +Â Â Â * rounded down is the phase in question.<br> +Â Â Â */<br> +Â Â if (width & 1)<br> +Â Â Â Â first = n_phases - 1;<br> +Â Â else<br> +Â Â Â Â first = (n_phases - 1) / 2;<br></blockquote><div><br></div><div>My version printed an extra point at 0,-width/2 here. Not sure if that is an improvement or a bug, you might want to try it just to get your opinion.</div><div>Â </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> +<br> +Â Â for (j = 0; j < width; ++j)<br> +Â Â {<br> +Â Â Â Â for (i = 0; i < n_phases; ++i)<br> +Â Â Â Â {<br> +Â Â Â Â Â Â int phase = first - i;<br> +Â Â Â Â Â Â double frac, pos;<br> +<br> +Â Â Â Â Â Â if (phase < 0)<br> +Â Â Â Â Â Â Â Â phase = n_phases + phase;<br> +<br> +Â Â Â Â Â Â frac = step / 2.0 + phase * step;<br> +Â Â Â Â Â Â pos = ceil (frac - width / 2.0 - 0.5) + 0.5 - frac + j;<br></blockquote><div><br></div><div>See below I think this has to be changed to the floor() version.</div><div>Â </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> +<br> +Â Â Â Â Â Â printf ("%g %g\n",<br> +Â Â Â Â Â Â Â Â Â Â pos,<br> +Â Â Â Â Â Â Â Â Â Â pixman_fixed_to_double (*(p + phase * width + j)));<br> +Â Â Â Â }<br> +Â Â }<br></blockquote><div><br></div><div>My version printed an extra point at 0,width/2 here.</div><div>Â </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> +<br> +Â Â printf ("e\n");<br> +Â Â fflush (stdout);<br> +}<br> +<br> +#endif<br> +<br> Â /* Create the parameter list for a SEPARABLE_CONVOLUTION filter<br> Â * with the given kernels and scale parameters<br> Â */<br> @@ -346,5 +457,9 @@ out:<br> Â Â Â free (horz);<br> Â Â Â free (vert);<br> <br> +#ifdef PIXMAN_GNUPLOT<br> +Â Â gnuplot_filter(width, subsample_x, params + 4);<br> +#endif<br> +<br> Â Â Â return params;<br> Â }<br> diff --git a/pixman/rounding.txt b/pixman/rounding.txt<br> index b52b084..1c00019 100644<br> --- a/pixman/rounding.txt<br> +++ b/pixman/rounding.txt<br> @@ -160,6 +160,7 @@ which means the contents of the matrix corresponding to (frac) should<br> Â contain width samplings of the function, with the first sample at:<br> <br> Â Â Â Â floor (frac - (width - 1) / 2.0 - e) + 0.5 - frac<br> +Â Â Â = ceil (frac - width / 2.0 - 0.5) + 0.5 - frac<br></blockquote><div><br></div><div>Unfortunately this produces numbers that don't agree with the filter generator or filtering code.</div><div><br></div><div>For a width==4 filter with n_phases==1, the generator seems to produce values at -1, 0, 1, 2, so the first sample is at -1. It also appears the filtering sampler is using the same rule, otherwise these filters would produce an obvious shift in the image.</div><div><br></div><div>Â Â frac = step/2 = .5</div><div>Â Â width = 3</div><div>Â Â floor (frac - (width - 1) / 2.0) + 0.5 - frac = floor(.5 - 1.5) + 0.5-0.5 = floor(-1) = -1.</div><div>Â Â ceil (frac - width / 2.0 - 0.5) + 0.5 - frac = ceil(.5 - 2.5) + 0.5-0.5 = ceil(-2) = -2</div><div><br></div><div>Whether by accident or not it looks like the correct formula is the floor one with the epsilon (e) removed, in disagreement with your rounding.txt document. I would recommend using that rather than changing the interpretation of the filter matrix.</div><div><br></div></div><br></div></div>