13
0
livetrax/libs/panners/1in2out/panner_1in2out.cc
Robin Gareus c02169a6c8
Micro-optimization: pre-calculate pan-law
powf() is very expensive to call, and as long as the pan-law
is fixed, there is no need to re-calculate coefficient scale
factor on every update.
2021-01-18 16:14:16 +01:00

375 lines
9.6 KiB
C++

/*
* Copyright (C) 2011-2012 Carl Hetherington <carl@carlh.net>
* Copyright (C) 2011-2017 Paul Davis <paul@linuxaudiosystems.com>
* Copyright (C) 2013 John Emmas <john@creativepost.co.uk>
* Copyright (C) 2014-2015 Robin Gareus <robin@gareus.org>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License along
* with this program; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
*/
#include <inttypes.h>
#include <cerrno>
#include <cmath>
#include <cstdio>
#include <cstdlib>
#include <float.h>
#include <locale.h>
#include <string>
#include <unistd.h>
#include <glibmm.h>
#include "pbd/cartesian.h"
#include "pbd/convert.h"
#include "pbd/enumwriter.h"
#include "pbd/error.h"
#include "pbd/failed_constructor.h"
#include "pbd/xml++.h"
#include "evoral/Curve.h"
#include "ardour/audio_buffer.h"
#include "ardour/buffer_set.h"
#include "ardour/debug.h"
#include "ardour/pannable.h"
#include "ardour/panner.h"
#include "ardour/profile.h"
#include "ardour/runtime_functions.h"
#include "ardour/session.h"
#include "ardour/utils.h"
#include "panner_1in2out.h"
#include "pbd/i18n.h"
#include "pbd/mathfix.h"
using namespace std;
using namespace ARDOUR;
using namespace PBD;
static PanPluginDescriptor _descriptor = {
"Mono to Stereo Panner",
"http://ardour.org/plugin/panner_1in2out",
"http://ardour.org/plugin/panner_1in2out#ui",
1, 2,
20,
Panner1in2out::factory
};
extern "C" ARDOURPANNER_API PanPluginDescriptor*
panner_descriptor ()
{
return &_descriptor;
}
Panner1in2out::Panner1in2out (boost::shared_ptr<Pannable> p)
: Panner (p)
{
if (!_pannable->has_state ()) {
_pannable->pan_azimuth_control->set_value (0.5, Controllable::NoGroup);
}
_can_automate_list.insert (Evoral::Parameter (PanAzimuthAutomation));
update ();
left = desired_left;
right = desired_right;
left_interp = left;
right_interp = right;
_pannable->pan_azimuth_control->Changed.connect_same_thread (*this, boost::bind (&Panner1in2out::update, this));
}
Panner1in2out::~Panner1in2out ()
{
}
void
Panner1in2out::update ()
{
#if 0
float const pan_law_attenuation = -3.0f;
float const scale = 2.0f - 4.0f * powf (10.0f, pan_law_attenuation / 20.0f);
#else
float const scale = -0.831783138f;
#endif
float const panR = _pannable->pan_azimuth_control->get_value ();
float const panL = 1 - panR;
desired_left = panL * (scale * panL + 1.0f - scale);
desired_right = panR * (scale * panR + 1.0f - scale);
}
void
Panner1in2out::set_position (double p)
{
if (clamp_position (p)) {
_pannable->pan_azimuth_control->set_value (p, Controllable::NoGroup);
}
}
bool
Panner1in2out::clamp_position (double& p)
{
/* any position between 0.0 and 1.0 is legal */
DEBUG_TRACE (DEBUG::Panning, string_compose ("want to move panner to %1 - always allowed in 0.0-1.0 range\n", p));
p = max (min (p, 1.0), 0.0);
return true;
}
pair<double, double>
Panner1in2out::position_range () const
{
return make_pair (0, 1);
}
double
Panner1in2out::position () const
{
return _pannable->pan_azimuth_control->get_value ();
}
void
Panner1in2out::distribute_one (AudioBuffer& srcbuf, BufferSet& obufs, gain_t gain_coeff, pframes_t nframes, uint32_t /* not used */)
{
assert (obufs.count ().n_audio () == 2);
pan_t delta;
Sample* dst;
pan_t pan;
Sample* const src = srcbuf.data ();
/* LEFT OUTPUT */
dst = obufs.get_audio (0).data ();
if (fabsf ((delta = (left - desired_left))) > 0.002) { // about 1 degree of arc
/* we've moving the pan by an appreciable amount, so we must
* interpolate over 64 samples or nframes, whichever is smaller */
pframes_t const limit = min ((pframes_t)64, nframes);
pframes_t n;
delta = -(delta / (float)(limit));
for (n = 0; n < limit; n++) {
left_interp = left_interp + delta;
left = left_interp + 0.9 * (left - left_interp);
dst[n] += src[n] * left * gain_coeff;
}
/* then pan the rest of the buffer; no need for interpolation for this bit */
pan = left * gain_coeff;
mix_buffers_with_gain (dst + n, src + n, nframes - n, pan);
} else {
left = desired_left;
left_interp = left;
if ((pan = (left * gain_coeff)) != 1.0f) {
if (pan != 0.0f) {
/* pan is 1 but also not 0, so we must do it "properly" */
mix_buffers_with_gain (dst, src, nframes, pan);
/* XXX it would be nice to mark that we wrote into the buffer */
}
} else {
/* pan is 1 so we can just copy the input samples straight in */
mix_buffers_no_gain (dst, src, nframes);
/* XXX it would be nice to mark that we wrote into the buffer */
}
}
/* RIGHT OUTPUT */
dst = obufs.get_audio (1).data ();
if (fabsf ((delta = (right - desired_right))) > 0.002) { // about 1 degree of arc
/* we're moving the pan by an appreciable amount, so we must
* interpolate over 64 samples or nframes, whichever is smaller */
pframes_t const limit = min ((pframes_t)64, nframes);
pframes_t n;
delta = -(delta / (float)(limit));
for (n = 0; n < limit; n++) {
right_interp = right_interp + delta;
right = right_interp + 0.9 * (right - right_interp);
dst[n] += src[n] * right * gain_coeff;
}
/* then pan the rest of the buffer, no need for interpolation for this bit */
pan = right * gain_coeff;
mix_buffers_with_gain (dst + n, src + n, nframes - n, pan);
/* XXX it would be nice to mark the buffer as written to */
} else {
right = desired_right;
right_interp = right;
if ((pan = (right * gain_coeff)) != 1.0f) {
if (pan != 0.0f) {
/* pan is not 1 but also not 0, so we must do it "properly" */
mix_buffers_with_gain (dst, src, nframes, pan);
/* XXX it would be nice to mark the buffer as written to */
}
} else {
/* pan is 1 so we can just copy the input samples straight in */
mix_buffers_no_gain (dst, src, nframes);
/* XXX it would be nice to mark the buffer as written to */
}
}
}
void
Panner1in2out::distribute_one_automated (AudioBuffer& srcbuf, BufferSet& obufs,
samplepos_t start, samplepos_t end, pframes_t nframes,
pan_t** buffers, uint32_t which)
{
assert (obufs.count ().n_audio () == 2);
Sample* dst;
pan_t* pbuf;
Sample* const src = srcbuf.data ();
pan_t* const position = buffers[0];
/* fetch positional data */
if (!_pannable->pan_azimuth_control->list ()->curve ().rt_safe_get_vector (start, end, position, nframes)) {
/* fallback */
distribute_one (srcbuf, obufs, 1.0, nframes, which);
return;
}
/* apply pan law to convert positional data into pan coefficients for
each buffer (output)
*/
#if 0
const float pan_law_attenuation = -3.0f;
const float scale = 2.0f - 4.0f * powf (10.0f, pan_law_attenuation / 20.0f);
#else
float const scale = -0.831783138f;
#endif
for (pframes_t n = 0; n < nframes; ++n) {
float panR = position[n];
const float panL = 1 - panR;
/* note that are overwriting buffers, but its OK
* because we're finished with their old contents
* (position automation data) and are
* replacing it with panning/gain coefficients
* that we need to actually process the data.
*/
buffers[0][n] = panL * (scale * panL + 1.0f - scale);
buffers[1][n] = panR * (scale * panR + 1.0f - scale);
}
/* LEFT OUTPUT */
dst = obufs.get_audio (0).data ();
pbuf = buffers[0];
for (pframes_t n = 0; n < nframes; ++n) {
dst[n] += src[n] * pbuf[n];
}
/* XXX it would be nice to mark the buffer as written to */
/* RIGHT OUTPUT */
dst = obufs.get_audio (1).data ();
pbuf = buffers[1];
for (pframes_t n = 0; n < nframes; ++n) {
dst[n] += src[n] * pbuf[n];
}
/* XXX it would be nice to mark the buffer as written to */
}
Panner*
Panner1in2out::factory (boost::shared_ptr<Pannable> p, boost::shared_ptr<Speakers> /* ignored */)
{
return new Panner1in2out (p);
}
XMLNode&
Panner1in2out::get_state ()
{
XMLNode& root (Panner::get_state ());
root.set_property (X_ ("uri"), _descriptor.panner_uri);
/* this is needed to allow new sessions to load with old Ardour: */
root.set_property (X_ ("type"), _descriptor.name);
return root;
}
string
Panner1in2out::value_as_string (boost::shared_ptr<const AutomationControl> ac) const
{
double val = ac->get_value ();
switch (ac->parameter ().type ()) {
case PanAzimuthAutomation:
/* We show the position of the center of the image relative to the left & right.
* This is expressed as a pair of percentage values that ranges from (100,0)
* (hard left) through (50,50) (hard center) to (0,100) (hard right).
*
* This is pretty wierd, but its the way audio engineers expect it. Just remember that
* the center of the USA isn't Kansas, its (50LA, 50NY) and it will all make sense.
*
* This is designed to be as narrow as possible. Dedicated
* panner GUIs can do their own version of this if they need
* something less compact.
*/
return string_compose (_ ("L%1R%2"), (int)rint (100.0 * (1.0 - val)),
(int)rint (100.0 * val));
default:
return _ ("unused");
}
}
void
Panner1in2out::reset ()
{
set_position (0.5);
update ();
}